The track record

In September, the portfolio lost 4.62%, while the S&P 500 added 0.36%. YTD, it is at -0.90% against +12.47%. Since December 2021, the portfolio is up 88.83% against 66.63% for the S&P 500, or 14.09% a year against 11.17%.

The YTD gap is ~13 points, and it has two sources. A third of the portfolio is in Treasuries: YTD, SGOV returned ~3%, and EDV lost ~12%. At the same time, FSLR and ADBE both lost ~32% YTD, and V, my largest position, gained only ~4%.

One comparison changed this month. QQQ is now ahead since inception: +93.34% against my +88.83%, or 14.65% a year against 14.09%. In #002, the portfolio was ahead of all four benchmarks. It is still ahead of the S&P 500, VUG (12.64% a year), and SPYG (13.06%). The Sharpe ratio decreased to 0.82 from 0.90; beta (0.95) and the maximum drawdown (-19%, 123 days) did not change.

As promised in #002, the final August number is now in the table: +1.62% against +2.54% for the S&P 500. It is lower than the +2.50% I published because that number did not include Aug 31.

Analyst’s note:

The “1M” column in the benchmark table is the trailing month from the day of the screenshot, not the calendar month. The calendar September is in the monthly table: -4.62% against +0.36%.

Analyst’s note:

All numbers in this issue are as of September 30, 2026, taken during the US trading session (around midday in New York), not after the close. The final September figure, with the full last day, will appear in the monthly table of #004.

Read more

TL;DR: Credo (CRDO) sells the connections inside AI clusters: active electrical cables (AECs), optical transceivers and the chips inside them, and retimers. The Q1 FY2027 report was another beat, and the company raised its full-year growth outlook to more than 85%. At the same time, the stock fell 20% the next day and is now 32% below its mid-August price. I see smaller beats, not a weaker business. The rest of the year depends on a new optical business, and stock-based compensation (SBC) is close to 20% of revenue in the Q2 guidance. Verdict: a very profitable company priced as if its growth were temporary. CRDO trades at 24x forward earnings for 50% consensus EPS growth, and the price is above the zone of my model; my full fair price math and the accumulation zone are at the end of this post.

Investment Thesis

Credo reported Q1 FY2027 on Sep 01 with record revenue, a higher full-year growth outlook, and record non-GAAP EPS of $1.20. CRDO fell 20% the next day and 27% in 2 weeks. It is now 42% below its June high, while consensus EPS for the next 12 months is at a record. My idea here is to check whether the report changed the long-term economics of the company or only the shape of this year.

• Growth is still very high, but the beats became small. Revenue increased 115% YoY to $479.0M, the seventh quarter in a row of triple-digit growth. In the three quarters before Q4 FY2026, Credo beat the top of its own revenue guidance by 12-18%. In the last two quarters, the beat was below 1%.

• The market prices in a much weaker future than analysts expect. The consensus 5Y EPS growth estimate is 50.0%, and CRDO trades at 24.1x forward earnings, a PEG ratio of 0.48. At a 12% required return and a 22x exit multiple, today’s price needs only ~17% annual EPS growth.

• The business is already very profitable. LTM gross margin is 67.07%, net margin 33.83%, and ROIC 20.8%. FCF over the last 12 months is $439M. Credo has no borrowings and $764M of cash and short-term investments.

• The multiple is near the bottom of its range. CRDO trades at 24.1x forward earnings, 47% above its 16.4x -1 standard deviation band. The 88.5x mean since the IPO comes from the loss-making years, so I do not use it. And this despite the fact that analysts expect record EPS for the next 12 months.

• The bear case: two customers bring 71% of revenue, the second half depends on optical products that have just started to ship, SBC is ~20% of revenue in the Q2 guidance, and the diluted share count increased 5% in 1 year. My model takes this into account by capping future growth at 20%, far below consensus, and by using the 17x -1 standard deviation band as the Bear Case exit. With these limits, the price of $176 is 16% above my Bear Case fair price ($151) and 10% below the Base Case ($196).

Analyst’s note:

EPS in this post is non-GAAP, since the estimates are built on it. The difference is large. In Q1, GAAP EPS was $0.67 against $1.20 non-GAAP, mostly because of SBC, acquisition costs, and the amortization of acquired intangibles. The 71% is the share of the two largest contracting customers in the 10-Q. By end customer, the top four brought 84% of revenue.

Company Overview

Next Earnings Date: Dec 02, 2026 (estimated)
Market Cap: $33.06B
Sector: Information Technology | Industry: Semiconductors
Type: Large Growth
Short Interest: 3.3%

Credo Technology Group was founded in 2008 in San Jose, California, and it is incorporated in the Cayman Islands. It started with serializer/deserializer (SerDes) technology - the circuits that move data between chips at high speed. Its engineers began to work on active electrical cables in 2017, and production of the first HiWire AECs started in 2019. An AEC is a copper cable with Credo chips at both ends, which clean and boost the signal. Credo created this product category, and AECs are still its largest business. The IPO was in January 2022 at $10 per share. Bill Brennan has been CEO for 13 years, and the two founders, Lawrence Cheng (CTO) and Job Lam (COO), are still in management and on the board.

The product range has four groups. AECs connect GPUs and servers with the first switch at distances of up to 7 meters. The optical business includes digital signal processors (DSPs) for optical transceivers, silicon photonics photonic integrated circuits (PICs) from the DustPhotonics acquisition, and ZeroFlap optical transceivers that Credo builds itself. Retimers (Screaming Eagle, Blue Heron, and Toucan) keep the signal clean inside servers and in scale-up networks. The rest is SerDes IP licensing and chiplets. Most products work with PILOT, Credo’s diagnostics software. The company is fabless - TSMC (TSM) makes its chips, and Amkor (AMKR) and ASE (ASX) package and test them. 616 of its 807 employees are engineers, according to the 10-K.

Credo sells to hyperscalers and neoclouds directly and through cable and optical module makers. By shipment destination (Q1 10-Q), 53% of Q1 revenue went to the US, 19% to Hong Kong, and 16% to Malaysia. Mainland China decreased from 23% of revenue a year ago to 0.2%.

Market Overview

The optical market is growing faster than almost any other part of the data center. According to LightCounting (January 2026), sales of Ethernet optical transceivers and co-packaged optics (CPO) for AI clusters reached $16.5B in 2025 and should reach $26B in 2026, +60% in both years. In April 2026, it forecast 65% growth for all Ethernet transceivers in 2026, after +93% in 2024. In March 2026, it wrote that there is “a reasonable chance” for AI cluster optics to reach $100B per year by 2030. To compare, Credo’s optical target of more than $600M for FY2027 is less than 3% of the 2026 market.

Copper still makes most of the connections, and AECs extend its reach. According to Brennan, copper is still more than 95% of the connections in a data center, the same share as 12 years ago. AECs cover the short end of the pluggable market, up to 7 meters. According to Brennan, the only switch between the two media so far went the other way - copper replaced optics at lengths of up to 7 meters, first of all because it is more reliable, and also because it uses less power. Credo does not report AEC revenue separately, but management says it more than doubled in FY2025 and more than tripled in FY2026. I did not find a reliable, recent third-party estimate of the AEC market. The published ones are far below Credo’s own AEC sales (AECs brought over $230M of extra revenue in Q1 alone), so I do not use them.

The buyers are few. LightCounting estimates that four cloud companies are responsible for more than 50% of AI spending on optical transceivers: Amazon (AMZN), Google (GOOGL), Meta (META), and Microsoft (MSFT). In its January report, it also expects optics to be 3.1% of the capex of the top 5 cloud companies in 2026, up from 2.7% in 2025. Credo has the same structure on a smaller scale, with two customers at 71% of revenue.

The risk is the timing, not the size. In its newest forecast (April 2026), LightCounting writes that demand exceeds supply by 30%, and that the shortages should go away by the end of 2026. It also warns that transceiver sales usually drop once the shortages ease, even while capex still grows: "This will happen again in the current cycle, but we just don't know when." In March, it pointed to a possible flat or negative year in 2027-2028. Credo plans its fastest growth for exactly this period: ~20% QoQ in Q3 (November-January) and ~30% in Q4 (February-April). In my view, this is the most important market question for CRDO today. A pause in orders across the industry would hurt a company whose guidance needs acceleration.

Economic Moat

Every LTM margin is far above its 5Y mean. Gross margin is 67.07% (mean: 62.41%), EBIT margin 31.74% (-3.07%), net margin 33.83% (-1.66%), and FCF margin 27.57% (-9.5%). The means include the loss-making years before FY2025, so the more useful number is the trend. The non-GAAP gross margin stayed between 67.6% and 68.6% for five quarters, while quarterly revenue more than doubled. Management guides FY2027 at the same level as FY2026 (68.1%).

Let’s see where this margin comes from. Part of it is intellectual property. Credo owns 86 issued US patents and 52 in mainland China. In March 2025, it filed a complaint with the US International Trade Commission (ITC) against Amphenol (APH), Molex, TE Connectivity (TEL), and Volex over its AEC patents. By March 2026, all four cases were settled with license or cross-license agreements, and Siemon and 3M (MMM) also signed licenses for Credo’s AEC patents. The terms are confidential, so I cannot say what these licenses are worth. What I like: four large cable makers chose a license over a long fight.

The system approach matters even more. Credo designs the SerDes, the chip, the firmware, and the cable, manages the supply chain, and tests its cables with the customer's own switches and network cards before it ships them. Management calls this "probably our biggest competitive moat" and says the company never fails a customer qualification. I cannot verify the second claim. At the same time, it explains why hyperscalers stay: once a cable is qualified for a cluster, a second supplier has to pass the same qualification again. Next, cost. According to the 10-K, Credo's SerDes reaches similar performance to competitors on an older and cheaper chip process.

The weak point is the customer's power. Two contracting customers brought 71% of Q1 revenue, and the remaining performance obligations were only $4.2M. According to the 10-K, Credo does not have long-term purchase commitments from its customers, and they can cancel or delay orders.

In my view, Credo has a narrow moat. The returns are high, and the patents held up, but the high profits are only 2 years old, and a customer can move volume to a second supplier after a new qualification.

The returns on capital support the moat, but the history is short. ROIC is 20.8% (5Y mean: 2.89%), ROE is 30.67% (4.0%), and ROA is 16.11% (1.81%). All three were negative or close to zero until 2025. ROIC decreased in the last quarter. GAAP operating income decreased 23% QoQ because of SBC and the amortization of acquired intangibles, and the DustPhotonics deal added $1.24B of goodwill and intangibles.

Business Strategy

Own the whole link. Credo builds its own SerDes, DSPs, and, since the DustPhotonics deal, silicon photonics PICs. On top of that, it designs complete cables and transceivers. Management says Credo is the only vertically integrated player in optical transceivers and expects this to give it an advantage on both the cost of goods and the price. Competitors split the same stack between partners. For example, Marvell (MRVL) sells AEC DSPs to cable makers, and in December 2025, it launched a Golden Cable program that gives them reference designs and software.

Sell reliability, not only bandwidth. In an AI cluster, one unstable link (a "link flap") can force a reset of the whole cluster. According to management, customers have reported losing 10-20% of GPU utilization because of it. Credo developed ZeroFlap with input from xAI and Oracle (ORCL). According to Bill Brennan, xAI can now bring a cluster up in 5 days instead of 6-8 weeks with ZeroFlap AECs, and its uptime has reached 99.9%. ZeroFlap optical transceivers do the same for longer links. They measure the health of every link in real time and take an unstable link down before it breaks the cluster. The company says it is not targeting the commodity IEEE 1.6T transceiver market, but the parts of the network where reliability justifies a higher price.

Optics as the second business. Management expects more than $600M of optical revenue in FY2027. ZeroFlap optics, silicon photonics PICs, and optical DSPs are each expected to bring more than $100M. The optical DSP business had record revenue in Q1, and the first PIC revenue came in the same quarter. The DustPhotonics team already has two design wins with major customers, with ramps late in FY2027 or in FY2028. By the end of FY2027, Credo plans to produce hundreds of thousands of optical units per month, and then double or triple that rate by the end of FY2028.

New products for FY2028. Active LED cables (ALCs) use micro-LEDs instead of lasers to reach 30 meters with the reliability of copper, and they are 75% smaller in volume than 7-meter AECs. OmniConnect gearboxes connect processors with LPDDR memory to solve the memory bottleneck in inference. Management estimates the Credo content at thousands of dollars per GPU. Both products should bring their first revenue in FY2028. Credo also joined the Open CPX MSA and has design wins for near-package optics (NPO) in scale-up networks, with ramps starting in FY2028.

Buy teams, not revenue. In 12 months, Credo made three acquisitions: Hyperlume (micro-LED technology, $92.0M, September 2025), CoMira (link layer and security IP, $35.1M, February 2026), and DustPhotonics (silicon photonics, $1.25B, May 2026). The 10-Q calls their effect on the results not material. As a result, the 115% growth in Q1 is almost entirely organic, and over 90% of the YoY increase came from higher AEC volumes.

Capital Allocation

The DustPhotonics deal halved the cash in one quarter. Cash and short-term investments fell to $764M from $1.44B in May. Credo paid $770M in cash, $169M in shares, and booked $310M of contingent consideration that depends on future milestones. As a result, goodwill and intangibles increased to $1.37B, which is 50% of the $2.73B equity. In my view, the price is high for a team with no material revenue yet, but it is 4% of the market cap, and ~38% of it is paid in shares or only if targets are met.

The balance sheet is still very strong. Total assets are $3.01B against $284M of liabilities. Credo has no borrowings, and the $26M of debt in Koyfin is operating lease liabilities. Debt/Equity is 1.0% against a 2.9% 5Y mean, and net cash is $738M.

FCF reached $439M over the last 12 months, a 27.57% margin, against a $50.1M 5Y mean. Capex is $62M, 3.9% of revenue. The weak spot is Q1. Operating cash flow was only $90.2M against $129.4M of net income, since working capital used $151.3M. Receivables increased $54.5M, and inventory increased $61.5M "to support unfulfilled backlog and related new product ramps". Another $25.3M went to refundable deposits that reserve manufacturing capacity. Inventory is now $313.1M, which is ~168 days of Q1 cost of revenue by my calculation.

The share count increases every year. Diluted shares increased 5.3% YoY to 194.4M, and management expects ~200M in Q2. The two largest increases were equity sales: a public offering of 8.94M new shares at $17.50 in December 2023, and an at-the-market (ATM) program that sold 4.8M shares for $736M of net proceeds from October 2025 to January 2026. Credo pays no dividend and does not buy back shares.

As a result, shareholder yield is -2.29% against a -3.43% 5Y mean. The negative number is the dilution.

SBC is the main cost that the non-GAAP numbers hide. LTM SBC is $235M, 14.78% of revenue, against a 14.67% 5Y mean. In Q1, SBC increased 148% YoY to $88.0M, and the Q2 guidance implies ~$104M, ~20% of revenue. In Q1 alone, Credo granted 2.1M restricted stock units (RSUs) and 1.9M performance units (PSUs). The PSUs include a special award for the CEO - up to 1.44M shares in six tranches, which require trailing revenue of $2.5-7.5B and a stock price of $244.70-489.40 before June 2031. What I like: both hurdles have to be met (with one exception for the first two tranches), and the first revenue hurdle is already the FY2027 consensus. LTM SBC equals 54% of FCF, so FCF after SBC is only $203M.

Operating costs grow slower than revenue, but faster than planned. Over the last 12 months, cost of revenue was $524M, R&D $341M, and SG&A $221M. In Q1, non-GAAP operating expenses were $95.2M, above the $86-90M guidance, since R&D for new products increased. Management expects non-GAAP operating expenses to grow ~55% in FY2027, against more than 85% revenue growth.

As I see it, Credo allocates capital in an aggressive but focused way. All the cash goes into R&D, inventory, and technology teams, and shareholders pay for it with dilution instead of receiving a return.

Advantages

• The fastest growth in the group at one of the lowest multiples. LTM revenue grew 165%, faster than Broadcom (AVGO), Marvell, and Astera Labs (ALAB). At the same time, CRDO trades at 24.1x forward earnings, against 44.8x for Marvell and 54.0x for Astera Labs.

• The inventor of the AEC category. Credo started production of AECs in 2019, and Microsoft was its first AEC customer. In 2025-2026, four cable makers settled Credo’s patent cases with licenses or cross-licenses, and two more companies licensed the patents.

• Stable margins during a very fast ramp. The non-GAAP gross margin was 67.6-68.6% in each of the last five quarters, while quarterly revenue increased from $223M to $479M. Management guides the same margin level for FY2027, even with the optical ramp.

• A net cash balance sheet after a large deal. Credo paid $770M in cash for DustPhotonics and still has $764M of cash and short-term investments, no borrowings, and a 1.0% Debt/Equity ratio.

• More products for the same customers. Credo works with 5 hyperscalers and a growing number of neoclouds. Optics should bring more than $600M in FY2027, and ALCs, OmniConnect, and near-package optics add new revenue from FY2028.

Disadvantages

• Customer concentration. Two contracting customers brought 71% of Q1 revenue, and the top four end customers brought 84%. The remaining performance obligations are only $4.2M, and customers have no long-term purchase commitments. The CEO himself describes the hyperscalers as “surging and then pausing”.

• The second half depends on products that have just started to ship. The CFO expects ~20% sequential growth in Q3 and ~30% in Q4. The first PIC revenue came only in Q1, and the first 1.6T DSP revenue is expected later this fiscal year. If the optical ramp slips by one or two quarters, the FY2027 growth outlook is too high.

• Heavy SBC and dilution. SBC is 14.78% of LTM revenue and ~20% in the Q2 guidance. The diluted share count increased 5.3% in 1 year. In Q1, GAAP EPS was only 56% of non-GAAP EPS.

• AECs become easier to copy. Marvell introduced a 1.6T DSP for AECs in 2024 and started a program with reference designs for cable makers in December 2025. Astera Labs sells smart cable modules for AECs, and the four cable makers from the patent cases now have licenses. Brennan says new entrants show that the market is large. In my view, more suppliers are the main threat to the 68% gross margin.

• Weaker cash conversion and a large inventory. Q1 operating cash flow was 70% of net income, and inventory increased 25% QoQ to $313M. Goodwill and intangibles are now $1.37B, so a failed optical ramp would also be an accounting problem.

What the bears get right: the beats became small exactly when the plan became more ambitious. Revenue in Q4 FY2026 and Q1 FY2027 came in only 0.5% and 0.8% above the top of the guidance, after 12-18% in the three quarters before. Even the top of the Q2 guidance ($535M) is +99.6% YoY, so the streak of triple-digit growth ends unless Credo beats it by more than $1M. The CFO's path for Q3 is only ~56% YoY, since Q3 FY2026 was a very strong quarter. A pause at a large customer already happened once. In February 2023, Credo's largest customer reduced its demand forecast; the company cut its quarterly revenue guidance to $30-32M (consensus was above $45M), and CRDO fell 47% in one day. FY2024 revenue then increased only 4.8%. The biggest risk, as I see it, is a repeat of 2023 at one of the two largest customers while the optical business is still small. My fair price does not include this risk, so it has to be handled with the size of a position.

Competitors

I compare Credo with 3 semiconductor companies and the sector ETF. In the 10-K, Credo names Broadcom, Marvell, and Astera Labs as its principal competitors, together with “various cable and optical transceiver suppliers”. Broadcom makes DSPs, SerDes, and custom AI chips for hyperscalers, and it also sells infrastructure software. Marvell sells optical DSPs and DSPs for AECs. Astera Labs makes retimers and switches for scale-up networks. In AECs, Marvell and Astera Labs sell chips and modules to cable makers, while Credo sells the complete cable. The cable makers from the patent cases are competitors too, but they sell many other products, so their multiples say little about AECs. The iShares Semiconductor ETF (SOXX) shows the level of the whole sector.

CRDO is not the cheapest name here, and Broadcom is the real alternative. The forward P/E is 20.7x for Broadcom, 24.1x for Credo, 44.8x for Marvell, and 54.0x for Astera Labs. SOXX trades at ~21.5x. Consensus 5Y EPS growth is 54.69% for Broadcom, 52.49% for Astera Labs, 50.0% for Credo, and 45.81% for Marvell. As a result, the PEG ratio is 0.38 for AVGO and 0.48 for CRDO, against 0.98-1.03 for MRVL and ALAB, and ~1.1 for SOXX. The strong point of Credo is LTM revenue growth: 165.11%, against 98.48% for Astera Labs and 30.62-48.69% for the other two.

On returns and margins, Credo is second after Broadcom. ROIC is 26.49% for Broadcom, 20.8% for Credo, 19.42% for Astera Labs, and 4.95% for Marvell. Broadcom also has the highest gross margin (75.52%), net margin (42.94%), and FCF margin (44.22%). Credo is second in net and FCF margin. On the balance sheet, Credo is the strongest, with a 1.0% Debt/Equity ratio, against 59.6% for Broadcom.

In my view, the next-best alternative is Broadcom. It has a lower P/E, a lower PEG ratio, higher margins, and a higher ROIC. The difference is the exposure. Broadcom is a much larger and more diversified company, with custom chips, networking, and software, and it has much more debt. Credo is a pure play on AI connectivity, with the fastest revenue growth in the group, net cash, and a much higher customer concentration. For me, CRDO makes sense only as a bet on this specific market, not as a cheaper Broadcom. On valuation alone, Broadcom wins.

Past

Q1 FY2027 results (quarter ended August 1, USD, YoY):

• Revenue: $479.0M, +115% (+10% QoQ), above the top of the $465-475M guidance

• Non-GAAP EPS +131% to $1.20; GAAP EPS +97% to $0.67

• Non-GAAP gross margin 68.0% (67.6% a year ago); non-GAAP operating margin 48.2% (43.1%)

• Non-GAAP operating expenses +75% to $95.2M, above the $86-90M guidance

• Operating cash flow $90.2M and FCF $82.9M; inventory +$62M QoQ to $313.1M

• LTM: revenue $1.59B, net income $538M, FCF $439M, and ROIC 20.8%

Why did CRDO fall 42% from Aug 14 to Sep 15? The stock was already 27% below its August high when the report came out. Here are the reasons, starting with the most important one:

• The reaction to the report. CRDO fell 8.65% on Sep 01, before the release, and 20.04% on Sep 02, while SOXX increased 0.23% that day.

• Small beats. Revenue was $4M (0.8%) above the top of the guidance and 0.7-1.2% above the consensus (Zacks and TipRanks). EPS of $1.20 beat the $1.17 consensus by 2.6%. In the first three quarters of FY2026, the beats were much larger, so the market read it as a slowdown.

• The GAAP gross margin. It decreased from 68.2% to 64.5%, and the Q2 guidance is 62.9-64.9%. Some analysts saw pricing pressure in it. The main reason is accounting: $11.0M of amortization of the DustPhotonics intangibles and $5.7M of SBC in cost of revenue. The non-GAAP margin was 68.0%, at the midpoint of the guidance.

• Operating expenses. Non-GAAP operating expenses were $95.2M, above the $86-90M guidance, and GAAP operating expenses more than doubled YoY.

• The sector. On Aug 18 and Sep 14, CRDO fell 13.03% and 7.89%, while SOXX fell 4.96% and 5.63%. Coherent (COHR) fell ~13% on both days, and Astera Labs fell 11.74% on Sep 14. From Aug 14 to Sep 15, SOXX lost 9%, and Astera Labs lost 21%.

Still, the numbers in the report were strong. The Q2 guidance midpoint ($530M) is 5% above the Zacks consensus before the report ($502.69M), and the company raised its full-year growth outlook. In my view, the reaction is only partly justified. The worry about the second half is fair, since it depends on new optical products. The worry about the gross margin is not, since the non-GAAP margin did not change.

Over a longer period, the record is uneven. Revenue increased from $58.7M in FY2021 to $1.34B in FY2026, and the 3Y revenue CAGR is 93.5%. The path was not smooth. FY2023 revenue ($184.2M) missed the company’s own target of “at least $200 million”, and in FY2024, revenue increased only 4.8% after the largest customer reduced its demand forecast. Credo had GAAP net losses from FY2021 to FY2024 and returned to profit in FY2025 ($52.2M). In FY2026, net income reached $472.3M, and GAAP EPS reached $2.51.

The stock was an excellent investment, but not a calm one. Since its first trading day on Jan 27, 2022, CRDO returned 1,410%, or 79.5% per year, against 14.66% per year for the S&P 500 (VOO). Right now, it is 42% below its June high, and earlier this year, it increased 245% from its March low in less than 3 months. Beta is 3.23.

Analyst’s note:

Koyfin shows a 70.74% CAGR “over 5 years” for CRDO, but the stock has traded for 4.6 years, and the chart starts from the first month-end close ($12.12). I calculated the returns from the daily closes instead: $11.65 on Jan 27, 2022 and $175.89 on Sep 18, 2026, with dividends reinvested for VOO.

Future

• Consensus revenue: FY2027 $2.50B (+87.36%), FY2028 $3.88B (+54.96%), FY2029 $5.07B (+30.86%) - the growth is expected to slow every year

• Consensus EPS: $6.31 -> $9.70 -> $12.45 (FY2027-FY2029); FY2027 EPS is 82% above the FY2026 non-GAAP EPS of $3.46

• 5Y forward EPS growth estimate: 50.0%, against a 33.56% mean since 2022 and a 59.83% +1 standard deviation band. To compare, non-GAAP EPS was $0.70 in FY2025 and $3.46 in FY2026

• Analysts: 18 of 19 rate the stock Buy or Strong Buy (4 Strong Buy, 14 Buy, and 1 Hold), average target $282.47, +60.59% from here (the range is $185.00-350.00)

• Company guidance: FY2027 revenue growth of more than 85% (raised from more than 80% in June), optical revenue of more than $600M, a non-GAAP gross margin broadly consistent with FY2026, ~55% growth in non-GAAP operating expenses, and a non-GAAP net margin of ~50%; Q2 revenue $525-535M, a non-GAAP gross margin of 67-69%, and non-GAAP operating expenses of $100-105M

What must go right. The CFO described the path at the Goldman Sachs conference on Sep 10: ~11% sequential growth in Q2, ~20% in Q3, and ~30% in Q4. That gives ~$2.47B of FY2027 revenue and a $1B quarter "within our sights" in early FY2028. In my view, the plan is realistic if three things happen: the optical products ramp on time, the AEC business keeps growing with the move to 1.6T, and the gross margin stays near 68% while optics becomes a larger part of revenue. The first one is the least certain. My model needs only 20% annual EPS growth over 5 years. That is why the stock does not need the full plan. It needs the business to keep growing after FY2027. The next checkpoints are the ALC demonstration at the Open Compute Project (OCP) summit in October and the Q2 report in early December.

Current Valuation

Current vs mean since IPO:
Price/Fwd Earnings: 24.1x vs 88.5x
Price/Fwd Sales: 11.4x vs 14.2x
Price/FCF: 75.4x vs 195.3x
Price/Book: 12.1x vs 12.4x
PEG: 0.48 vs 2.88
Forward Earnings Yield: 3.6%

All five multiples are below their means since the IPO, but these means are not a good anchor. The forward P/E is 24.1x against a 16.4x -1 standard deviation band, and Price/Sales is 11.4x against an 8.9x band. On GAAP earnings, CRDO trades at 62.2x LTM EPS, and the FCF yield is only 1.33%. To compare, SOXX trades at ~21.5x forward earnings. In my view, CRDO is priced slightly above an average semiconductor company, although its PEG ratio is 2.3x lower (0.48 vs ~1.1).

Analyst’s note:

The means start in 2022, when Credo had no profit, so the forward P/E reached almost 400x in 2023, and the 88.5x mean is not usable. Price/FCF has data only from mid-2024, when FCF became positive. Also, Koyfin’s 24.1x uses next-12-months EPS ($7.30), which includes part of FY2028. On FY2027 EPS of $6.31, the P/E is 27.9x. The exit multiples in my model start from the -1 standard deviation band, not from the mean.

The growth estimate is above its mean, and the PEG is near its lowest level. The 5Y EPS growth estimate is 50.0% against a 33.56% mean, and the PEG ratio is 0.48 against a 2.88 mean. The mean PEG is also distorted by 2023, but even SOXX has a PEG of ~1.1, which is 2.3x higher.

What does the price already assume? If I require 12% per year, CRDO needs ~24% annual EPS growth for 5 years when the multiple decreases to 17x. At a 22x exit, ~17% is enough, and at 30x, ~10%. Consensus expects ~53% per year from FY2026 to FY2029. A reverse DCF shows that the current price implies 41.4% annual revenue growth for 5 years, while consensus expects ~56% per year until FY2029.

One chart frames this whole section. Its three dashed lines are the consensus forward EPS multiplied by my exit multiples (17x, 22x, and 30x), and the black line is the price.

Fair Price

The model starts from the FY2027 consensus EPS of $6.31 and grows it for 5 years. Consensus expects 50.0% per year, which is 2.5x my 20% cap, so the model uses 20%. Credo pays no dividend. As a result, EPS reaches $15.70 in FY2032, while consensus already expects $12.45 in FY2029.

The discount rate is 12%; the margin of safety is 30%. The exit multiples are 17x, 22x, and 30x. 17x is the -1 standard deviation band (16.4x today), so this case assumes that CRDO stays at the bottom of its range for 5 years. At the 20% cap, 17x is a PEG of 0.85. 22x is where CRDO traded right after the September 1 report, a PEG of 1.1. 30x is the ceiling of my method, a PEG of 1.5 at the 20% cap.

Bear Case (exit P/E 17x): fair price $151 - MoS price $106
Base Case (exit P/E 22x): fair price $196 - MoS price $137
Bull Case (exit P/E 30x): fair price $267 - MoS price $187

At $176, CRDO trades 10% below the Base Case. It is also 16% above the Bear Case fair price. My accumulation zone is $106-151, so the price is above it. Growth is the input that moves the result most. If EPS grows 15% per year, not 20%, the fair prices fall to $122/$158/$216, and CRDO is 44% above the Bear Case. At the same time, 15% would already mean a very sharp slowdown from the ~53% that consensus expects until FY2029. Worth noting that the average street target ($282) is above my Bull Case ($267), so my corridor is more conservative than the street.

Due Diligence

Profitability (12 of 15):

• Positive Gross Profit: $1.07B

• Positive Operating Income: $505M

• Positive Net Income: $538M

• Positive FCF: $439M

• Gross margin >= 40%: Yes (67.07%)

• Net margin >= 10%: Yes (33.83%)

• FCF margin >= 10%: Yes (27.57%)

• Management (ROIC, ROE, ROA) >= 10%: Yes (20.8%/30.67%/16.11%)

• Strong 3Y Revenue Growth: Yes (~94% per year)

• Revenue Growth Forecast: Yes (~56% per year over the next 3 years)

• ROE is increasing: negative in FY2024 -> 30.67%

• ROIC is increasing: negative in FY2024 -> 20.8%

• Revenue surprises since the IPO in a row: No (in line in Q3 FY2023 and Q4 FY2024)

• EPS surprises since the IPO in a row: No (a miss in Q2 FY2023)

• EPS growth YoY 5Y in a row: No (losses until FY2024)

Financial Strength (5 of 6):

• Total assets ($3.01B) exceed total liabilities ($284M) by 10.6x

• Negative Net Debt: Yes ($738M of net cash, with cash and short-term investments of $764M against $26M of lease liabilities)

• Low Debt/Equity: 1.0% (5Y mean: 2.9%)

• Debt/Capital: 0.95% (5Y mean: 2.83%)

• Interest coverage (FFO): Yes (no borrowings)

• Piotroski F-Score: 6 of 9 (not passed: operating cash flow below net income, lower current ratio YoY, and more shares outstanding)

Valuation and Advantage (3 of 4):

• Valuation < mean since IPO: Yes

• Valuation < the industry: No (P/Fwd E 24.1x vs SOXX at ~21.5x and AVGO at 20.7x)

• Does it have a moat: Yes (narrow)

• Outperformed the S&P 500 since the IPO: Yes (79.5% vs 14.7% CAGR)

Shares (1 of 3):

• Insider ownership >= 5%: Yes (9.06%)

• Fewer shares outstanding YoY: No (194M vs 185M diluted)

• Insider buys in the last 6M: No ($178M of sales, mostly under 10b5-1 plans)

Price (3 of 4):

• 1Y price forecast > 10%: +60.59%

• Next 5Y EPS growth estimate (CAGR) > 10%: Yes (50.0%)

• DCF Value: ~$102 (base case: revenue growth 28.2% - below consensus, 10.1% discount rate); overvalued by ~42%

• Short Interest < 5%: Yes (3.3%)

Verdict

CRDO belongs in a long-term portfolio as a small, high-risk growth position, not as a core holding. Credo created the AEC category and turned it into a very profitable business, with a 67% gross margin, a 21% ROIC, net cash, and revenue that grew 115% in the last quarter. CRDO trades at 24.1x forward earnings and a PEG of 0.48, not far from its -1 standard deviation band and slightly above the whole semiconductor sector, although analysts expect 50% annual EPS growth. I cap growth at 20% in my model and use exit multiples of 17x/22x/30x. My fair prices range from $151 to $267, with an accumulation zone of $106-151. At $176, CRDO is 16% above the top of that zone. For me, the realistic bear case is not the end of AI spending but a pause at one of the two largest customers while the optical ramp is late, as in 2023, when the stock fell 47% in one day. For this reason, the position should stay small. Accumulate gradually inside the zone with a multi-year horizon, and judge the thesis on optical revenue in the second half and the non-GAAP gross margin, not the share price.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

TL;DR: Adobe (ADBE) makes the standard software for documents and creative work (Acrobat, Photoshop, Premiere) and sells marketing software to large companies. The Q3 FY2026 report was a beat and a raise, and the business is still one of the most profitable in software. At the same time, the stock is at 9.4x forward earnings, since the market expects AI tools to replace paid Adobe seats. I see a slowdown, not a decline. Without the Semrush acquisition, annual recurring revenue (ARR) grows ~9%, and buybacks provide half of the EPS growth. Verdict: a high-quality business priced for decline. Organic growth and the operating margin still move in the wrong direction. ADBE is just above my accumulation zone; my full fair price math and the accumulation zone are at the end of this post.

Investment Thesis

Adobe reported Q3 FY2026 on Sep 10 with record revenue, a higher full-year EPS target, and record Q3 operating cash flow. Over 5 years, consensus forward EPS more than doubled, and the share price lost 62%. My idea here is to separate what is really slowing inside the business from what the price already assumes.

• Growth is slower than the headline, but it is still there. Revenue increased 13% YoY to $6.76B, and total ARR increased 11.2% to $27.50B. Both numbers include Semrush, which Adobe bought in April. Without it, I estimate revenue growth at ~11% and ARR growth at ~9%.

• The price assumes much less than consensus. The consensus 5Y EPS growth estimate is 13.92%, and ADBE trades at 9.4x forward earnings with a PEG ratio of 0.67. At a 12% required return, the free cash flow after stock-based compensation (SBC) implies only ~3.5% long-term growth.

• The business is still one of the most profitable in software. LTM gross margin is 89.25%, the FCF margin is 40.79%, ROIC is 39.19%, and FCF reached $10.59B. The diluted share count decreased by 6.8% in 1 year.

• The multiple already fell below every historical anchor. ADBE trades at 9.4x forward earnings against a 24.5x 5Y mean and a 15.2x -1 standard deviation band. And this despite the fact that consensus EPS for the next 12 months has increased every year.

• The bear case: AI tools make a paid Creative Cloud seat less necessary, organic ARR growth slides below 8%, the non-GAAP operating margin keeps declining, and a new CEO takes over on Dec 01. My model takes this into account by keeping today’s 9x multiple as the Bear Case exit. Even then, ADBE is only 5% above my Bear Case fair price of $240.

Analyst’s note:

Adobe’s fiscal year ends in late November, so FY2026 ends soon. EPS in this post is non-GAAP because the estimates are built on it. GAAP EPS is ~$18 for FY2026, mostly because of SBC. The organic numbers are my estimates. The company disclosed ~$40M of revenue and ~$480M of ARR from Semrush for its first month. Its FY2026 targets include ~$280M of Semrush revenue for 7 months, which is ~$120M per quarter.

Company Overview

Next Earnings Date: 10 Dec 2026 (estimated)
Market Cap: $100.26B
Sector: Information Technology | Industry: Software
Type: Large Value
Short Interest: 4.5%

Adobe Inc. is a software company from San Jose, California, founded in 1982 by John Warnock and Charles Geschke. The first product was PostScript - the page description language behind desktop publishing. Photoshop followed in 1990, and the PDF format in 1993. In 2013, the company stopped selling boxed software and moved the creative suite to Creative Cloud subscriptions. Subscriptions are now 97% of revenue. Shantanu Narayen has been CEO since 2007. On Dec 01, 2026, Anil Chakravarthy (at Adobe since 2020, before that the CEO of Informatica) becomes CEO, and Narayen moves to Executive Chair.

The product range splits into two customer groups. The first one is Business Professionals & Consumers ($1.91B in Q3, +16% YoY): Acrobat, Reader, Acrobat Sign, and Express (the document and quick-design tools for everyday users). The second one is Creative & Marketing Professionals ($4.65B, +13% YoY): Creative Cloud (Photoshop, Illustrator, Premiere Pro, Lightroom, InDesign, After Effects), Firefly (the generative AI studio, plus its enterprise version), and Adobe CX Enterprise for marketers (Experience Platform, Experience Manager, and GenStudio). Semrush is reported inside Creative & Marketing Professionals.

Distribution is direct (web, app stores, and inside sales) plus resellers and multi-year enterprise agreements. By geography, 59% of revenue comes from the Americas, 27% from Europe, the Middle East, and Africa, and 14% from Asia.

Market Overview

The market itself is large. At its March 2024 investor meeting, Adobe estimated its addressable market at ~$205B for 2024 and ~$293B for 2027, which is ~13% growth per year. Customer experience software is ~$155B of the 2027 figure, creative software ~$91B, and document software ~$47B. In April 2026, the company described the opportunity as expanding to new surfaces such as AI chatbots and agentic browsers. If the estimate is right, Adobe’s organic ARR grows slower than its market. The risk is not fewer users, but fewer paid seats per user.

Adobe’s audience is growing faster than its revenue. More than 1 billion people use Adobe products every month, up more than 20% YoY. The creative freemium base (Firefly, Express, and the web and mobile versions of Photoshop, Premiere, and Lightroom) crossed 100 million monthly active users (MAU), up more than 70% YoY. Acrobat and Express together are used by more than 900 million people (up more than 25% YoY), and Acrobat AI Assistant users doubled in one quarter.

The competitors grew faster from much smaller bases. Figma (FIG) grew revenue 48% YoY to $370M in Q2 2026 and guides ~39% growth for the year, but it still reported a GAAP net loss of $112M in the quarter. Canva is private. It ended 2025 with ~$4B of ARR and more than 265 million MAU. Recently, it cut its revenue growth forecast from 30% to 20%, since it slowed its AI rollout to reduce the cost of AI tasks. In addition, image and video models from OpenAI and Google (GOOGL) produce good-enough visuals for free or inside a chatbot subscription. To compare, Adobe’s LTM revenue of $25.97B is ~18x Figma’s full-year guidance.

AI revenue is small, but it grows fast. AI-first ending ARR is more than $650M, growing more than 150% YoY. That is 2.4% of total ARR. Firefly ARR from the app and credit packs grew 40% QoQ. In my view, the model matters more than the size - Adobe sells the generative features as credits and higher plans, instead of giving them away to defend the base.

The marketing business grows faster than the creative one, but part of it is bought. Ending ARR for Experience Manager, GenStudio, and Experience Platform each grew more than 20% YoY. Digital Experience subscription revenue increased 22% YoY in Q3, and I estimate ~9 points of that is Semrush. Without it, the marketing side grew ~13%, against 11% for Digital Media (the creative and document products). More than 20,000 enterprises use Adobe’s customer experience software. CX Enterprise Coworker (an AI agent for marketing workflows) had more than 1,700 customers and early adopters after it became generally available in June.

The problem is monetization, not the audience. Total ARR growth decreased from 13.8% in Q1 FY2024 to 11.2% now. The FY2026 target is 10.2%, and it already includes the ~$480M of ARR that Adobe bought with Semrush. Net new ARR in Q3 was $0.40B against $0.66B a year ago. In June, the company said it had decided "to defer previously planned Creative Cloud second half line optimizations" (planned price changes) to push the free products first. The CEO put half of the ARR impact on this decision and half on the freemium push.

Economic Moat

The LTM margins are still at or above their 5Y means: gross margin is 89.25% against an 88.48% mean, EBIT margin is 35.70% against 35.71%, net margin is 28.05% against 28.01%, and FCF margin is 40.79% against 39.47%. The quarterly trend in 2026 is weaker. The non-GAAP gross margin decreased from 90.5% in Q3 FY2025 to 89.5%, since subscription cost of revenue increased 24% YoY against 14% growth in subscription revenue. Semrush explains only a small part of it, since its gross margin was 80.5% in 2025. The company does not explain the rest, but I think the main reason is the compute cost of generative AI features.

Let’s see where this margin comes from. Most of it is switching costs. File formats: PDF is an ISO standard that Adobe created, and the PSD, AI, and INDD files of Photoshop, Illustrator, and InDesign are the working formats of design studios and print shops. Next, training: design schools teach Adobe tools, and job descriptions ask for them. Next, the bundle: Creative Cloud sells 20+ applications in one subscription, so an alternative has to replace the whole workflow, not one app. The weak point is pricing power. Adobe deferred its planned Creative Cloud price changes this year, so the moat has not been tested with higher prices in the AI era.

Figma is the clearest test of the moat. In 2022, Adobe agreed to buy Figma for ~$20B, half in cash and half in stock. In December 2023, the companies terminated the deal, since there was no clear path to approval from the European Commission and the UK Competition and Markets Authority. Adobe paid a $1B termination fee. Today, Figma is a listed competitor with 136% net dollar retention, which means its existing customers spend 36% more every year.

Worth mentioning also the enterprise contracts. Large customers sign multi-year agreements, and remaining performance obligations (RPO) of $22.16B equal 85% of LTM revenue. 67% of them are expected to be recognized over the next 12 months. Management reports “continued strength in retention” across enterprise customers, but it does not disclose a retention rate, so I cannot check it.

The marketing software is different. Adobe built most of it through acquisitions (Omniture in 2009, Marketo and Magento in 2018, Semrush in 2026), and it competes directly with Salesforce (CRM). I count it as the main source of growth, not as part of the moat.

The returns on capital still support the moat: ROIC is 39.19% (5Y mean: 30.93%), ROE is 61.90% (41.90%), and ROA is 19.73% (16.14%). Worth noting that ROE is inflated by buybacks, since $55.6B of repurchased stock reduces the equity to $11.76B. ROIC is the cleaner number here, and it decreased slightly in the last quarter after 2 years of increases.

Business Strategy

Free first, paid later. Express, Firefly, Reader, and the web versions of the flagship apps have free tiers, and the paid step is a credit pack, a higher plan, or a team seat. The company states that this builds “long-term durable growth”. The data confirms the first half of the plan. The second half, conversion into ARR, is not visible yet. On the Q3 call, Anil Chakravarthy said that engagement “is starting to translate into ARR”, but by my estimate, organic ARR growth decreased from ~10.5% in Q2 to ~9% in Q3. The newest example is Saudi Arabia, where the Ministry of Communications and Information Technology and HUMAIN make Firefly and Express available to 27 million citizens and residents.

Agents inside the flagship apps. The Creative Agent now works in Photoshop and Premiere, the Productivity Agent in Acrobat turns documents into reports, slides, and audio summaries, and CX Enterprise Coworker executes marketing workflows. Adobe also routes tasks to third-party AI models inside its own apps. What I like: Adobe positions Firefly as commercially safe, which matters for enterprises that publish the output. Disney (DIS) Imagineering is integrating Firefly Foundry into its theme park design toolkit.

Go where the documents are opened. Acrobat is now available inside ChatGPT, Chrome, Claude, Edge, and WhatsApp. Instead of waiting for users to open Acrobat, Adobe puts it inside the tools they already use. Student Spaces in Acrobat targets universities, where the next generation of business professionals learns its software.

Buy the missing pieces. Semrush (closed on Apr 28) is now part of Adobe Brand Visibility - a tool that shows brands how they appear in ChatGPT, Google AI Mode, Copilot, and Perplexity. Paid customers of the brand visibility solutions doubled QoQ. Topaz Labs (an AI video and image enhancement company with more than 1 million users) is the next deal, expected to close in Q4.

A new leadership team. In March, Shantanu Narayen said he would step down once the board found a successor. On Sep 03, the board chose Anil Chakravarthy, who manages the customer experience business and worldwide field operations. David Wadhwani, president of the creativity and productivity business and seen as a candidate, then said he would leave. CFO Dan Durn left on Jun 15 for Marvell (MRVL), and Steve Day is the interim CFO. As I see it, the board chose the enterprise marketing side to lead the company, and the creative side lost its leader in the same week.

Capital Allocation

The balance sheet moved from net cash to net debt in 9 months. Cash and short-term investments decreased from $6.60B at the end of FY2025 to $5.64B, and borrowings increased from $6.21B to $6.36B. As a result, net cash of $0.39B became net debt of $0.72B. The Semrush deal explains most of it, since buybacks already use ~90% of FCF. Adobe paid ~$1.9B, ~4x the $471M of ARR that Semrush reported at the end of 2025. In 2025, Semrush grew revenue 18% with a 12% non-GAAP operating margin. In my view, the price is reasonable, but the deal adds growth, not margin. Worth noting that Koyfin's total debt of $6.76B also includes $0.40B of lease liabilities.

Leverage is still low relative to cash flow. Debt/Equity is 57.5% against a 41.1% 5Y mean, and Debt/Capital is 36.50% against 28.57%. Interest coverage based on funds from operations (FFO) is 41.6x against a 55.0x mean. The equity is small because of buybacks, so I prefer to compare debt with cash flow - total debt is 64% of one year of FCF.

FCF reached $10.59B over the last 12 months, a 40.79% margin, against an $8.05B 5Y mean. Capex is $214M, less than 1% of revenue. Operating cash flow of $2.52B was a Q3 record.

Adobe pays no dividend and returns cash through buybacks. The buyback yield is 9.41% against a 4.67% 5Y mean. The share count decreased from ~477M in 2021 to 397.5M. In Q3 alone, the company bought 9.5M shares for $2.23B, 92% of Q3 FCF, and $24.55B remains under the April 2026 authorization (~25% of the market cap). What I like: at 9.4x earnings, each dollar of buybacks retires ~2.6x more shares than at the 24.5x mean. Worth noting that past buybacks were made at much higher prices. In FY2025, Adobe bought 30.8M shares for $11.28B (~$366 per share on average), and ADBE now trades 31% below that average.

Shareholder yield is 9.17% against a 4.42% 5Y mean. It is 0.24 points below the buyback yield, since Adobe added a small amount of debt over the last 12 months.

SBC is $2.07B, 7.97% of revenue, down from ~8.9% in 2023. The buybacks are ~4.6x SBC in dollars, so the share count decreases even after the dilution. At the same time, SBC is the main reason why the FY2026 GAAP EPS target (~$18) is ~26% below the non-GAAP one.

Operating costs grow faster than revenue this year. Over the last 12 months, sales and marketing were $7.15B, R&D $4.69B, G&A $1.92B, and cost of revenue $2.79B. In Q3, non-GAAP operating expenses increased 16% YoY against 13% revenue growth, and GAAP R&D expenses increased 18%. As a result, the non-GAAP operating margin decreased from 46.3% to 44.0%. By my estimate, Semrush explains ~0.6 points of that, since its margin is much lower. Management targets ~45% for FY2026 against 46.2% in FY2025.

In my view, the capital allocation is disciplined but not yet proven. The FY2025 buybacks lost value, and the extra spending has created users, not yet ARR.

Advantages

• The cheapest high-return company in its group. ADBE trades at 9.4x forward earnings with a PEG ratio of 0.67, while Salesforce, Intuit (INTU), Autodesk (ADSK), and ServiceNow (NOW) trade at 13.3-29.3x. Adobe also has the highest ROIC and FCF margin of the 5.

• Two standards nobody has replaced. PDF and Photoshop have been the defaults for more than 30 years, and more than 400 billion PDFs are opened in Acrobat every year. So far, AI tools have added users to Adobe’s free products instead of taking them away.

• Cash flow that pays for the AI investment. $10.59B of FCF, capex of $214M, and $22.16B of contracted revenue in RPO. Adobe increased R&D 18% YoY in Q3 and still bought back ~$9.4B of stock over the last 12 months.

• Buybacks at a low price. The diluted share count decreased 6.8% in 1 year, and $24.55B of authorization remains. Consensus expects ~9% revenue growth and ~13% EPS growth for FY2027. Most of the gap comes from buybacks.

• The derating has already happened. From ~50x forward earnings in 2021 to 9.4x, while ROIC increased from ~26% to 39.19%. The downside from further derating is smaller at 9x than it was at 25x.

Disadvantages

• AI competes for the seat. If a marketing team needs 3 Creative Cloud seats instead of 10, Adobe’s ARR decreases even if its AI features are the best. That would push revenue growth below the ~9% consensus for FY2027 and make the EPS estimates in my model too high.

• Organic growth is slowing. Without Semrush, I estimate Q3 ARR growth at ~9% and the Q4 revenue guidance at ~8% growth. JPMorgan (JPM) notes that Adobe expects net new ARR to decline for the full FY2026.

• Margins are declining. The non-GAAP operating margin decreased 2.3 points YoY in Q3, and the subscription gross margin decreased from 91.2% to 90.4%. If compute costs keep growing faster than revenue, EPS growth will come mostly from buybacks.

• New management in a difficult year. A new CEO from Dec 01, the head of the creative business leaving, and an interim CFO since June. The planned Creative Cloud price changes are deferred, and the next CEO has to decide when to bring them back.

• Net debt and GAAP earnings. Net cash became net debt of $0.72B in 9 months. On GAAP EPS, ADBE trades at 13.9x FY2026 earnings, not 10.3x. The difference is mostly SBC, which is a real cost to shareholders even when buybacks offset the dilution.

What the bears get right: the free products grow much faster than the revenue. ARR growth decreased to 11.2%, and ~2 points of that was bought with Semrush. Even an analyst with a positive rating says so. On Sep 11, JPMorgan cut its target from $340 to $315 and kept its Buy rating. The analyst pointed to 8% RPO growth and the deferred Creative Cloud pricing. If organic ARR growth falls below 7% and stays there, my 9x Bear Case multiple is too optimistic, not too conservative. In my view, the key risk is the conversion of free users into paying ones, not the quality of the AI features. Position size should reflect the chance that I am wrong here.

Competitors

I compare Adobe with 4 large application software companies. Salesforce is the leader in customer and marketing software; Autodesk has the closest business model (design software sold as subscriptions); Intuit is a quality benchmark in application software; the market treats ServiceNow as a winner of AI in software. Figma and Canva compete more directly in design, but Figma has no 5Y history as a public company, and Canva is private.

ADBE is the cheapest of the 5 on every price multiple. The forward P/E is 9.4x against 13.3x for Intuit, 16.3x for Autodesk, 16.9x for Salesforce, and 29.3x for ServiceNow. Growth does not explain the gap. Consensus 5Y EPS growth is 13.92% for Adobe, 14.02% for Intuit, 16.22% for Salesforce, 16.42% for Autodesk, and 24.45% for ServiceNow. As a result, the PEG ratio is 0.67 for ADBE and 0.99-1.20 for the others. To compare, ServiceNow is expected to grow EPS ~1.8x faster and trades at ~3.1x the multiple. The weak spot is LTM revenue growth - Adobe's 12.03% is ahead of Salesforce (11.23%) only.

On returns and margins, Adobe leads the group. ROIC is 39.19% against 26.54% for Autodesk, 17.20% for Intuit, 8.51% for Salesforce, and 6.55% for ServiceNow. Adobe also has the highest net margin (28.05%), FCF margin (40.79%), and EBITDA margin (37.56%). Autodesk has the higher gross margin (92.50% against 89.25%). Debt/Equity of 57.5% is in the middle: Intuit is at 44.3%, ServiceNow at 67.5%, and Salesforce and Autodesk are above 100%.

In my view, no other company here has a ROIC above 30%, an FCF margin above 40%, and a P/E below 10x at the same time. The next-best alternative on quality is Autodesk, with a 26.54% ROIC and 16.42% expected EPS growth, but at 16.3x earnings. There are reasons for the discount. Adobe has the most direct exposure to AI-generated images and video of the 5, and its revenue growth is the second lowest. The market pays ~29x earnings for ServiceNow, which it sees as a winner of AI, and ~9x for ADBE, since it sees Adobe's creative software as a loser. So far, the data supports a much smaller difference than that.

Past

Q3 FY2026 (ended August 28, all USD, YoY):

• Revenue +13% to $6.76B (+12% constant currency), including ~$120M from Semrush, by my estimate

• Non-GAAP EPS +15% to $6.13, while non-GAAP net income increased 7%; GAAP EPS +11% to $4.62

• Total ARR +11.2% to $27.50B; net new ARR $0.40B against $0.66B a year ago

• Non-GAAP operating margin 44.0% against 46.3% a year ago

• Operating cash flow $2.52B (a Q3 record); 9.5M shares repurchased for $2.23B

• LTM: revenue $25.97B, net income $7.29B, FCF $10.59B, and ROIC 39.19%

Why did ADBE fall 14% in the first 2 weeks of September? Most of the move happened before the report. The reasons, from the largest to the smallest:

• The leadership change. On Sep 03, the board chose Anil Chakravarthy as the next CEO, and David Wadhwani said he would leave. ADBE fell 6.73% on Sep 04 and 3.47% on Sep 08.

• The Q4 revenue guidance. The $6.825B midpoint is 0.4% below the $6.85B LSEG consensus and implies ~10% growth after 13% in Q3. Adobe beat the midpoint of its own Q3 revenue guidance by $65M but raised the full-year target by only $51M, so the implied Q4 decreased by ~$14M.

• The organic slowdown. Without Semrush, the Q4 guidance is ~8% growth, by my estimate.

• The forward indicators. RPO grew 8% YoY to $22.16B, after 13% in Q2.

• Freemium without conversion yet. The free products bring users, but ARR growth decreased, and the planned Creative Cloud price changes are deferred.

At the same time, the report itself was solid. Revenue, EPS, and both customer groups came in above the top of Adobe’s own guidance. Revenue beat the LSEG consensus by 0.9% and EPS by 0.7%, and consensus FY2027 EPS increased from $27.50 to $27.67 the day after the report. On Sep 11, ADBE closed 1.4% higher. In my view, the leadership change explains the size of the move better than the numbers do, since the report changed the implied Q4 revenue by only ~$14M.

The longer history is more important. GAAP EPS increased from $10.02 in FY2021 to $17.91 over the last 12 months, and LTM EPS growth is 11.61% against a 9.82% mean. In FY2021, GAAP EPS decreased 7% after a one-time tax benefit in FY2020, and in FY2022, it increased only 0.8%. Consensus forward EPS more than doubled over 5 years, from ~$13 to $26.89.

At the same time, the stock was a poor investment. The total return was -17.49% per year over 5 years against 12.60% for the S&P 500 (VOO), and 9.41% per year over 10 years against 15.25%. The whole difference is the multiple, which decreased from ~50x forward earnings in 2021 to 9.4x.

Future

• Consensus revenue: FY2026 $26.61B (+11.96%), FY2027 $29.02B (+9.05%), FY2028 $31.47B (+8.44%) - the slowdown is already in the estimates

• Consensus EPS: $24.48 -> $27.67 -> $31.42 (FY2026-FY2028); the FY2026 number is inside the company’s guidance, and FY2027 EPS increased 1.0% since Aug 24

• 5Y forward EPS growth estimate: 13.92%, against a 15.88% 5Y mean and below the -1 standard deviation band (14.52%); it decreased from ~17.6% at the end of 2024 to ~13% in early 2026. To compare, non-GAAP EPS grew ~14% per year from FY2021 ($12.48) to FY2025 ($20.94)

• Analysts: 23 of 40 rate the stock Hold (4 Strong Buy, 8 Buy, 23 Hold, 4 Sell, and 1 Strong Sell), average target $279.05, +10.63% from here (the range is $190.00 to $377.07)

• Company guidance: FY2026 revenue $26.576-26.626B, non-GAAP EPS $24.45-24.50, ending ARR growth 10.2% (Semrush included), and non-GAAP operating margin ~45%; Q4 revenue $6.80-6.85B and non-GAAP EPS $6.30-6.35

What has to happen for the thesis to work. Consensus expects ~9% revenue growth and ~13% EPS growth in FY2027. In my view, this is realistic if three things happen: organic ARR growth stays at ~8-9%, the non-GAAP operating margin stops declining at ~44%, and buybacks keep reducing the share count by ~5% a year. The buybacks are the most reliable part, since they depend on FCF, not on AI. The first two are the ones to check. My model assumes 13.92% EPS growth per year for 5 years, so all three have to hold well beyond FY2027. The next checkpoints are Adobe MAX in November and the December report, which normally includes the targets for the next fiscal year and will be the first one under the new CEO.

Current Valuation

Current vs 5Y mean:
Price/Fwd Earnings: 9.4x vs 24.5x
Price/Fwd Sales: 3.5x vs 9.1x
Price/FCF: 9.5x vs 25.9x
Price/Book: 8.5x vs 13.7x
PEG: 0.67 vs 1.57
Forward Earnings Yield: 9.7%

All five multiples are below their -1 standard deviation bands. The forward P/E is 9.4x against a 15.2x band, Price/Sales is 3.5x against 5.6x, and Price/FCF is 9.5x against 15.6x. ADBE trades at a 9.7% forward earnings yield and a ~10.6% FCF yield. To compare, the 4 competitors trade at 13.3-29.3x forward earnings. In my view, these yields are high for a company whose revenue still grows ~11% organically.

Analyst’s note:

The 5Y means include late 2021, when ADBE traded at 40-50x forward earnings, so the 24.5x P/E mean is inflated, and I do not use it as the Base Case. Also, Koyfin’s 9.4x uses next-12-months EPS (mostly FY2027, since the fiscal year ends in November). On FY2026 EPS of $24.48, the P/E is 10.3x. The exit multiples in my model start from today’s 9.4x.

The market cut the multiple much more than the growth. The 5Y EPS growth estimate decreased from a 15.88% mean to 13.92%, while the PEG ratio decreased from a 1.57 mean to 0.67. Even at the mean growth rate, today’s P/E would be a PEG of ~0.6.

What does the price already assume? With my 12% required return, ADBE needs ~15% annual EPS growth if the multiple stays at 9x. If the multiple recovers only to 12x, ~9% EPS growth is enough, and at 15x, ~4% is enough.

One chart frames this whole section. It applies the three exit multiples of my model (9x, 15x, and 20x) to the consensus forward EPS and compares the result with the price.

Fair Price

My model projects consensus EPS 5 years forward. It takes the risk of overly optimistic estimates into account by capping future growth at 20% per year. Here the cap does not bind. The consensus 5Y estimate is 13.92% per year, and Adobe pays no dividend, so the model uses 13.92%. Starting EPS is $24.48 (FY2026), which compounds to $46.97 in FY2031.

The discount rate is 12%; the margin of safety is 30%. The exit multiples are 9x, 15x, and 20x. 9x is today’s multiple rounded down, so this case assumes the market never re-rates ADBE, even with 14% EPS growth. 15x is the -1 standard deviation band (15.2x), where ADBE traded at the end of 2025, and a PEG of ~1.1. 20x is a PEG of ~1.4 and still below the 24.5x 5Y mean, which the 2021 period inflates.

Bear Case (exit P/E 9x): fair price $240 - MoS price $168
Base Case (exit P/E 15x): fair price $400 - MoS price $280
Bull Case (exit P/E 20x): fair price $533 - MoS price $373

At $252, ADBE trades 5% above the Bear Case fair price and 37% below the Base Case. The accumulation zone is $168-240. The model is sensitive to growth. At 10% annual EPS growth instead of 13.92%, the fair prices decrease to $201/$336/$447, and ADBE would be 25% above the Bear Case. Since growth is the least certain input here, the lower end of the zone ($168) is a more conservative anchor than the Bear Case fair price. Worth noting that the Base Case of $400 is 6% above the highest street target ($377). The corridor is more optimistic than the street on the multiple, with the same consensus EPS.

Due Diligence

Profitability (11 of 15):

• Positive Gross Profit: $23.18B

• Positive Operating Income: $9.27B

• Positive Net Income: $7.29B

• Positive FCF: $10.59B

• Gross margin >= 40%: Yes (89.25%)

• Net margin >= 10%: Yes (28.05%)

• FCF margin >= 10%: Yes (40.79%)

• Management (ROIC, ROE, ROA) >= 10%: Yes (39.19%/61.90%/19.73%)

• Strong 3Y Revenue Growth: No (~10.5% per year)

• Revenue Growth Forecast: No (~10% per year over the next 3 years)

• ROE is increasing: ~36% -> 62%

• ROIC is increasing: ~26% -> 39%

• Revenue surprises in the last 5Y in a row: No (in line in Q3 FY2022 and Q4 FY2022)

• EPS surprises in the last 5Y in a row: Yes

• EPS growth YoY 5Y in a row: No (GAAP decline in FY2021 after a one-time tax benefit in FY2020)

Financial Strength (5 of 6):

• Total assets ($29.98B) exceed total liabilities ($18.22B) by 1.65x

• Negative Net Debt: No ($0.72B of net debt, with cash and short-term investments of $5.64B against $6.36B of borrowings)

• Low Debt/Equity: 0.58 (5Y mean: 0.41)

• Debt/Capital: 0.37 (5Y mean: 0.29)

• Interest coverage (FFO): 41.6x

• Piotroski F-Score: 8 of 9 (not passed: higher current ratio YoY)

Valuation and Advantage (3 of 4):

• Valuation < 5Y mean: Yes

• Valuation < the industry: Yes (the lowest P/E, P/S, and P/FCF among the competitors)

• Does it have a moat: Yes (narrow)

• Outperformed the S&P 500 over 5Y: No (-17.49% vs 12.60% CAGR)

Shares (2 of 3):

• Insider ownership >= 5%: No (0.45%)

• Fewer shares outstanding YoY: Yes (395M vs 424M diluted)

• Insider buys in the last 6M: Yes (a director in June)

Price (4 of 4):

• 1Y price forecast > 10%: +10.63%

• Next 5Y EPS growth estimate (CAGR) > 10%: Yes (13.92%)

• DCF Value: ~$579 (base case: revenue growth 7.7% - below consensus, 8.8% discount rate); undervalued by ~56%

• Short Interest < 5%: Yes (4.5%)

Verdict

ADBE belongs in a long-term portfolio as a satellite value position, not as a core holding, until organic growth and the operating margin stabilize. The company is the standard in documents and creative software, with more than 1 billion monthly users, an 89% gross margin, a 39% ROIC, and $10.6B of FCF that goes into buybacks. ADBE is priced at 9.4x forward earnings and a PEG of 0.67, below its -1 standard deviation band on every multiple. My model uses the consensus 13.92% growth and today’s 9x multiple as the Bear Case. The fair price corridor is $240-533, and the accumulation zone is $168-240. At $252, ADBE is just above that zone. In my view, the realistic bear case is not a crash but dead money: organic ARR growth slides to ~7%, the multiple stays at 9x, and the return is mostly the buyback yield. Accumulate inside the zone with a multi-year horizon, and judge the thesis on organic ARR growth and the non-GAAP operating margin, not the share price.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

TL;DR: Uber (UBER) is the world's largest ride-hailing and delivery platform. Over 200 million people use it at least once a month, across 70+ countries. The business is in its best shape ever - bookings increased 24% in Q2, trailing free cash flow passed $10B for the first time, and ROIC reached 24%. At the same time, the stock is down 16% over the last year, at 19.7x forward earnings for ~27% consensus EPS growth - a PEG of 0.74. The market is pricing one fear: autonomous vehicles. This post is my attempt to weigh that fear against the numbers. Verdict: at $76, the stock is 20% below my Bear Case fair price, deep inside the accumulation zone. My full fair price math and the accumulation zone are at the end of this post.

Investment Thesis

The gap is simple to state. Uber's users, bookings, and free cash flow are at all-time highs, yet the stock lost 16% over the past year. Everything in this post hangs on whether the reason for that loss (the robotaxi story) deserves the price it already collected.

• Growth remains strong where it matters. Gross bookings increased 24% YoY in Q2, to a record $58.0B, and 208 million people now use the platform monthly (+16% YoY) - the company added more first-time users over the last 12 months than in any period of the last 5 years. Uber One members now account for more than 70% of delivery bookings.

• The growth is not priced in. Consensus expects ~27% EPS growth per year over the next 5 years: $3.36 this year, $4.38 next, $5.46 in 2028. UBER trades at 19.7x forward earnings against that - a PEG of 0.74.

• Quality stopped being the question. ROIC is 23.97%, ROE is 37.16%, the FCF margin is 18.32%, and Debt/Equity fell to 0.52 against a 0.88 historical mean. The company that burned cash until 2022 now buys back 4% of its shares a year.

• The derating already happened. Uber’s early P/E history is useless (the company lost money until 2023), but the readable metrics agree: P/B is at 5.8x, exactly its -1 standard deviation band; P/S is below its mean; PEG is 0.74.

• The bear case: Robotaxis remove the driver, and the driver is the reason an aggregator exists. Waymo already sells rides through its own app in San Francisco and Los Angeles, Tesla (TSLA) promises the same, and a vertical winner does not need Uber in the cities where it operates. My model takes this into account by capping growth at 20% against the 26.8% consensus, and by using today’s 20x multiple as the Bear Case exit - a world where the market never changes its mind about Uber. Even that scenario prices the stock above today’s quote. The math is in the Fair Price section.

Company Overview

Next Earnings Date: Nov 03, 2026 (estimated)
Market Cap: ~$155B
Sector: Industrials
Industry: Ground Transportation
Type: Large Growth
Short Interest: 2.3%

Uber is a marketplace. It matches riders with drivers, hungry people with restaurants and couriers, and shippers with carriers. 208 million people use the platform at least once a month, in over 70 countries. The company reports in three segments, and Q2 shows their weight: Mobility generated $29.0B of bookings (+22% YoY), Delivery $27.5B (+26%), and Freight $1.6B (+25%). Worth noting that delivery now grows faster than rides.

The product range grew far beyond a taxi app. In Mobility: everything from Moto and UberX Share at the cheap end to Black and Reserve at the expensive end. In Delivery: restaurants, groceries and retail, and increasingly advertising on top of the order flow. Uber One ties the segments together, and membership reached another all-time high in Q2.

The deal history shows the same pattern for years: buy the regional leader (Careem), buy the missing piece (Postmates, Transplace), keep equity when leaving a market (Didi in 2016, Grab (GRAB) in 2018, a ~25% stake in Aurora after selling its own AV unit in 2020).

The company is managed by Dara Khosrowshahi (LinkedIn), CEO since 2017. He came from Expedia, took Uber public in 2019, and led the company through the turn from heavy losses to real profitability. The CFO is Balaji Krishnamurthy (LinkedIn). The marketplace behind the leadership grows on both sides: a record 10.2 million drivers and couriers earned over $25B on the platform in Q2, more than 400,000 merchants participate in Uber One offers (+50% YoY), and the membership program is backed by 35+ outside partners, from Delta (DAL) and Marriott (MAR) to PayPal (PYPL) and Disney+ (DIS).

Market Overview

The core question for Uber is what autonomous vehicles (AV) do to a demand aggregator. Everything else about the company is going well, and the market knows it - which is exactly why the stock trades on this one question.

The market itself keeps growing - trips on the platform increased 18% YoY to 3.9 billion in a single quarter, and Uber remains the largest player in both categories it serves. AVs today are still a small share of the market. Waymo (the clear technology leader) has a fraction of a percent of Uber’s quarterly trip count.

Worth noting the geography. Only about 30% of Uber's US gross bookings (and ~25% of its profits) come from the top 20 cities - exactly where robotaxis arrive first. The rest of the volume lives in suburbs, smaller towns, and 70+ countries, where the AV economics do not work yet and will not for years.

The demand side of ride-hailing is not linear - it spikes at 8 a.m. and 5 p.m. A robotaxi fleet has to own enough cars for the peak (and burn capital at noon) or own too few (and lose the customer). Uber’s human drivers flex with the peak. And the company answers the fear with data: in the most mature AV markets (Los Angeles, San Francisco, Phoenix), Uber reports a higher category share than a year ago - both citywide and inside the AV operating zones. So far, robotaxis are growing the category instead of eating the aggregator. In my view, that is why the most likely outcome is not replacement but partnership - the AV companies build the cars, and the aggregator with 208 million users fills them.

The chart above is the whole situation in one picture: revenue estimates at a record, price 22% below its high. As a result, the bear narrative needs a strong argument to justify the gap.

Economic Moat

Gross margin is 40.75% against a 34.7% mean, the FCF margin is 18.32% against an 8.73% mean, and the operating margin went from deeply negative to 12.13%. The one line that fell is the net income margin - it spiked above 30% in 2025, when the one-time 2024 tax benefit was inside the trailing window, and normalized to 17.34% as the benefit rolled off. The company that lost 40 cents per revenue dollar in 2022 now keeps 17 cents of it as net income.

The moat is the network effect, and it can be measured. Uber has 208 million monthly users, and they take 18-19 trips per quarter on average. To compare, DoorDash (DASH) gets 12-15 orders per user and Lyft (LYFT) gets 8-9 rides over the same period. Uber has also averaged ~15% user growth since 2022 against ~10% at Lyft - and this despite the fact that Uber’s base is several times larger. More riders attract more drivers, more drivers cut wait times, shorter waits attract more riders. Every trip provides data for the pricing and matching algorithms, so the marketplace gets better as it gets bigger. That is the flywheel.

The returns confirm it. ROIC went from deeply negative to 23.97%, ROE to 37.16%, ROA to 6.88%. A marketplace at scale is a very good business.

Business Strategy

Mobility and Delivery share one subscription, one map, and one payment method. Uber One membership is at a record; members order more often, and ~40% of eligible members are active across both Mobility and Delivery - the subscription is the switching cost for this business. Advertising grows on top of the order flow as an almost pure-margin layer.

The largest deal in company history. Uber agreed to acquire Delivery Hero (DHER), which expands the platform to nearly 100 markets and almost doubles the number of markets where Uber offers both rides and delivery - from 34 to 58. The company counts 50 million+ newly eligible cross-platform users, and Uber One extends to all of them.

Commercialize everyone's cars, build none. Uber sold its own autonomous unit to Aurora in 2020 (keeping ~25%), and its stated ambition now is to be the commercialization platform for autonomous mobility. The numbers behind the words: AVs are already live on Uber in 7 cities, with a target of up to 15 by year-end, and partners have committed ~120,000 vehicles to the network. The 2026 deployment roster counts more than 20 companies: Waymo, Wayve, Nuro, Zoox, WeRide, and Pony.ai on the self-driving side, with carmakers and fleet operators around them. Wayve's cars received London licenses in August and launched supervised rides in the UK in early September. The logic is simple: if many AV makers compete, the aggregator picks the terms. So Uber works to make sure many compete, including through an AV Labs team that collects millions of miles of driving data each month for its partners.

Cost discipline. Sales and marketing falls as a share of revenue year after year. Stock-based compensation fell from over 7% of revenue in 2022 to 3.5%. Capital expenditure is $308M on $55B of revenue - the company owns software, not cars. And on Sep 02, Uber announced 3,300 job cuts (10% of the workforce, the largest reduction since the pandemic) and a 20% reduction in managers. The company states the goal is fewer management layers and more resources for the delivery and AV investments.

Capital Allocation

The balance sheet is solid but not excellent. Total debt is $14.7B against $5.4B of cash; Uber has a net debt position of ~$9.3B. The trend is what matters: Debt/Equity fell to 0.52 against a 0.88 historical mean, Debt/Capital is 0.34, and FFO covers interest 22.6x against a 9.0x mean. Worth noting that the debt edged up over the last year - the balance sheet improvement comes from earnings, not from repayment.

Free cash flow is the story of the decade for this company - from a trailing burn of more than $1B in early 2022 to $10B over the last 12 months. Capex is minimal, so operating cash and free cash are almost the same number.

There is no dividend. Instead, Uber buys back stock: $510M in Q2 alone, with a stated plan of directing roughly half of free cash flow to repurchases. The diluted share count declined ~4% YoY, and the trailing buyback yield is 4.2%. With SBC down to 3.5% of revenue, the buyback is real: it reduces the count instead of hiding dilution.

Worth mentioning also the Q2 exception. Buybacks were smaller than usual because ~$4B went into a ~37% economic interest in Delivery Hero - the first step of the full acquisition, with the rest funded through existing liquidity and debt, and closing targeted for the second half of 2027. Management states gross leverage stays below 2x through the deal.

Advantages

• The network effect. 200 million users taking 18-19 trips a quarter - the highest engagement in the industry, and it reinforces itself.

• Asset-light economics. $308M of capex on $55B of revenue. Uber owns the marketplace, while drivers own the cars.

• The cash machine. Trailing FCF passed $10B for the first time, an 18.3% margin, and a buyback that retires 4% of the shares annually.

• Diversification. Two large segments plus freight, 70+ countries, and only ~30% of US bookings from the top 20 cities. A robotaxi launch in one place does not touch most of the business.

• The data flywheel. Billions of trips provide data for pricing and matching algorithms that competitors cannot replicate at smaller scale - and that AV companies will need.

Disadvantages

• The AV question. Waymo and Tesla can build their own demand apps and bypass the aggregator. The Waymo relationship is already cooling: the exclusive partnerships in Austin and Atlanta expire in the first half of 2028, and the two companies lobby for opposite regulatory frameworks.

• Regulation. The business model depends on drivers being contractors, not employees. California’s Proposition 22 went Uber’s way, but there are many jurisdictions, and one unfavorable ruling can reprice the whole cost structure.

• GAAP noise. Uber’s stakes in Didi, Grab, and Aurora swing reported earnings every quarter. In Q2 alone, revaluations added a $1.6B pre-tax benefit - GAAP EPS came in at $1.17 against $0.81 Non-GAAP. Trailing EPS of $4.58 is inflated the same way, so I use consensus EPS everywhere in this post.

• Low switching costs. A rider with two apps installed pays whoever is cheaper. Uber One softens this, but the platform still has to win on price and wait time every single day.

• Ambitions cost money. Uber plans to commit over $10B (equity stakes, infrastructure, and vehicle offtake commitments) to bring AVs to market, and the Delivery Hero acquisition adds debt to the balance sheet on top. Both strain the asset-light story - the company states it is evaluating financing structures with third-party capital providers for the AV deployments.

What the bears get right: if Waymo proves the vertical model in the top places, Uber's pricing power in its densest, most profitable markets erodes first - and the market will not wait for the suburbs to confirm it.

Competitors

The competitor set depends on the segment. In US ride-hailing, it is Lyft; in delivery, it is DoorDash and Instacart (CART); internationally, it is Didi, Grab, and Bolt; in the AV scenario, potentially Waymo and Tesla. I made the comparison in numbers.

DoorDash is the one real operational rival. Consensus expects 45% EPS growth from DoorDash over the next 5 years against Uber’s 27%, and charges 34.6x forward earnings against Uber’s 19.7x. On PEG, the two are almost identical: 0.77 against 0.74. The difference is what the buyer gets today. Uber trades at 15.3x free cash flow with a 23.97% ROIC; DoorDash trades at 37.1x free cash flow with a 6.27% ROIC. At the same price per unit of growth, Uber is the one that already produces the cash and the returns.

Lyft looks like the cheapest stock in the table. The 9.2x P/E comes with negative expected EPS growth over the next 5 years, and the PEG of 2.55 is the worst in the set - the multiple is low because the earnings are expected to decline. The 42.32% net income margin with a 64.21% ROIC is not operational either - in Q4 2025, Lyft released the valuation allowance on its deferred tax assets and booked a $2.9B one-time benefit. The 0.69% EBITDA margin shows the operating reality. Uber recorded the same kind of benefit in 2024, so I treat both the same way and look at cash instead. On cash, Lyft is genuinely cheap at 5.9x P/FCF, but cheap with declining earnings is a value case, not a growth case.

Instacart is the cheapest per unit of growth. A PEG of 0.47 with a 72.58% gross margin is the best-looking line in the table. But Instacart is a one-category company, and both Uber and DoorDash are pushing into groceries with much larger user bases. The 11.4x multiple prices in the risk that the growth does not survive that competition. Grab is the opposite case - the market pays 3.0x forward sales for Southeast Asian growth with a negative FCF margin and a 1.65% ROIC today. Worth noting that Uber holds equity stakes in both Grab and Didi, so the international competition is partly owned by Uber itself.

In my view, Uber is the only company in the set that combines expected EPS growth above 25%, a ROIC above 20%, and a free cash flow multiple below 20x. DoorDash has the growth without today’s returns, Lyft has the cash without the growth, and Instacart’s discount is the price of competing with the first two. Debt changes nothing here: Debt/Equity ranges from 1.3% at Instacart to 51.9% at Uber, and every company in the table generates positive free cash flow against it.

Past

The last report (Q2 2026) was strong: gross bookings reached a record $58.0B, +24% YoY (+22% in constant currency), once again above the high end of the company’s own outlook. Revenue grew 12% to $14.2B - the gap versus bookings comes from business model changes that cut ~8 points off reported revenue growth. Adjusted EBITDA increased 33% to $2.8B (4.9% of bookings), and Non-GAAP EPS grew 35% to $0.81. The only soft spot was Brazil, the highest-volume market on the platform - a driver shortage in Moto cost ~2 points of global trips growth. The trailing 12 months: revenue $55.2B, net income $9.6B, FCF $10.1B.

The EPS chart tells the whole 5-year arc: from -$5 in 2022 to +$4.58 today. There is no smoothing in that line - the whole swing happened in about 30 months.

A 14.8% 5Y CAGR against 12.8% for the S&P 500 (VOO) - ahead, but not by what the business did. The last year explains it: -16%, driven not by results but by the AV narrative. The selling started with the February guidance (19% bookings growth at the midpoint - a deceleration) and accelerated on every Waymo headline since. Why did the stock fall on strong numbers? Because the market is not pricing this year. It is pricing the terminal value of an aggregator in a robotaxi world.

Future

• Revenue consensus: $57.96B this year (+11.42%), $66.84B in 2027 (+15.32%), $76.22B in 2028 (+14.05%).

• EPS consensus: $3.36 -> $4.38 -> $5.46 (+37.34%, +30.2%, +24.56%).

• The 5Y forward EPS growth estimate is 26.8%.

• Analysts: 51 cover the stock; the average rating is Strong Buy, the average target is $101.81 (range $70-150), which is +34% above the current price.

• Company guidance for Q3: gross bookings of $58.25-60.25B (+18-22% constant currency, with ~1 point of currency headwind) and Non-GAAP EPS of $0.84-0.88 (+28-35%), which translates to Adjusted EBITDA of $2.86-2.96B. The midpoint implies a step down from Q2’s pace - that deceleration, not the results, is what the market keeps selling.

Current Valuation

Current vs 5Y mean:
P/E (NTM): 19.7x (mean not usable)
P/S (NTM): 2.5x vs 2.7x
P/FCF (LTM): 15.3x (mean not usable)
P/B (LTM): 5.8x vs 8.7x (at the -1 SD band)
PEG (NTM): 0.74 vs 0.72
EV/EBITDA (NTM): 13.0x

Analyst’s note:

Uber’s 5Y valuation means are mostly unusable. The company was unprofitable until 2023, so the P/E “mean” of 41x and the P/FCF mean of 50x average in the loss-making years - I do not use them anywhere in my model. And I use the forward P/E, not the trailing 17x: trailing GAAP earnings include revaluations of the Didi, Grab, and Aurora stakes, so the consensus $3.36 for this year is the honest denominator. The metrics that survive the history check: P/S, P/B, and PEG.

The readable metrics agree with each other: P/S slightly below its mean, P/B exactly at its -1 standard deviation band, PEG at 0.74. For a company growing bookings 24%, none of that reads expensive.

One chart frames this whole section:

Fair Price

The model takes the consensus EPS for this year ($3.36) and grows it for 5 years. Consensus says 26.8% a year, but my model caps growth at 20% (the cap is a rule). The result is multiplied by three exit multiples and discounted back.

The discount rate is 12%; the margin of safety is 30%. The exit multiples are 20x/25x/30x: the Bear Case is today’s multiple (19.7x, rounded to 20x) - a market that never re-rates Uber; the Base Case is 25x - a PEG of 1.25 on my capped growth, since the polluted historical mean gives me nothing to anchor on; the Bull Case is 30x - a PEG of 1.5, the most my own rules allow me to pay for any stock.

Bear Case (exit P/E 20x): fair price: $95; MoS price: $66
Base Case (exit P/E 25x): fair price: $119; MoS price: $83
Bull Case (exit P/E 30x): fair price: $142; MoS price: $100

At $76, the stock is 20% below the Bear Case fair price - deep inside the accumulation zone, and below even the Base Case MoS price of $83. To say it plainly, the current quote assumes less than a never-re-rated multiple on a growth rate cut by a quarter.

Verdict: UBER belongs in a long-term portfolio as a large-cap growth holding. The business is the largest ride-hailing and delivery aggregator in the world: 208 million users, an 18% FCF margin, and a buyback of 4% of the shares a year. The stock trades at 19.7x forward earnings with a PEG of 0.74, and P/B is at its -1 standard deviation band. My corridor is $95-142, and the accumulation zone is everything up to the Bear Case fair price of $95; at $76, the price is 20% below it. The realistic bear case here is not a crash but dead money - the multiple stays where it is while every Waymo headline resets the AV debate, and the stock stays cheap until the terminal-value question gets an answer. Buy with a multi-year horizon, and judge the thesis on bookings growth and the FCF margin, not the share price.

Due Diligence

Profitability (11 of 15):

• Positive Gross Profit: $22.5B

• Positive Operating Income: $6.7B

• Positive Net Income: $9.6B

• Positive FCF: $10.1B

• Gross margin >= 40%: Yes (40.75%)

• Net margin >= 10%: Yes (17.34%)

• FCF margin >= 10%: Yes (18.32%)

• Management (ROIC, ROE, ROA) >= 10%: Mostly (23.97%/37.16%/6.88%)

• Strong 3Y Revenue Growth: ~18% per year

• Revenue Growth Forecast: ~14% per year over the next 3 years

• ROE is increasing: negative in 2022 -> 37.16%

• ROIC is increasing: negative in 2022 -> 23.97%

• Revenue surprises in the last 5Y in a row: No (2025: -0.05%)

• EPS surprises in the last 5Y in a row: No (2025: -11.71%)

• EPS growth YoY for the last 5Y in a row: No (losses until 2023)

Financial Strength (4 of 6):

• Total assets exceed total liabilities by 1.76x

• Negative Net Debt: No ($14.7B of debt against $5.4B of cash)

• Low Debt/Equity: 0.52 (historical mean: 0.88)

• Debt/Capital: 0.34

• Interest coverage (FFO): 22.6x

• Piotroski F-Score: 6 of 9 (not passed: Higher ROA yoy, Higher Current Ratio yoy, and Higher Asset Turnover yoy)

Valuation and Advantage (3 of 4):

• Valuation < mean: Yes (on the usable metrics: P/S, P/B, PEG)

• Valuation < the industry: Mixed (below DASH and GRAB, above LYFT and CART)

• Does it have a moat: Yes (narrow, network effects)

• Outperformed the S&P 500 over 5Y: Yes (14.8% vs 12.8% CAGR)

Shares (1 of 3):

• Insider ownership >= 5%: No (3.95%)

• Fewer shares outstanding YoY: Yes (-4%)

• Insider buys in the last 6M: No (the CFO bought in February)

Price (4 of 4):

• 1Y price forecast > 10%: +34% (average target $101.81)

• Next 5Y EPS growth estimate (CAGR) > 10%: Yes (26.8%)

• DCF Value: undervalued by 12% (86.85)

• Short Interest < 5%: Yes (2.3%)

Watchlist Note

The largest ride-hailing and delivery aggregator. 19.7x fwd earnings, PEG 0.74, P/B at -1SD; 5Y means unusable (losses until 2023). Consensus +27%/yr, model capped at 20% -> fair value $95/$119/$142, zone up to $95 - price deep inside, below Base MoS $83. Watch: bookings growth >= 20%, FCF margin >= 15%, Waymo/AV headlines, gig-worker rulings. Next report: Nov 03 (estimated).

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

Why trust these zones

The fair price models here are the same ones I publish in my analyses, so their track record is public. The August 4 Microsoft analysis put the fair price at $567 with the stock at ~$457 - three weeks later, the stock was $514. The July 28 NVIDIA analysis identified an accumulation zone of $181-258 while the market argued about peak AI demand - the stock is $217 today, still inside that zone, and the quarter reported since then showed +106% revenue growth. And the honest side of the same coin: the Reddit analysis identified a $149-213 accumulation range at a $160 price - the stock is $145 now, below the bottom of it. I also bought On Holding inside its zone at ~$38 - the stock is below $28. Zones are not a prediction of the bottom - they mark the price where the math stops needing optimism.

Worth noting that the same screens that feed this watchlist hold 5 of my 9 portfolio stocks today. The list below is not a separate universe - it is the queue my next buys come from.

How names enter and leave this list

A name enters in two steps. The screens find the numbers - the same quality tests my portfolio stocks pass. Then I review the business myself and build the fair price model you see in every entry. A screen finds numbers - it does not find moats. I try to find moats in my detailed analyses.

To be clear about what this list is not: a watchlist entry is not a finished deep dive. The full analysis comes before any buy - that rule stays. A name leaves through two doors only: I buy it, and it moves to the portfolio; or a quality test breaks, and I remove it with the reason stated. A name never leaves for going up in price. A stock that ran away from its zone simply waits in the list. The list holds 15 names at most, because every name has a live fair price model that must be refreshed after every report. The zones get rechecked in every monthly portfolio issue.

Explanations

The method is the one from Portfolio #002: consensus EPS, grown for 5 years at consensus growth plus the dividend yield (capped at 20% in total), multiplied by three exit P/E multiples and discounted at 12% a year. Fair price = the Base Case. Buy zone = the Bear Case fair price.

The exit multiples are never invented numbers. Each one is a real anchor from the stock’s own history: today’s multiple, the -1 standard deviation band, the 5Y mean, or the +1 standard deviation band. The Bear Case is the pessimism that already happened - the lower of today’s multiple and the -1 standard deviation band, so the buy zone assumes no recovery at all. The Base Case is the norm I actually believe - the 5Y mean, unless the mean is poisoned by a bubble or pre-profit years. Then I step down to the next anchor I trust, and I say so in the entry. The Bull Case is the most I would pay myself - and it never exceeds 30x, because at my 20% growth cap, 30x is a PEG of 1.5, the ceiling from the Manifest. When two multiples land next to each other, that is not a rounding accident. It is the model saying the derating has already happened, or that the ceiling already binds.

NFLX

Next earnings: Oct 16 (estimated)
Fair price: $122
Buy zone: up to $117
Curr Price/Fair Price: 0.68 (32% undervalued)
PEG: 1.10

Netflix (NFLX) is the largest paid streaming service in the world, with 325 million subscribers and the biggest TV entertainment audience both in the US and internationally. The model is on-demand series, films, and documentaries, without regular live or sports. The advantage is economies of scale - one content budget spreads over a base no competitor matches. And since 2022, advertising has grown as a second revenue stream, and the company expects it to roughly double this year.

The stock fell 33% over the last year while consensus still expects ~21% EPS growth over the next 5 years. NFLX is at 23.3x forward earnings - below its -1 standard deviation band (24.6x) and far below its 33.6x 5Y mean. My exit multiples are 23x/24x/30x: today’s multiple, the -1 standard deviation band, and my 30x bull ceiling. At $83, the stock is 29% below the Bear Case fair price - the scenario where the multiple never recovers at all.

What I watch: the 5Y EPS estimate. It already came down from a 25.7% mean to 21.2%; my model caps growth at 20% anyway, so the fair prices move only if the estimate falls below that.

Analyst’s note:

Netflix’s 5Y mean P/E of 33.6x includes the 2021 period at 55-60x forward earnings. In my view, that mean is inflated, so the model does not use it. The Base Case takes the -1 standard deviation band instead, and the Bull Case is capped at 30x - at my 20% growth cap, that is a PEG of 1.5, the most I allow myself to pay for any stock.

Below is the list of the other 9 entries. 5 of them are in their buy zones right now. The cheapest one trades at a PEG of 0.37, the deepest fell 56% in a year, and one is on the list precisely because I refuse to buy it at today’s price. Each entry has the fair price model, the buy zone, and the one thing that would change my mind. At the end - the whole list in one comparison table.

Read more

The track record

Since inception in December 2021, the portfolio compounds at 15.71% a year - ahead of the S&P 500 (11.37%), QQQ (14.39%), VUG (14.46%), and SPYG (14.49%). Beta is 0.95; the max drawdown is -19%. The method is unchanged and documented in issue #001: TWR returns via Portseido, percentages and not dollars, 4 benchmarks.

August was an average month: +2.50% vs +2.88% for the index. YTD, I remain behind: +4.79% vs +12.44%. The reason is the same as in #001: a large treasury block while the market keeps running. That block got smaller in August - the details are below.

Analyst’s note:

All numbers in this issue are as of August 30, 2026 - the last trading day included is Friday, August 28.

Read more

TL;DR: First Solar (FSLR) is the largest thin-film solar module manufacturer in the Western Hemisphere, and its Q2 2026 report was built on margin, not sales: earnings beat estimates by 43%, even as revenue fell 3.7% YoY to $1.06B. Gross margin reached 44.0% LTM, helped by Section 45X manufacturing tax credits and a growing mix of US-made modules. The stock trades at 10.8x forward earnings and a PEG ratio of 0.51, with $1.5B of LTM free cash flow and roughly $1.5B of net cash on the balance sheet. Strong backlog through 2030. Even so, the stock is down 14% YTD on tariff uncertainty, global module oversupply, and its dependence on US clean-energy policy. Verdict: the earnings power is improving faster than the price. My full fair price math and the accumulation zone are at the end of this post.

Investment Thesis

First Solar reported second-quarter 2026 results with earnings 43% above consensus and a fourth straight quarter of margin expansion, even as revenue fell 3.7% YoY. The stock is still down 14% YTD. Below are my main points of view.

• The growth is in earnings and backlog, not yet in revenue. The contracted backlog stands at 45.1 gigawatts, worth $13.6B through 2030, and consensus expects EPS to grow from $17.60 in FY2026 to $23.09 in FY2027 and $29.19 in FY2028, even as reported revenue guidance for 2026 is roughly flat at $4.9-5.2B.

• The market is not paying for that growth. Consensus expects a 21.3% EPS growth rate over the next 5 years, and the PEG ratio is now 0.51 - the cheapest of its solar-equipment competitors that are still growing at all.

• The margin profile keeps improving. 44.0% gross margin LTM, 32.5% net margin, 16.5% ROIC, $1.5B of FCF, and roughly $1.5B of net cash - both margins are the highest among its competitors.

• The multiple has compressed. The forward P/E is 10.8x against a 30.4x 5Y mean (inflated by 2022-23’s near-zero earnings), even as EPS estimates for 2026 through 2028 keep increasing.

• The bear case: a $60-80M net tariff headwind assumed for 2026, a pending Section 232 investigation that could affect close to 1.8 gigawatts of finished Southeast Asia capacity, intense competition from Chinese crystalline-silicon manufacturers, and a 2026 guidance built on $2.10-2.19B of Section 45X tax credits. My model takes this into account. Even then, the stock trades below my Bear Case fair price.

Company Overview

Next Earnings: 27 Oct 2026 (estimated)
Market Cap: $24.24B
Sector: Information Technology
Industry: Semiconductors and Semi-Equipment
Type: Mid Core
Employees: ~7,900
Short Interest: 9.20%

First Solar, Inc. is headquartered in Tempe, Arizona, and is the world's largest thin-film photovoltaic (PV) solar module manufacturer and the largest PV solar module manufacturer in the Western Hemisphere. It was founded in 1999 and incorporated in 2006. Unlike most of the solar industry, which relies on crystalline silicon, First Solar uses a proprietary cadmium telluride (CdTe) thin-film technology - a process that represents only about 5% of the global solar module market. Worth noting that this is also why the stock is filed under Semiconductors and Semiconductor Equipment rather than a solar-specific industry: a CdTe module is, technically, a semiconductor product.

The company designs, manufactures, and sells solar modules that convert sunlight into electricity, and it provides operations and maintenance (O&M) services to system owners after installation. Modules are its only reportable segment.

Production runs across 4 countries: the United States, India, Malaysia, and Vietnam. The international lines are being scaled back as manufacturing shifts toward the US. Domestic capacity is expected to reach about 14 gigawatts once current expansions are complete, and the company sells almost entirely to project developers, system integrators, and utility-scale renewable energy operators, with an estimated 30% historical share of the US utility-scale solar market.

Market Overview

Global solar module manufacturing is a scaled, commoditized industry. Production is heavily concentrated in Asia, with China alone responsible for roughly two-thirds of global capacity, and crystalline silicon is the dominant technology at about 95% of the market. Falling module prices have made solar economically competitive against traditional generation almost everywhere, but the module makers themselves have historically captured little of that value - the segment has a long history of thin and volatile returns on invested capital.

First Solar is outside that commodity race in two ways. First, its cadmium telluride technology is a genuinely different manufacturing process, with a simpler supply chain and no dependence on polysilicon pricing. Second, and more important, it is the largest domestic solar manufacturer in a market that US trade and tax policy has increasingly closed to low-cost imports. The Section 45X manufacturing credit pays roughly $0.17 per watt for US-made modules, and FSLR domestic factories ran at 98% utilization in Q2 2026.

The bear case says this is a policy-dependent position that disappears once incentives fade, or global oversupply pushes prices down again. The contracted backlog is the counterargument: 45.1 gigawatts worth $13.6B through 2030, with US production substantially booked through 2028. That is more revenue visibility than most cyclical manufacturers get, even if the multiple the market pays for it should stay modest.

Economic Moat

As I see it, First Solar does not have a structural moat in the traditional sense. Solar modules are a commoditized product, and crystalline-silicon competitors can copy a price cut quickly. What protects the business today is policy, not technology: trade barriers and the Section 45X credit have fenced off the domestic market from the low-cost Chinese supply that dominates the rest of the world.

What still separates First Solar from its competitors is that its cadmium telluride process is a genuinely different technology - simpler, with a shorter supply chain, and no dependence on polysilicon pricing. That gives it a lower carbon footprint and stronger yields in hot, humid climates, but it does not give it a durable cost advantage. On a pre-credit basis, FSLR cost per watt runs modestly above its crystalline-silicon competitors. Without the 45X credit, the moat argument would be much weaker.

Analyst’s note:

I am calling this a no-moat business on purpose, even though the balance sheet and the backlog are genuinely strong. A moat has to survive a change in policy, and First Solar’s current advantage mostly does not.

Contracted demand is the more durable protection. The backlog runs 45.1 gigawatts and $13.6B through 2030, and US production is substantially booked through 2028. Developers pay a premium for that certainty: the average selling price on new US bookings was $0.36 per watt in Q2 2026, with room to charge more through so-called technology adjusters as CuRe and perovskite improvements roll out.

The returns on capital confirm the improvement, even without a classic moat. ROIC is 16.5% against a 14.0% 5Y mean, ROE is 18.5% against 16.2%, and ROA is 8.6% against 7.3%. All three increased sharply from below 6% in 2023 to today's levels, as the tax-credit-driven margin expansion flowed straight to the bottom line. Among its solar-equipment competitors, only Nextpower (NXT) does better on this measure, at 23.6% ROIC; Enphase (ENPH), Sunrun (RUN), and SolarEdge (SEDG) are all in the mid-single digits or negative.

Business Strategy

Here, I would like to define four choices that create the strategy. First, US-first manufacturing. Domestic capacity is on track to reach about 14 gigawatts once current expansions are complete, and a new finishing facility in South Carolina will let the company complete modules started at its international plants inside the United States. Management gives priority to integrated US factories when allocating production, which reduces dependence on short-cycle international demand.

Second, contract discipline over volume. Management continues to focus on pricing, contract quality, and balanced risk terms rather than booking volume for its own sake. The result is the backlog of 45.1 gigawatts - demand from large corporate energy users and data-center customers adds to that pipeline.

Third, a technology roadmap built around CuRe and perovskites. FSLR has started customer notifications tied to CuRe contract adjusters, the first step in converting performance improvements into future revenue. Management says factory and field results have exceeded expectations across several climates. The company is also advancing a perovskite program, with a Series 6 pilot line expected in the first half of 2027 - the next differentiation wave once crystalline silicon closes the current efficiency gap.

Fourth, capacity as the only real margin lever. Roughly 80% of operating expenses are fixed, and long-term pricing power on the panels themselves is limited, so further capacity growth (not price increases) is what expands operating margins from here. Every gigawatt added toward the 14-gigawatt US target flows mostly to the bottom line once it is running.

Capital Allocation

The balance sheet is very conservative. Total assets are $13.4B against $10.3B of equity and just $3.1B of liabilities. Cash and short-term investments are $1.7B against only $194M of total debt - net cash of roughly $1.5B. FFO interest coverage is 55.8x against a 43.5x 5Y mean. Debt/Equity is just 1.9%, next to 51.9% for Enphase, 98.9% for SolarEdge, and 292.5% for Sunrun. Whatever tariffs or module pricing do next, First Solar is not a solvency risk.

FCF only turned reliably positive in 2025, after years of funding capacity expansion from the balance sheet. LTM FCF is $1.5B, a 27.9% margin, against a 5Y mean of -6.1%. The company was still burning cash as recently as 2023 and 2024 while it built out its US and India capacity. There is no dividend and effectively no buyback.

Capex is funding the US and India build-out: $655.6M LTM, down from a peak above $1.5B in early 2025 as the domestic expansion phase matures. R&D spending is comparatively small at $269.7M LTM - this is a manufacturing-scale business, not an R&D-intensive one. Stock-based compensation is only 0.44% of revenue LTM, low even by the standards of a capital-intensive industry.

Shares outstanding have increased slightly to 107.47M on modest stock-based compensation.

Advantages

• A strong balance sheet in a cyclical industry. Net cash of roughly $1.5B, FFO interest coverage of 55.8x, and almost no debt. FSLR has the resources to outlast competitors that have failed in past solar-industry downturns.

• Demand visibility most manufacturers do not have. A 45.1 gigawatt, $13.6B backlog running through 2030, with US production substantially committed through 2028.

• US policy has fenced off its core market. Section 45X tax credits worth $2.10-2.19B are built into 2026 guidance, trade barriers keep low-cost imports out, and the company holds an estimated 30% share of the US utility-scale market.

• Cheap on every basis. 10.8x forward earnings, a PEG of 0.51, a Price/Book of 2.3x, and a 9.3% forward earnings yield - all near the low end of their own history. The PEG is the cheapest among competitors still posting positive growth.

• Margins keep expanding. 44.0% gross margin LTM against a 31.4% 5Y mean, 32.5% net margin against 20.1%, and a 16.5% ROIC that is still rising - both margins are the highest of its competitors.

Disadvantages

• Earnings depend on a tax credit that could change. 2026 guidance assumes $2.10-2.19B of Section 45X credits - a large share of projected profitability. Any reduction in incentive value or added compliance complexity would reduce margins directly.

• No structural cost advantage. On a pre-credit basis, First Solar’s cost per watt runs modestly above its crystalline-silicon competitors, and the broader industry is fiercely competitive and commoditized. SolarEdge’s own swing from a deeply negative gross margin in early 2025 back to positive within a year shows how quickly conditions can turn even for a US-listed equipment name.

• Tariffs cut both ways. 2026 guidance assumes a net $60-80M tariff impact. A pending Section 232 investigation could affect close to 1.8 gigawatts of finished Southeast Asia capacity, and the company is already carrying about $30M of quarterly underutilization costs there.

• Revenue is flat to down in 2026. Guidance of $4.9-5.2B implies a roughly flat year after the Southeast Asia capacity pullback. This year’s story is funded by margin and tax credits, not volume.

• Little cushion in a drawdown. No dividend and effectively no buyback; a Beta of 1.75 and 1Y volatility near 54% mean large swings around every catalyst.

What the bears get right: a 9.2% short interest is real money betting the policy tailwind fades faster than the market expects. Their case is straightforward: the solar module industry has a long history of poor returns on invested capital, Chinese crystalline-silicon manufacturers have far more capacity than global demand needs, and roughly a third of First Solar’s own manufacturing footprint is in Southeast Asia, exactly where US tariff policy is least settled. If Section 45X credits shrink or a Section 232 ruling goes the wrong way, the earnings growth behind today’s cheap multiple could prove temporary. As I see it, the balance sheet and the contracted backlog are real protections, but the position only works if the policy backdrop holds, and that is not something a spreadsheet can fully price in.

Competitors

The competitors set here is: Enphase Energy (the quality benchmark, though its own growth has stalled), SolarEdge Technologies (a turnaround still working through a brutal 2024-2025 drawdown), Sunrun (the residential installer, financed with far more debt than the rest of the group), and Nextpower (a smaller name with the best returns on capital of the 5).

The obvious comparison set of large crystalline-silicon module manufacturers like JinkoSolar and Canadian Solar has too little fresh data on the tools I use for this series, so this is the US-listed solar-equipment group.

On forward P/E, First Solar is at 10.8x. Sunrun is lower at 9.3x; Enphase is at 20.3x, Nextpower at 21.1x, and SolarEdge at a distorted 84.3x - a symptom of earnings that only recently turned positive again.

Consensus 5Y EPS growth tells the more interesting story: First Solar's 21.3% is the highest in the group. Nextpower is next at 15.9%, and then the numbers turn negative: Sunrun's is roughly flat at -1.0%, Enphase's is -9.5%, and SolarEdge's is -92%. As a result, FSLR's PEG of 0.51 is the cheapest growth-adjusted multiple among the names still actually growing; Sunrun is at 0.76 and Nextpower at 1.33. Enphase and SolarEdge's PEG numbers are not meaningful here, since a PEG built on shrinking or near-zero forward earnings does not measure the same thing.

On returns on capital, Nextpower actually leads the group at a 23.6% ROIC, though that is down sharply from above 50% in 2022 as its base of invested capital has grown. First Solar is next at 16.5% and rising; Enphase is at 4.9%, Sunrun is close to breakeven at 0.3%, and SolarEdge is destroying capital at -17.1%, still recovering from its drawdown.

First Solar's Debt/Equity is 1.9%, and Nextpower's is 1.5% (both essentially unlevered) against 51.9% for Enphase, 98.9% for SolarEdge, and 292.5% for Sunrun.

On margins, First Solar also leads: 44.0% gross margin and 32.5% net margin, both the highest of the 5.

In my view, First Solar is not the best business in this group by ROIC alone - Nextpower's returns are still better. But it is the only one that combines real growth, the highest margins, and a sub-1 PEG all at once.

Past

Q2 2026 (ended June 30, all USD, YoY):

• Net sales $1.06B, down 3.7% (missed consensus by 0.4%)

• EPS $3.92, beat consensus by 43.1%, up 23.3% from $3.18

• Gross profit $605M, up 21% from $499.9M; gross margin expanded about 12 points to 57% for the quarter

• Adjusted EBITDA $644M, up 15%, with margin expanding 990 basis points to 60.9%

• Cash and equivalents $1.69B, down from $2.80B at year-end 2025; no long-term debt, versus $282.6M at year-end 2025

• Contracted backlog: 45.1 gigawatts worth $13.6B through 2030, with 1.9 gigawatts of new US bookings during the quarter at $0.36 per watt

Why is the stock still down 14% YTD, after a quarter like this? I see three reasons, in order of importance:

• The 2026 top line is flat to down. Guidance of $4.9-5.2B implies a roughly flat-to-negative year after the Southeast Asia capacity pullback, and Q2 sales themselves missed consensus and fell 3.7% YoY.

• The policy overhang never really lifted. A $60-80M net tariff impact, a pending Section 232 investigation, and a 2026 guidance built on $2.10-2.19B of Section 45X credits keep the stock trading on Washington news as much as on its own numbers.

• Cash on the balance sheet fell sharply on paper. $1.69B at the end of Q2, down from $2.80B at year-end 2025, even though this mostly reflects the normal seasonal timing of credit monetization and capacity spending rather than a change in the business.

Worth noting also what a headline-only read misses: gross margin expanded 12 points, and adjusted EBITDA grew 15% despite the sales miss, and the backlog still runs through 2030. A business shedding revenue for structural reasons does not usually see margin expand like this.

The longer history matters more. LTM revenue reached $5.38B, net income $1.75B, and FCF $1.50B. Every margin is above its 5Y mean: gross margin 44.0% against 31.4%, net margin 32.5% against 20.1%, and FCF margin 27.9% against a negative 6.1% mean - the company was still burning cash on a trailing basis as recently as 2023 and 2024.

Total return of 162.16% over 5 years (21.26% per year) against 90.01% for VOO (13.70% per year); and the gap holds over 10 years too, at 16.97% per year against 15.42%. FSLR has already beaten the market by a wide margin, and at 10.8x forward earnings, it does not look like a stock that needs to keep doing that to still be attractively priced.

Future

• Consensus revenue: FY2026 $5.05B (-3.31%), FY2027 $5.91B (+17.07%), FY2028 $6.53B (+10.49%) - the flat year is 2026 only

• Consensus EPS: $17.60 -> $23.09 -> $29.19 (FY2026 - FY2028)

• 5Y forward EPS growth estimate: 21.30%, almost exactly in line with its own 20.41% mean - the growth outlook itself has held steady while the multiple compressed

• Analysts: Buy (9 Strong Buy / 15 Buy / 10 Hold / 1 Sell / 1 Strong Sell, 36 covering), average target $270.97, +20.13% from here (the range is $150.00 to $402.00)

• Company guidance: FY2026 sales of $4.9-5.2B, gross profit of $2.4-2.6B, module shipments of 17-18.2 gigawatts, and capital expenditure of $0.8-1B

Current Valuation

Current vs 5Y mean:
Price/Fwd Earnings: 10.8x vs 30.4x
Price/Fwd Sales: 4.4x vs 4.1x
Price/FCF: 16.2x vs 27.3x
Price/Book: 2.3x vs 2.6x
PEG: 0.51 vs 2.23
Forward Earnings Yield: 9.3%

Most multiples here are at or below their 5Y mean. Price/Book is 2.3x against a 2.6x mean, and Price/FCF is 16.2x against a 27.3x mean. On a Price/Sales basis, FSLR is actually slightly above its own mean, at 4.4x versus 4.1x - the one multiple here that has not compressed. First Solar now trades at a 9.3% forward earnings yield. To me, that is a high yield for a business consensus expects to grow EPS at over 20% a year.

Analyst’s note:

The 5Y mean here is distorted by 2022 and early 2023, when First Solar’s earnings were near zero. A near-zero denominator sent the forward P/E and PEG to extreme, meaningless levels for several quarters, which drags the 5Y average up to 30.4x and 2.23 respectively. I do not use those means anywhere in my model. The exit multiples in the Fair Price section start from today’s levels, not from the distorted average.

One chart frames this whole section. The corridor below takes the consensus forward EPS estimate and multiplies it by my 3 exit multiples: 11x, 15x, and 20x.

Fair Price

My standard model projects EPS 5 years forward and takes the slowdown risk into account by capping future growth at 20% per year. Consensus for the next 2 years runs above that (+31.2% for FY2027, +26.4% for FY2028), so my projection is below the consensus average. There are no dividends to add. Starting EPS is $17.62 (FY2026), which gives $43.84 in FY2031.

Analyst’s note:

Consensus moved to $17.60 for FY2026 after I had built the model, which starts from $17.62. The 0.1% difference is well within the 30% margin of safety, so I did not rebuild it. On the newer number, the Bear Case is $273 instead of $274.

The discount rate is 12%; the margin of safety is 30%. The exit multiples are 11x, 15x, and 20x. 11x is close to where the stock already trades today - the current NTM multiple is actually a touch below even that. 15x is a modest re-rating, still well under the 30x-plus levels First Solar traded at during 2022-23’s near-zero-earnings stretch. 20x assumes First Solar earns something closer to a quality-industrial multiple once the policy backdrop proves durable over a full cycle. The 30.4x historical mean plays no role, because it is distorted by the near-zero earnings of 2022 and early 2023.

Bear Case (exit P/E 11x): fair price $274 - MoS price $192
Base Case (exit P/E 15x): fair price $373 - MoS price $261
Bull Case (exit P/E 20x): fair price $498 - MoS price $348

At $225.56, the stock trades 18% below the Bear Case fair price, inside the accumulation zone of $192-274. Worth noting that my Base Case of $373 is still below the highest street target of $402.

Verdict: First Solar belongs in a long-term portfolio as a policy-levered value-and-growth position - not a core holding, and sized like a cyclical industrial with real regulatory risk attached. The company is the largest thin-film solar module manufacturer in the Western Hemisphere, protected less by technology than by US trade and tax policy. The stock trades at 10.8x forward earnings and a PEG of 0.51, among the cheapest multiples in its competitor group despite having the best margins of the 5. My model’s corridor runs $274-498, with an accumulation zone of $192-274; at $225.56, the stock is inside that zone. I’ve owned the company for more than 1.5 years and am going to hold it further. The realistic bear case here is not a demand decline but a policy shock - a cut to Section 45X credits or an adverse Section 232 ruling would reduce earnings directly, since a large share of 2026’s projected profitability runs through the tax credit line rather than the module business itself. Buy with a multi-year horizon, and judge the thesis on the durability of US manufacturing policy and the pace of the backlog converting into shipped, credit-qualified gigawatts - not the share price.

Due Diligence

Profitability (10 of 15):

• Positive Gross Profit: $2.37B

• Positive Operating Income: $1.81B

• Positive Net Income: $1.75B

• Positive FCF: $1.50B

• Gross margin >= 40%: Yes (44.0%)

• Net margin >= 10%: Yes (32.5%)

• FCF margin >= 10%: Yes (27.9%)

• Management (ROIC, ROE, ROA) >= 10%: Mostly (16.5% / 18.5% / 8.6%)

• Strong 3Y Revenue Growth: ~26% per year

• Exceptional Revenue Growth Forecast: Mixed (FY2026 guidance is down 3.3%, but FY2027-2028 consensus implies ~14% CAGR)

• ROE is increasing: ~7% -> 18.5%

• ROIC is increasing: ~6% -> 16.5%

• Revenue surprises in the last 5Y in a row: No (Missed: 2021, 2022, and 2023)

• EPS surprises in the last 5Y in a row: No (Missed: 2024 and 2025)

• EPS growth YoY 5Y in a row: No (Decline in 2022)

Financial Strength (6 of 6):

• Total assets ($13.39B) exceed total liabilities ($3.07B) by 4.4x

• Negative Net Debt: -$1.54B (cash and short-term investments of $1.73B against $194M of debt)

• Low Debt/Equity: 0.02 (5Y mean: 0.07)

• Debt/Capital: 0.02 (5Y mean: 0.07)

• Interest coverage (FFO): 55.8x

• Piotroski F-Score: 8 of 9

Valuation and Advantage (2 of 4):

• Valuation < 5Y mean: Yes

• Valuation < the industry: Mixed (cheaper than most named competitors on forward P/E, but a forward Price/Sales of 4.62x that runs above the broader solar sub-industry’s 2.13x)

• Does it have a moat: No

• Outperformed the S&P 500 over 5Y: Yes (21.26% vs 13.70% CAGR)

Shares (1 of 3):

• Insider ownership >= 5%: Yes (5.54%)

• Fewer shares outstanding YoY: No (slight, steady dilution)

• Insider buys in the last 6M: No

Price (2 of 4):

• 1Y price forecast > 10%: Yes (+20.13%)

• Next 5Y EPS growth estimate (CAGR) > 10%: Yes (21.30%)

• DCF Value: $205, overvalued by 10%

• Short Interest < 5%: No (9.20%)

Watchlist Note

Largest thin-film solar module manufacturer in the Western Hemisphere. No economic moat but a strong balance sheet: net cash ~$1.5B, Debt/Equity 0.02, FFO coverage 55.8x. Gross margin 44.0%. Backlog 45.1GW/$13.6B through 2030. Trades at 10.8x forward earnings, PEG 0.51, 9.3% forward earnings yield. The best margins in its competitor group at one of the cheapest multiples. Down 14% YTD. Fair price corridor $274-498; accumulation zone $192-274. Risks: heavy dependence on Section 45X tax credits ($2.10-2.19B of 2026 guidance), tariff and Section 232 uncertainty, no structural cost advantage over crystalline silicon, no dividend or buyback.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

TL;DR: On (ONON) reported its highest gross margin ever and raised its margin outlook: 65.4% (best in sportswear), with full-year gross margin guidance raised to 65%. Still, the stock fell on weaker US demand, HOKA (DECK) and a rebirth of Nike (NKE), the strong franc, and 6.4% short interest. It now trades at 17.3x forward earnings against a 53x mean since IPO. Consensus expects ~25% annual EPS growth over the next 5 years; the PEG ratio is now 0.70. The strategy is premium-only, DTC-first, beyond running shoes, and sustainability. On sells its shoes at full price; the innovation pipeline and the athletes support the premium prices. ROIC is 17.1% and rising while the company is still building stores and warehouses. I see a running brand that becomes a sportswear house with a young audience. The stock trades below my Bear Case fair price, inside the accumulation zone. The full model is at the end of this post.

Investment Thesis

On reported Q2 2026 results with 21.6% constant currency growth, its highest gross margin ever, and a raised margin outlook. As a result, the stock fell 14% on the report; YTD decline is 33%. The business and the price now tell two different stories, and this post is my intention to explain which one is correct and why.

• Growth remains strong where it matters. Direct-to-consumer (DTC) grew +34.3% constant currency, Asia-Pacific +54.7%, and full-year guidance expects low-20% constant currency growth with gross margin raised to 65%.

• Valuation is low relative to growth. Consensus expects ~25% annual EPS growth over the next 5 years. The PEG ratio is now 0.70.

• The fundamentals remain strong. 65.4% gross margin (best in sportswear), 12.3% net margin, 17.1% ROIC, $542M of FCF, and ~$0.8B of net cash.

• The stock derating has already happened. 17.3x forward earnings against a 53x mean since IPO - around the -1 standard deviation band. Despite the fact that forward EPS estimates have continued to rise.

• The bear case: weaker US demand, HOKA and a rebirth of Nike, the strong franc, and 6.4% short interest. My model takes this into account by capping future growth at 20%. Even then, the stock trades below my Bear Case fair price.

Analyst’s note:

All prices and multiples in this post are in USD, because ONON is the NYSE listing. On itself is a Swiss company and reports in CHF.

Company Overview

Next Earnings Date: 10 Nov 2026 (estimated)
Market Cap: $10.36B
Sector: Consumer Discretionary | Industry: Textiles, Apparel and Luxury Goods
Type: Mid Growth
Short Interest: 6.4%

On Holding AG is a Swiss premium sportswear company from Zurich. It was founded in 2010 by former professional triathlete Olivier Bernhard together with David Allemann and Caspar Coppetti. Roger Federer has been an investor and product partner since 2019. CloudTec cushioning is their core technology. Newer technologies include the LightSpray one-piece upper (a robot sprays the shoe in minutes) and the SURREAL superfoam that debuts in the Cloudsurfer 3 later this year.

The product range covers road running (Cloud, Cloudsurfer, Cloudmonster, Cloudrunner), racing (Cloudboom Strike, LightSpray), trail (Cloudvista, Cloudultra), tennis (THE ROGER line), training shoes, lifestyle (Cloudtilt, Cloudnova), plus a fast-growing apparel and accessories business. Distribution is a mix of premium wholesale and DTC: own stores, e-commerce, and the app. DTC is 45.7% of sales and rising - the main source of the high gross margin.

Market Overview

The primary question I see for On is whether the Q2 slowdown is real demand weakness or mostly a temporary effect. The answer determines if 17x forward earnings is an investment opportunity or rather a fair price for a slowing business.

Global sportswear is a slow-growth market where the leader stumbled. Nike spent 2 years fixing itself and is only now recovering; Adidas is still in a turnaround. The premium running niche they both neglected was taken over by two brands: HOKA and On. From my perspective, the difference between those two matters a lot. HOKA scaled fast through run-specialty stores and is now decelerating - consensus expects only 7.2% EPS growth for Deckers (HOKA) over the next 5 years. On took the other (slower) road with its own stores, its own app, and full price. And On is the one still compounding at 20%+.

The brand is still early in its growth. On sells in 90+ countries but has sold only 50M+ pairs of shoes in its entire history. To compare, Nike sells that amount in a season. Global brand awareness is only 30%. This is the growth story - most future customers have not met the brand yet. Where awareness is already high, the results follow. Asia-Pacific grew +54.7% constant currency in Q2 (now over 20% of sales) - Japan, South Korea, and Greater China all performed well. The first stores in Sao Paulo and Copenhagen were opened only weeks ago. I see there is still a lot of room to expand geographically.

The customer mix indicates that we are still in the early stages. Consumers under 34 buy the Cloudtilt as streetwear, not as running equipment. Apparel grew +56.2% constant currency and accessories +102.2%, both from small bases. This is how a running brand becomes a sportswear house: shoes build the brand, apparel monetizes it at scale. To compare, Lululemon (LULU) needed 20 years for the same transition in the opposite direction.

Shoes became a fashion category with a performance core. The winners are the brands that own a recognizable silhouette. On owns two (the Cloud and the Cloudtilt) and sells them at full price, while most of the industry discounts.

Analyst’s note:

Personally, I run regularly. Right now my primary running shoes are from Adidas, but I’ve been looking at On for a long period of time.

Economic Moat

As I see it, the moat is narrow but real. It is shown in one number: 65.4% gross margin. To compare, Nike is in the mid-40s, Deckers (HOKA) in the mid-50s, Lululemon just under 60.

Let’s see what produces this margin. Full-price discipline: On sells out instead of marking down, even in a promotional market. Next, the DTC mix of 45.7% of sales removes the middleman on nearly half the business.

The innovation pipeline protects the price point. CloudTec created the brand in 2010. The first prototype was pieces of garden hose glued to an outsole; the improved version won the ISPO BrandNew award before the company had real revenue. Today the stack includes thirteen named technologies: the CloudTec family (Phase, Sphere, Connect), the Helion superfoams, the Speedboard plate, Missiongrip for trail, and now SURREAL, which ships in the Cloudsurfer 3 later this year. A premium brand needs a reason to charge premium prices - On keeps creating new ones.

LightSpray is a special case. I see it as a manufacturing moat, not just a shoe. What is it? A robot arm sprays the entire upper in minutes; there is no stitching and no assembly line. The process produces up to 65% less CO2 than a conventional racing upper; production can happen close to the customer instead of in an Asian factory. A foam recipe can be copied in a season, but a robotic production process is much harder to copy. LightSpray started as an elite racing product and now is rolling into core franchises like the Cloudmonster.

Worth mentioning also the athletes. On Labs in Zurich puts elite athletes in the same room as the engineers. Hellen Obiri wins marathon majors in the shoes, Kristian Blummenfelt validates them in triathlon, the On Athletics Club builds track credibility, and Iga Swiatek and Ben Shelton carry the tennis line after Federer’s playing career. For a performance brand, podium wins are the best advertising. The under-34 audience that buys the Cloudtilt follows the same athletes on social media.

Every margin line confirms the model works: gross margin 64.8% LTM against a 59.7% 5Y mean, EBIT margin 13.8% against 5.0%, net margin 12.3% against 2.0%, FCF margin 13.6% against 2.7%. All four are at or near record highs and continue to rise.

The returns on capital show the same trend: ROIC is 17.1% against a 14.2% 3Y mean, ROE is 24.0% against 13.8%, ROA is 9.6% against 7.7%. ROIC increased from ~11% to 17.1% in 3 years while the company was still building stores and warehouses. The 24% ROE comes with very little leverage (Debt/Equity is only 0.30). Every franc On reinvests creates value thanks to a cost of capital around 9-10%. To compare, Deckers (HOKA) is still ahead at 33.8% - that is the level On's returns can grow into as the store base matures.

Business Strategy

Here, I would like to define four choices that create the strategy. First, premium-only - On does not chase volume. Wholesale sell-in is deliberately managed, discounts are rare, and management repeats in every call that they only pursue growth that protects and promotes the brand. This quarter proved their statement - they accepted +4.8% in wholesale instead of pushing extra volume into the channel.

Second, DTC-first. Management describes its own stores as “highly profitable premium brand hubs”. Worth noting that the store network is still small by competitors’ standards - the Sao Paulo and Copenhagen openings were the first ones in those markets (15 years into the company’s life). Every point of DTC mix adds gross margin. At 45.7% of sales, there is still a long way to go.

Third, beyond running shoes. Apparel and accessories are the fastest-growing product lines. At the same time, tennis is becoming a real franchise: Federer’s THE ROGER line started it, and the current team keeps it visible on tour. Though these businesses are still small next to running shoes. Each new category is sold through stores and the app that the shoe business already paid for.

Fourth, the next generation of customers. I do not buy stocks for sustainability reports, but I pay attention when sustainability is a sales argument. Their stated goal is fossil-free materials with circular systems. The LightSpray emissions cut is a product feature; Cyclon is a subscription where the customer returns the shoe for recycling. Together with the Right to Run community program and a memorable mission (“ignite the human spirit through movement”), this is a brand that the youngest cohort of buyers picks for its responsibility. What I like: that cohort is already more than a third of On’s customers.

Capital Allocation

The balance sheet is simple and conservative. Cash and short-term investments of $1.49B against $696M of total debt - that means roughly $0.8B of net cash. Debt/Equity is only 0.30, and FFO interest coverage is 21.8x. There is no dividend and no buyback; the share count grows slightly (buyback yield is -0.12%). Worth noting that stock-based compensation is only 2.25% of revenue, down from over 20% in the IPO era. In my opinion, the dilution problem is solved.

FCF of $542M funds the store network and inventory without new debt. The FCF history shows the maturation of the business: negative through 2022 - 2023 while the company built inventory and logistics; positive and growing from 2024.

Advantages

• The best growth and margin combination among the competitors. Consensus expects ~25% EPS growth over the next 5 years, and the 65.4% gross margin is above every major sportswear competitor. Normally, an investor pays for one of those. Here both come in one stock.

• Margin expansion has a built-in driver. DTC is 45.7% of sales and rising; every extra point of the mix adds gross margin. The store network is still small by competitors’ standards, so this driver has years left.

• The balance sheet is safe. Net cash of ~$0.8B, FCF margin of 13.6%, interest coverage of 21.8x. Whatever the consumer does the next year, On is not a solvency risk.

• The brand is still early. 90+ countries but only 30% global awareness; 50M+ pairs sold in the company’s history; apparel and accessories growing 50%+ constant currency from small bases. Most future customers have not met the brand yet.

• The derating has already happened. From 200x+ at IPO to 17x forward, while net income margin increased from negative to 12.3% and ROIC increased to 17.1%.

Disadvantages

• A consumer cyclical in a promotional market. Americas at +4.5% reported is the weaker part, since shoes are discretionary spending. A potential US recession would hit both volume and the wholesale channel.

• Currency and tariffs. On reports in CHF, sells mostly in dollars, and absorbs US import tariffs. The strong franc removed about 8 points of reported growth this quarter and can keep doing it. This does not touch the business quality, but it does touch the headline numbers the market trades on.

• Competition is a meaningful risk. HOKA is bigger in run specialty, Nike is recovering, and every brand now sells a thick-cushion shoe. The Cloudtilt is a fashion item, but fashion cycles end eventually.

• Nothing supports the stock in a drawdown. No dividend, no buyback, slight dilution - when sentiment turns, there is no mechanical buyer. The 33% YTD fall during a year of 20%+ growth shows how this feels.

• The stock is volatile. 1Y volatility of ~50, short interest of 6.4%, and a 22% intraday swing on earnings day. Position sizing is critical here.

What the bears get right: a 6.4% short interest is serious money betting that the slowdown is real. Their case is simple. The Americas grew only 4.5% reported in the most important shoes market; “wholesale discipline” is what every brand says when sell-in gets soft. Deckers (HOKA) already showed how this story can go: HOKA decelerated, the multiple fell to 12x and stayed there. If On’s constant currency growth slides to the mid-teens, my 17x Bear Case is not a floor but a ceiling. As I see it, the real risk is the growth, not the franc. Q2’s DTC +34.3% constant currency and the raised margin guidance tell me the brand is fine. Even then, the position must be sized for the chance that this view is wrong.

Competitors

The competitors set here are: Nike (the giant), Deckers (HOKA, the closest rival), Lululemon (the premium athleisure benchmark), and Adidas. On forward P/E, ONON is at 17.3x and in the middle of the group; Nike is at 23.7x, Adidas at 15.3x, Deckers at 12.1x, and Lululemon at 11.1x.

Consensus 5Y EPS growth: On 24.8%, Adidas 21.5%, Nike 13.2%, Lululemon 9.3%, Deckers 7.2%. As a result, On's PEG of 0.70 is the cheapest in the group. To compare, Nike trades at 1.80 per unit of growth and Deckers at 1.69. Deckers and Lululemon are cheap because their growth stalled. ONON is priced near them while growing 3x faster.

On returns on capital, Deckers leads with a 33.8% ROIC, Lululemon is at 20.9%, ONON is at 17.1% and rising, and Adidas and Nike are in the low teens. On Price/Sales, the whole group has compressed: ONON at 2.2x is nearly level with Deckers at 2.1x; Nike and Lululemon are at 1.3x. In my view, On is not the best business here yet (Deckers is), but it is the one improving fastest. And the only one that grows above 20% while priced like the slow ones.

Past

Q2 2026 (ended June 30, all CHF, YoY):

• Net sales +13.5% to 850.3M (+21.6% constant currency); DTC +26.0% to 388.4M (+34.3% cc)

• By region: EMEA +15.4%, Americas +4.5%, Asia-Pacific +43.1% (cc: +20.5% / +13.0% / +54.7%)

• Apparel +47.7%, accessories +88.3% - the beyond-shoes business is compounding

• Gross margin 65.4%, up 3.9 points, with US tariffs absorbed and zero refunds booked

• Net income 105.0M against a 40.9M loss a year ago; adjusted EBITDA margin 19.8%

• Cash increased by 18% to 1,205.6M

Why did the stock fall 14% on these numbers? I see five reasons, in order of importance:

• The FX effect. Net sales grew 13.5% as reported but 21.6% in constant currency. The Swiss franc got stronger, and roughly 8 points of growth disappeared in translations.

• The sales miss. Consensus expected more than CHF 850.3 million, and did not get it.

• Americas looked weak. +4.5% was reported in the biggest region. In constant currency, it was +13.0%. Still the slowest region, but far from the near-zero the reported number suggests.

• Weak wholesale. The channel grew only 4.8%. The company states it was deliberate: full-price discipline in a heavily promotional market and a clean runway for the new product cycle. The market read it as demand weakness.

• Guidance. Full-year constant currency growth “in the low-20% range” after +24% in H1 implies some deceleration in H2, even if part of it is the wholesale choice above.

Worth noting also what the sellers skipped: On raised its full-year gross margin guidance to at least 65% and kept the adjusted EBITDA margin at 19.5 - 20%. A brand that is losing demand does not raise its margin guidance while absorbing US tariffs at full price. In my view, this combination of a slower reported top line and stronger profitability is a currency and channel story, not a brand story.

The longer history matters more. In 2022, the company lost money and burned cash: LTM EPS bottomed at -$0.65, and FCF was negative for 2 years. Since then, LTM revenue reached $3.99B, net income $490M, and FCF $542M. Every margin is above its 5Y mean. Net income margin increased from a 2% average to 12.3% today.

And the stock? Total return of 3.62% over 5 years (0.71% per year) against 89.05% for the S&P 500. And this despite the fact that the business roughly tripled over the same period. That is what makes the valuation interesting.

Future

• Consensus revenue: FY2026 $4.32B (+13.57%), FY2027 $5.13B (+18.67%), FY2028 $6.18B (+20.57%) - note the reacceleration as FX normalizes

• Consensus EPS: $1.73 -> $2.06 -> $2.57 (FY2026 - FY2028)

• 5Y forward EPS growth estimate: 24.79%, versus a 40.18% mean since IPO - the street already cut its expectations nearly in half

• Analysts: Strong Buy (7 Strong Buy / 16 Buy / 4 Hold / 1 Sell, 28 covering), average target $46.15, +48% from here (the range is $19.98 to $72.91)

• Company guidance: FY2026 constant currency sales growth in the low-20% range, gross margin at least 65.0%, adjusted EBITDA margin 19.5 - 20.0%

Current Valuation

Current vs 5Y mean:
Price/Fwd Earnings: 17.3x vs 53.0x
Price/Fwd Sales: 2.2x vs 5.0x
Price/FCF: 19.1x vs 36.6x
Price/Book: 5.8x vs 9.4x
PEG: 0.70 vs 1.20
Forward Earnings Yield: 5.7%

Every multiple is below its mean since IPO. The forward P/E touched its -1 standard deviation band (16.0x) recently. ONON now trades at a 5.7% forward earnings yield. To me, that is a value-stock yield for a business with ~25% expected EPS growth.

Analyst’s note:

The averages here cover the period since the September 2021 IPO. In the zero-rate era of 2021 - 2022 the stock traded at 150 - 220x forward earnings, which inflates the 53x mean. I do not use that mean anywhere in my model. The exit multiples below start from today’s 17x, not from the IPO-era levels.

One chart frames this whole section. The corridor below takes the consensus forward EPS estimate and multiplies it by my three exit multiples: 17x, 25x, and 30x.

Fair Price

My standard model projects EPS 5 years forward and takes the slowdown risk into account by capping future growth at 20% per year. The cap is a rule - I never model above it. Consensus for the next 2 years is +19% and then +25%, and the 5Y consensus is ~25% per year, so the cap binds, and my path is below the consensus average. There are no dividends to add. Starting EPS is $1.75 (FY2026), which gives $4.35 in FY2031.

Analyst’s note:

Worth noting that consensus moved to $1.73, when I had already built the model. The 1% difference is far inside the 30% margin of safety, so I keep the model as it is.

The discount rate is 12%; the margin of safety is 30%. The exit multiples are 17x, 25x, and 30x. 17x assumes the market keeps pricing ONON exactly like today for 5 more years, despite 20% growth. 25x is a normal multiple for premium growth (a PEG of 1.25). 30x is a multiple for a brand with recovered momentum - still far below the IPO-era levels. The 53x historical mean plays no role, for the reason in my Analyst’s note in the section above.

Bear Case (exit P/E 17x): fair price $42 - MoS price $29
Base Case (exit P/E 25x): fair price $62 - MoS price $43
Bull Case (exit P/E 30x): fair price $74 - MoS price $52

At $31, the stock trades 26% below the Bear Case fair price. Not below the Base Case - below the Bear Case, the scenario where the multiple never recovers at all. Worth noting that the Bull Case is within a dollar of the highest street target of $73, so the corridor is not built on my private optimism.

Analyst’s note:

For transparency about the timing: I bought at ~$38 in early August, before the report. The model said the zone starts at $42, and I paid inside it. Two weeks later, the market set the price 18% lower. This is why the margin of safety exists - it does not protect from a drawdown, it decides whether the drawdown is a problem or an opportunity.

Verdict: ONON belongs in a long-term portfolio as a premium-brand growth position. Not a core holding, and sized like a volatile mid-cap consumer cyclical. The company is the strongest growth story in sportswear: the best gross margin in the industry (65.4%), 20%+ constant currency growth, ~$0.8B of net cash, and rising returns on capital. The stock trades at 17x forward earnings with a PEG of 0.70, priced next to competitors that grow 3x slower. My model takes the risks into account by capping future growth at 20% and still produces a fair price corridor of $42 - 74. At $31, the stock is 26% below even the Bear Case, deep inside the accumulation zone of $29 - 42. I own the stock from ~$38, as documented in the July portfolio update. In my view, the realistic bear case is not a crash but dead money - constant currency growth slides to the mid-teens, and the stock moves sideways with 20% swings on every report. Buy with a multi-year horizon, and judge the thesis on constant currency growth and gross margin, not the share price.

Due Diligence

Profitability (11 of 15):

• Positive Gross Profit: $2.58B

• Positive Operating Income: $549M

• Positive Net Income: $490M

• Positive FCF: $542M

• Gross margin >= 40%: Yes (64.8%)

• Net margin >= 10%: Yes (12.3%)

• FCF margin >= 10%: Yes (13.6%)

• Management (ROIC, ROE, ROA) >= 10%: Mostly (17.1% / 24.0% / 9.6%)

• Strong 3Y Revenue Growth: ~30% per year

• Exceptional Revenue Growth Forecast: ~19% per year over the next 3 years

• ROE is increasing: ~8% -> 24%

• ROIC is increasing: ~11% -> 17%

• Revenue surprises in the last 5Y in a row: No (2023 missed)

• EPS surprises in the last 5Y in a row: No (2021 and 2023 missed)

• EPS growth YoY 5Y in a row: No (Decline in 2025)

Financial Strength (6 of 6):

• Total assets ($4.01B) exceed total liabilities ($1.65B) by 2.4x

• Negative Net Debt: -$0.79B (cash and short-term investments of $1.49B against $696M of debt)

• Low Debt/Equity: 0.30 (5Y mean: 0.24)

• Debt/Capital: 0.23 (5Y mean: 0.19)

• Interest coverage (FFO): 21.8x

• Piotroski F-Score: 8 of 9

Valuation and Advantage (2 of 4):

• Valuation < 5Y mean: Yes

• Valuation < the industry: Mixed

• Does it have a moat: Yes (narrow)

• Outperformed the S&P 500 over 5Y: No (0.71% vs 13.58% CAGR)

Shares (2 of 3):

• Insider ownership >= 5%: Yes (founder-led, ~31%)

• Fewer shares outstanding YoY: No (slight dilution)

• Insider buys in the last 6M: Yes (May)

Price (3 of 4):

• 1Y price forecast > 10%: +48%

• Next 5Y EPS growth estimate (CAGR) > 10%: Yes (~24.8%)

• DCF Value: ~$74.75; undervalued by ~58%

• Short Interest < 5%: No (6.4%)

Watchlist Note

Premium Swiss sportswear compounder. +21.6% cc sales growth, 65.4% GM (best in sportswear), net cash ~$0.8B, FCF margin 13.6%. 17x fwd P/E vs 53x avg since IPO, PEG 0.70 - cheapest growth in the competitors group. Fell 33% YTD on FX translation and a deliberate wholesale decision, not brand weakness. Fair price corridor $42 - 74; accumulation zone $29 - 42. Risks: US consumer, CHF translation, HOKA/Nike competition, 6.4% short interest.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

Why I am publishing this

On this website, I analyze companies that I find appealing. This series is the other side of the story, showing the portfolio where I invest my own money based on those analyses. Each month, you will see the returns, every trade, and the reasoning behind my decisions. I will include the losers too. I currently have two positions down 31% and 38%, and I will discuss both today.

The rules I follow (what I buy, how much I invest, and when I sell) are documented separately in the LTP Portfolio Manifest. This update demonstrates how those rules work.

The method

I track everything in Portseido. Returns are time-weighted (TWR), so deposits and withdrawals do not distort the numbers. What you see is pure investment performance. I compare my results against four benchmarks: S&P 500, QQQ, VUG, and SPYG. My portfolio combines growth stocks with value stocks and bonds. Relying on just one index would misrepresent my performance. I share percentages, not dollar amounts.

The track record

Since it began in December 2021, the portfolio has compounded at a rate of 16.12% per year, outperforming the S&P 500 (11.64%), QQQ (14.96%), VUG (14.45%), and SPYG (14.59%). The aspect that matters most to me: this has been achieved with a beta of 0.95 and a maximum drawdown of -19%, which is no deeper than the index over the same period. It has delivered greater returns than the market while involving no more risk than the market - that is the very objective.

Full honesty. 2026 YTD, I am behind: +5.72% vs +13.04% for the index. I was also behind in 2024. The reason for this in both cases is the same, and it is a deliberate choice, not an accident, since 40% of the portfolio has been held in treasuries while the market has continued to move. The following two sections explain why I accept that.

Below: all 10 positions with weights and entry prices, why 40% of the portfolio is in treasuries, where every dollar of gains and losses came from, and the three buys I just made with the exact reasoning behind each.

Read more

TL;DR: Reddit is the biggest archive of human conversation on the internet and is currently the fastest-growing ad platform on a large scale. The share dropped 21% following the Q2 report and has fallen 30% so far this year, even though revenue increased by 61% (achieving eight quarters in a row of growth above 60%) with a 91% gross margin, a record 43% EBITDA margin, and $2.8 billion in net cash. At a Price/Fwd Earnings ratio of 18.7 and with a consensus earnings per share growth rate of 47%, the PEG ratio is 0.40, indicating that the market is treating hypergrowth as if it will terminate at the end of next year. The bear case: US daily users have stopped growing, and Google has control over both the traffic and half of the licensing aspects. Even if growth is capped at 20% (as against the 47% consensus forecast), my Bear Case fair price is $213, and at around $160 the stock is trading within my accumulation range.

Investment Thesis

Reddit ended the June quarter with revenue up 61%, a net margin of 31%, and a FCF margin of 32%. It is now trading 21% lower than it was a year ago. This difference between the business's performance and its stock price is the focus of this post.

• The growth at this scale is exceptional. The company reported $805 million in quarterly revenue, an increase of 61%, marking the eighth consecutive quarter of over 60% growth, which puts it on a $3.2 billion run-rate. This revenue growth, combined with the EBITDA margin, categorizes Reddit as a "Rule of 104" company.

Analyst’s Note:

What the “Rule of 104” is: the traditional software benchmark (the Rule of 40) combines revenue growth with profit margin. A company in good shape should achieve more than 40. The rule exists since growth and profitability generally tend to offset each other. Reddit’s revenue growth of 61% together with a 43% EBITDA margin makes 104, and it has now been above 100 for five quarters in a row (111 -> 108 -> 115 -> 109 -> 104). To put this in context, both Meta and Microsoft operate in the 60s and 70s respectively and almost no company of this size manages to do both. Reddit is one of them.

• Profitability came quickly. The gross margin is 91.3%, the adjusted EBITDA margin is 43%, and FCF reached $1.02 billion over the past year. Revenue per employee has just surpassed $1 million.

• There is a clear path for monetization. The average revenue per user (ARPU) of $6.18 increased by 36%. The international market presents an opportunity, as 59% of users contribute only 21% of revenue, earning $2.26 per user compared to $11.85 in the US.

• The balance sheet is strong. $2.79 billion in cash against $21 million in debt, and buybacks have already begun, two years after the IPO.

• The stock multiple has dropped. It stands at 18.7x of Price/Fwd Earnings compared to a 47.9x average since the IPO, which is below the -1 standard deviation band. Additionally, the PEG ratio is 0.40, while consensus anticipates a 47% growth in EPS.

• The bear case highlights some concerns. Daily users in the US saw a sequential decline, search referrals are unpredictable as Google implements AI responses, and nearly all licensing value comes from just two partners. The expected slowdown is set at around 48% for the next quarter after experiencing 61% growth.

Company Overview

Next Earnings Date: Oct 30, 2026 (estimated)
Market Cap: ~$30.8B
Sector: Communication Services | Industry: Interactive Media and Services
Type: High Growth
Beta (5Y Monthly): 2.03 | Short Interest: 8.80%
Dividend Yield: none
Employees: ~2,555

Founded in 2005 and going public in March 2024, Reddit is a community of communities with over 100,000 active groups, 26 billion posts and comments, 130 million daily visitors, and 515 million weekly visitors. Users visit to ask questions, discuss topics, compare opinions, and make decisions. This makes Reddit a vast collection of genuine human opinions on the open internet. It is important to note that it is still led by its co-founder, Steve Huffman.

Analyst’s note:

I also use this social media platform for 7 years.

The advertising division generates about 95% of revenue: Reddit Ads include full-funnel formats, AI-powered Reddit Max campaigns, new Shopping Listing Ads, and a Shopify integration for merchants. The remainder comes from data licensing for AI training, input grounding, and search indexing (partners include Google and OpenAI), along with Reddit Premium. Besides the main app, Reddit also offers Reddit Answers, agentic search, machine translation for non-English markets, and a Developer Platform with games that encourage daily engagement.

Market Overview

A key question for Reddit is how a platform based on human conversation will adapt when AI saturates the internet with synthetic content. Reddit believes that the value of verified human perspectives will increase.

Daily active unique users reached 130.3 million, an 18% rise, and weekly active users surpassed half a billion (514.6 million), marking a 24% increase. International daily users grew by 28%, and logged-out users increased by 27%.

Advertising revenue rose sharply by 64%, fueled by higher pricing and more impressions. The number of active advertisers grew by over 70%, with 11 out of 15 industry sectors experiencing growth of more than 50%.

Reddit is where consumers rely on confirmation before making purchases. About 50% of US shoppers check recommendations on Reddit prior to buying. Third-party research indicates that shoppers from Reddit tend to generate greater lifetime value across various categories.

AI benefits Reddit in two ways: the labs license the archive for training, while users who are tired of AI-generated content seek out human responses.

AI simplifies web content into summaries, whereas Reddit offers the opposite: context, opinions, and first-hand experiences. A potential risk comes from Google, which connects Reddit to a significant share of its traffic. As search transitions from links to AI-driven answers, that traffic channel could shrink.

Economic Moat

Reddit has 26 billion posts and comments built over two decades of niche experiences. It has the largest legally licensable archive of genuine conversations on the internet. AI has increased its value, not decreased it, prompting Reddit to take legal action against those who scrape its content without payment.

Over 100,000 active communities are moderated by the users themselves. This structure cannot be replicated by any algorithmic feed. People search specifically for "reddit". Advertisers reach audiences through context instead of relying on personal data, placing Reddit in a strong position as privacy regulations tighten.

The gross margin stands at 91.4% compared to a 5Y average of 87.5%. The EBIT margin is 28.3%, against a negative historical average. The FCF margin is 36.7%, surpassing the 1.5% mean. The ROIC became positive only in 2025 and has already hit 23.9%.

This advantage is real but narrower than what you'd find in a mega-cap. Attention is contested every day, part of the traffic belongs to Google, and revenue from the archive is contingent on two licensing agreements. The moat lies in the content, while the vulnerability is in the distribution.

Business Strategy

First, turn weekly visitors into daily ones. Only one in four users who visit weekly returns daily. Management is working on improving the feed to encourage web users to shift to the app, where engagement is much higher. They are also adding features like videos in comments and games. User retention for new app users has improved by 50% YoY. The ambition is to reach one billion daily users worldwide and 100 million in the US.

Second, narrow the performance gap with leading advertising platforms. Machine learning has increased click-through rates by over 40% YoY. Revenue from Reddit Max campaigns increased by more than 150% QoQ. Shopping Listing Ads have launched, and the Shopify integration is now widely available. The mid-market and small business channel has doubled. This mirrors Meta's strategy, but Reddit is doing it earlier.

Third, monetize AI. The archive is licensed for training, post-training, grounding, and search indexing, creating multiple revenue streams instead of relying on just one contract. Reddit Answers converts the same archive into a search product. Both search-related activities and the volume of searches have grown during the quarter.

The cost structure highlights where the funds are allocated. R&D expenses totaled $814 million over the past year, making it the largest line item - about 29% of revenue is invested in product development. SBC decreased from 19% to 13% of revenue. Revenue grew 1.6 times faster than adjusted costs.

Capital Allocation

In the last twelve months, FCF reached $1.02 billion with a 36.7% margin, which has improved from a negative margin two years ago. This change is due to Reddit's model: the company rents its infrastructure from third-party clouds and doesn’t own data centers. Therefore, CapEx was only $1 million for the quarter, making up 0.1% of revenue. This contrasts sharply with AI infrastructure companies. Hyperscalers typically spend a third of their revenue on capital expenses, while Reddit converts a third of its revenue into cash.

Reddit has $2.79 billion in cash and investments against $20.9 million in total debt, with net cash amounting to about 9% of the market cap. Its total liabilities of $351 million are covered 10x by assets. The accumulated deficit fell from $671 million to $214 million in just six months. The company is earning quicker than it is spending.

Capital returns began this quarter. The company repurchased 1.5 million shares for $235 million at an average price of $157.57 (near the current price). Fully diluted shares increased by only 0.2% YoY, or 0.9% when excluding buybacks, meaning the impact of SBC dilution has practically been balanced out. There is no dividend at this stage of growth.

Advantages

Reddit is experiencing hypergrowth, with revenue growing 61% on a $3.2 billion run-rate. This marks the eighth consecutive quarter surpassing 60% growth, aligning with a "Rule of 104" profile, which is 6,400 basis points above the software benchmark.

The asset that AI can't replicate is the enormous volume of 26 billion human posts across over 100,000 communities. As the internet fills with synthetic content, the verified human perspective becomes a unique asset. Both AI labs and human users turn to Reddit for this value.

Reddit maintains software-like economics without a large CapEx burden. It boasts a gross margin of 91%, an EBITDA margin of 43%, a FCF margin of 37%, with CapEx at just 0.1% of revenue and $1.08 million of revenue per employee.

The company has built a financial stronghold early on with $2.8 billion in net cash, insignificant debt, and share buybacks initiated two years after the IPO (essentially offsetting dilution).

The stock multiple is at 18.7x Price/Fwd Earnings compared to a 47.9x average since the IPO, which is below the -1 standard deviation mark. The PEG ratio stands at 0.40, contrasting with a 0.86 average. This multiple is 2.2 times Meta’s expected growth.

Disadvantages

The core US user base has stalled. Daily unique visitors in the US reached 53.2 million but decreased sequentially and increased by only 6% YoY. Almost all new users are international and monetize at $2.26 each compared to $11.85 in the US. A decline in the premium user base may lead to reduced advertising revenue.

Google is both a partner and an obstacle. Search referrals have been unstable and weakened toward the end of the quarter. AI-generated summaries are replacing the links that Reddit used to rank for. Management has identified their visibility as low. A recent shareholder letter (PDF) revealed that management disclosed the impact of Google too late. The company is now building direct habits to counteract the shrinking traffic source.

Licensing revenues present a concern. Much of the content licensing value comes from just two partners, Google and OpenAI. These agreements are medium-term, carrying renewal risks. Other revenue streams are limited, contributing only $43 million a quarter and growing at 24%, which is slower than advertising growth.

The expected deceleration is concerning. Current growth is at 61%, but the implied growth rate for Q3 is around 48%, and consensus projects a revenue growth of 31%.

The stock is volatile. It has a beta of 2.03, 1Y volatility near 70, short interest of 8.8%, and a single-day drop of 21%.

SBC remains at 13% of revenue.

What the bears highlight: there is a real gap between revenue and users. Revenue increased by 61%, but the users who support the high ARPU (US, logged-in) have hardly grown. If US daily users continue to decline while the stock waits for proof, it will remain cheap for a reason. I take that scenario seriously.

Competitors

Reddit competes for attention and advertising budgets with platforms like Meta, Pinterest, Snap, TikTok, YouTube, and X. It also faces competition from AI assistants. ChatGPT and similar tools increasingly handle questions that people used to ask on Reddit.

The charts below compare RDDT with META, PINS, and SNAP. On forward earnings, Reddit at 18.7x costs about the same as Meta at 18.4x, but Reddit's consensus EPS growth is 2.2 times higher (46.7% vs 21.0%). Its net margin of 31.4% beats Meta’s 29.8%, and its ROIC of 23.9% is the highest in the group. Pinterest has a net margin of 6.0%, while Snap is negative. PEG ratios show Reddit at 0.40, Meta at 0.88, and Pinterest at 0.60. Only Snap is cheaper at 0.30, but it has a negative net margin and offers negative returns on capital.

Considering quality per dollar, Reddit is the most affordable option on this list.

Past

• Q2 2026 results: revenue of $805M, up 61% (US +56%, international +84%). Gross margin is 91.3%. Net income stands at $253M (31.4% margin) with diluted EPS at $1.25, a rise of 178%. Adjusted EBITDA is $343M with a 42.6% margin. FCF is $261M.

• H1 2026: revenue reaches $1.47B, up 65%, along with $572M of FCF (39% margin).

• Two years ago, this company lost money on every line. LTM diluted EPS rose from -$9.66 at the post-IPO low to +$4.29. This is one of the fastest shifts to profitability in the consumer internet sector.

The stock went public in March 2024 at $34. Total return since then is +224% (vs +75% for the S&P 500 (VOO)), reflecting a 47.9% CAGR over three years. However, the journey has been tough, featuring two drawdowns of over 40% within the two years since going public, including a collapse of over 50% in spring 2025 and the current decline of -30.4% year-to-date, with -17.9% in the past month alone. Reddit can gain value quickly and also fall sharply. That's the reality.

Future

• Guidance for Q3: revenue between $860M and $870M. This represents a +48% YoY increase at the midpoint, marking a record quarter in absolute terms. Adjusted EBITDA is expected between $385M and $395M, also a record with about a 45% margin.

• Consensus revenue projections: FY2026 $3.38B (+53%), FY2027 $4.42B (+31%), FY2028 $5.50B (+24%).

• Consensus EPS: $7.56 -> $9.63 -> $11.79 (FY2026-FY2028). Growth rates are projected at +67%, +27%, and +22%. The 5Y forward EPS CAGR is 46.7%.

• Analysts have a Buy rating (5 Strong Buy, 17 Buy, 11 Hold, and 1 Sell, totaling 34 analysts), with an average target of $216.35, indicating a +35% upside. The lowest target is $120, and the highest is $300.

Valuation

Price/Fwd Earnings: 18.7x vs 47.9x average (below the -1 SD band at 25.0x)
Price/Fwd Sales: 8.0x vs 12.2x (at the -1 SD band)
Price/FCF: 30.2x vs 134.5x
Price/Book: 9.3x vs 11.2x
PEG: 0.40 vs 0.86 (below the -1 SD band)
Fwd Earnings Yield: 4.7%

The stock is down 21% over the past year, but earnings estimates kept rising. As a result, the Price/Fwd Earnings dropped from the mid-40s to 18.7x, making Reddit the cheapest it has ever been. If you exclude the $2.8 billion in net cash, the enterprise trades at 16.0x forward EBITDA and 7.2x forward sales. Every multiple is at or below its -1 standard deviation band, except for Price/Book.

Analyst's note:

The averages here only cover the 2.5Y since the March 2024 IPO. In the early period, the company had triple-digit multiples on small earnings, so the 47.9x mean makes the comparison look better. That is why I focus on the PEG of 0.40 and the peer table, which considers Meta’s multiple for double Meta’s growth, instead of the historical mean.

The entire section is illustrated by one chart. Taking the forward EPS estimate that represents the consensus and multiplying it by the three exit multiples from my fair price model - specifically 20x, which is about the current multiple and the level at the post-IPO trough; 25x, which is the -1 standard deviation band; and 30x, which is a small premium reflecting the group's leading growth - the corridor currently extends from about $171 to about $256. The present price of about $160 is therefore below the whole corridor, including the bear line. All the scenarios in my model, even the one in which the multiple never recovers from its all-time low, begin at a level above today's price.

Fair Price

I have set a cap on EPS growth at 20% (my maximum value). The consensus estimate for 5Y EPS growth is 46.7%. Even after applying this reduction, the consensus EPS of $7.56 for FY2026 rises to about $18.81 by 2031.

The exit multiples are 20x, 25x and 30x. 20x is about the current multiple (18.8x) and also represents the level at the end of the period following the IPO, so in the Bear Case it is assumed that the market will never re-rate a company with 20% growth from its all-time low. 25x is the minus one standard deviation figure and 30x is a moderate premium over Meta's 18.4x for twice the expected growth (although this is well below Reddit's own post-IPO average of 47.9x), which I don’t regard as a reliable benchmark. Future prices are discounted at 12% per year, with an additional 30% margin of safety.

Bear Case (exit P/E 20x): fair price $213 - MoS price $149
Base Case (exit P/E 25x): fair price $267 - MoS price $187
Bull Case (exit P/E 30x): fair price $320 - MoS price $224

The share price, at about $160, is within the accumulation range of $149 to $213. The fair value under the Bear Case, with growth reduced to 20% and a trough multiple that is permanent, is 33% higher than the current price. Compared to the fair value based on the Base Case of $267, the potential upside is about 67%.

The verdict is that Reddit is not a core compounder such as Microsoft but rather a growth position which has a 2.03 beta, has suffered two drawdowns of 40% or more in two of its public years, and has a real bear case if Google takes control of it. That is the very reason the price is within the range in question. I am a buyer within the $149 to $213 range, taking into account the level of volatility, and I would increase my position aggressively towards the Base Case MoS price of $187 should there be any further weakness. The market rewards you for holding on to this share through the noise. You should buy with a multi-year time horizon and evaluate the argument based on US daily unique users and revenue growth remaining above 40%, not on the share price.

Checklist

• Gross margin >= 40%: 91.4%

• Net margin >= 10%: 31.4%

• FCF margin >= 10%: 36.7%

• Management (ROIC, ROE, ROA) >= 10%: Yes (23.9%, 30.7%, 15.7%)

• Piotroski F-Score: 7 of 9

• Revenue surprises in last 5Y in a row: Yes (for 3Y; based on TradingView)

• EPS surprises in last 5Y in a row: Yes (for 3Y; based on TradingView)

• EPS growth YoY 5Y in a row: No (for 3Y; based on TradingView)

Valuation and Advantage:

• Valuation < its 5Y averages: Yes

• Valuation < the industry: Mostly

• Does it have a moat: Yes (narrow)

• Outperformed the S&P 500 10Y CAGR: Yes (public only since March 2024: +224% total return vs +75% for VOO)

Shares:

• Insider ownership >= 5%: Yes (27.24%; based on FinViz’s data)

• Fewer shares outstanding YoY: No (fully diluted count +0.2%, but Q2 buybacks offset SBC dilution for the first time)

• Insider buys last six months: Yes (Mar; based on FinViz’s data)

Price:

• 1Y stock price forecast >= 10%: Yes (+35.3%)

• Next 5Y EPS growth estimates (CAGR) >= 10%: Yes (46.7%)

• DCF Value: ~$158; fairly valued (5Y, LTM FCF of $1.02B grown at my capped 20%, discount rate: 10%, terminal growth: 3%, equity model: FCFF plus net cash)

• Short Interest < 5%: No (8.8%)

Due Diligence

Profitability (11 of 12):

• Positive Gross Profit: $2.54B

• Positive Operating Income: $785M

• Positive Net Income: $871M

• Positive FCF: $1.02B

• Exceptional 1Y Revenue Growth: 65% (H1 2026 - 61% in the June quarter)

• Exceptional 3Y Revenue Growth: ~58% (per year for the last 3 years: ~$0.7B -> $2.78B)

• Exceptional Revenue Growth Forecast: ~36% (per year over the next 3 years, consensus: +53%, +31%, +24%)

• Exceptional ROE: 30.7%

• Exceptional 5Y Average ROE: No (the average is negative)

• ROE is increasing: Yes (loss-making in 2024 -> 30.7%)

• Exceptional ROIC: 23.9%

• ROIC is increasing: Yes (turned positive in 2025 and has climbed every quarter since)

Solvency (6 of 6):

• Total assets ($3.64B) exceed total liabilities ($351M) by 10.4x

• Negative Net Debt: -$2.77B (cash and investments of $2.79B against $20.9M of total debt)

• Low Debt/Equity Ratio: 0.006

• Debt/Capital: 0.63% (5Y mean: 1.14% - there has never been real leverage here)

• Interest coverage: not needed - Reddit has net interest income (~$49M in H1 2026); the balance sheet pays the company

• Altman Z-Score: ~55 (anything above 3 is the safe zone)

Watchlist Note

The internet's archive of human conversation + fastest-growing ad platform at scale (+61%, 8th quarter above 60%). 18.7x fwd P/E vs 47.9x post-IPO mean (below -1 SD) for 46.7% consensus EPS growth; PEG 0.40. 91% GM, 43% EBITDA margin, 37% FCF margin, $2.8B net cash, capex 0.1% of revenue. Bears: US DAU flat, Google referrals volatile, licensing = two partners. Fair price: bear $213 / base $267 / bull $320 (20% capped growth vs 47% consensus). Accumulation zone $149-213; price ~$160 inside it. Watch: US DAUq, growth >=40%, licensing renewals. Earnings: Oct 30.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

TL;DR: Microsoft is the enterprise software monopoly plus the fastest-growing hyperscaler. The stock is down 11% in a year, yet Azure just crossed $100B in annual revenue growing 43%. Contracted future revenue (RPO) jumped 84% to $678B. At 23.9x forward earnings versus a 29.8x five-year average (below the -1 standard deviation band), the market is pricing one of the highest-quality businesses at roughly a market multiple. The bear case: capex at 35% of revenue has crushed FCF. Verdict: my bear case fair price is $453, and the stock trades right at it. My full fair price math and the accumulation zone are at the end of this post. I have been holding this giant for years and will continue to do so.

Investment Thesis

Microsoft closed fiscal 2026 with revenue up 18%, operating income up 21%, and Azure accelerating to 43% growth. The stock still trades with an 11% decline over the past year. That gap between the business and the price is the reason for this post.

• Azure crossed $100 billion in annual revenue and is accelerating. Full-year growth of 41% versus 35% in FY2025, a 43% exit rate in the June quarter, and guidance of ~45% for the next quarter. Management is doubling total data center capacity in two years. And demand still exceeds supply.

• Revenue is unprecedented. Commercial remaining performance obligation of $678B, up 84% (up 25% even excluding OpenAI), with a weighted duration of just 2.3 years.

• High quality. 68% gross margin, 40% net margin, 23% ROIC, net cash balance sheet, and the 21st consecutive year of dividend increases.

• The multiple has already de-rated. 23.9x forward earnings versus a 29.8x five-year mean (below the -1 standard deviation band), while forward EPS estimates rose 23% over the same year.

• The bear case is the capex supercycle. $116B of capital expenditures (35% of revenue) cut FCF margin from a 29% average to ~20%. My model absorbs that by capping growth at consensus 15% and using a 20x bear exit multiple. And even then, the stock trades right at my bear case fair price.

Company Overview

Next Earnings Date: Oct 29, 2026 (expected)
Market Cap: ~$3.40T
Sector: Information Technology | Industry: Software
Type: Mega Growth
Beta (5Y Monthly): 1.13 | Short Interest: 1.20%
Dividend Yield: 0.80%
Employees: ~223,000

Microsoft is the largest software company in the world and the second-largest cloud provider. Its base is a set of near-monopolies in enterprise computing: Windows, Microsoft 365 (Office, Teams, Outlook), and Entra identity run the day-to-day work of most large organizations on the planet. And every one of those parts is now a distribution channel for AI.

The product stack today: Azure cloud infrastructure and AI services (including Azure AI Foundry with 11,000+ models from OpenAI, Anthropic, Mistral, xAI, and Microsoft’s own MAI family); the Copilot family (Microsoft 365 Copilot, GitHub Copilot, Security Copilot); Microsoft 365 and Office; Dynamics 365 business applications; LinkedIn; GitHub and Visual Studio developer tools; Microsoft Fabric and SQL Server for data and analytics; the Power Platform; the security suite (Defender, Sentinel, Entra, Purview); Windows and Surface devices; Bing and Edge search advertising; Xbox with Activision Blizzard content (Call of Duty, Candy Crush); and first-party silicon (Maia AI accelerators) underneath it all.

Analyst's note:

The chart above shows LTM FCF of $72.9B and a 22.9% margin because Koyfin's data does not yet include the June 2026 quarter, where capex stepped up to $41B. On the fresh press-release numbers, FY2026 FCF is $67.0B, and the margin is ~20%.

Market Overview

The core question for Microsoft is the same one hanging over all of AI infrastructure: does the spending continue, and does it pay back? Microsoft’s own numbers are currently the best answer the bulls have:

• Commercial remaining performance obligation reached $678B, up 84%, and up 25% even excluding OpenAI’s commitments. This is signed, contracted future revenue, and ~30% of it is converting within 12 months.

• Microsoft added another gigawatt of data center capacity in the June quarter alone and remains on track to roughly double total capacity in two years, while still being demand-constrained.

• Monetization is visible, not hypothetical. Microsoft 365 Copilot passed 30 million paid seats, GitHub Copilot serves millions of developers, Fabric has 40,000+ paid customers (up 60%), and Copilot workload throughput is up 4x since the start of the year.

• Management guides calendar 2026 capex to approximately $175 billion, and Azure growth in H1 FY27 to accelerate versus H2 FY26.

Unlike a chip vendor, Microsoft does not need the AI capex boom to continue at any particular pace - it needs enterprises to keep paying for software. The AI buildout runs through the same subscription machine that survived every previous cycle: if AI works, Microsoft sells it per seat per month; if AI disappoints, the seats are still there.

Economic Moat

Microsoft has one of the widest moats in business, built on three layers.

The first is distribution. Hundreds of millions of commercial seats across Microsoft 365, Windows, and Entra. It means every new product (including every Copilot) launches into an installed base competitors spend decades trying to build.

The second is switching costs. Identity, documents, collaboration, security, and now AI workflows are woven into a single stack.

The third is the full-stack AI position. Microsoft owns the infrastructure (data centers, Maia silicon), the model layer (the OpenAI partnership, Anthropic and other frontier models in Foundry, plus its own MAI family), the platform (Azure AI Foundry), and the applications (Copilot). It collects revenue at whichever layer.

But the margins show that the moat is under pressure. Even with the largest capex cycle in software history, EBIT margin reached 46.8% - above its own five-year average of 43.9%. Because operating leverage in software keeps offsetting infrastructure costs. Net margin reached 40.3% versus a 36.4% mean.

More enterprise seats -> more data and workflows on the platform -> better Copilot and agents -> more Azure consumption -> which funds more capacity and models. Each turn deepens the dependency of the world's back office on one vendor.

Business Strategy

First, the infrastructure grab. Doubling data center capacity in two years, ~$175B of calendar 2026 capex, and first-party silicon to bend the cost curve. Maia 200 delivers 30% better performance per dollar than the latest merchant hardware in Microsoft’s fleet and already serves both OpenAI and MAI models.

Second, Copilot monetization on the installed base. Microsoft 365 Copilot passed 30 million paid seats. It is priced as a premium on top of existing subscriptions. The same motion repeats across GitHub, security, and Dynamics - AI is sold as an upsell to customers who already cannot leave.

Third, model independence. The OpenAI partnership remains central (the investment produced $4.96B of net gains in FY2026). But Microsoft now hedges it from every direction: Anthropic models in Foundry and Copilot (a $3.2B investment gain in Q4 alone), the in-house MAI family in production, and 11,000+ models in the catalog. Whoever wins the model race, the enterprise buys it through Azure.

Capital Allocation

Operating cash flow grew from $136B in FY2025 to $183B in FY2026, up 34%. The machine works.

What changed is where the cash goes. Capital expenditures reached $116B, or 34.9% of revenue versus an 18.3% five-year average, taking FY2026 FCF down to $67.0B ($182.9B of operating cash flow minus $115.9B of capex) and the FCF margin to ~20% versus a 28.9% five-year mean. In the June quarter alone, capex was $41.0B, and FCF fell 23% year-over-year.

Management's defense. Roughly two-thirds of capex is short-lived assets (CPUs and GPUs) bought against demand that is already contracted in the $678B RPO, and the June quarter came in above guidance. The balance sheet can afford the bet: $76.8B of cash and investments against $56.8B of total debt (including leases). A net cash position, with debt-to-equity of just 12.8% versus a 38.9% five-year mean.

Shareholder returns continue. $48.7B returned in FY2026 ($26.4B in dividends, $22.3B in buybacks), the 21st consecutive annual dividend increase (quarterly $0.91, payout ratio under 20%). And a share count that still declines slightly every year. Stock-based compensation is a disciplined 3.9% of revenue. The trade-off is explicit: shareholder yield of 1.47% versus a 1.83% five-year mean, because the marginal dollar goes into data centers.

Advantages

• The distribution monopoly. Hundreds of millions of commercial seats across Microsoft 365, Windows, and Entra. Every AI product Microsoft ships lands on an installed base no competitor can replicate, at near-zero customer acquisition cost.

• The fastest-growing hyperscaler. Azure crossed $100B in annual revenue while full-year growth accelerated to 41% from 35% in FY2025. And the exit rate is even higher: 43% in the June quarter, with guidance of ~45% for the next one.

• Unmatched revenue. $678B of contracted future revenue (up 84%, up 25% excluding OpenAI) with 2.3 years average duration. No other company at this scale shows its next two years of demand in a single line.

• A financial fortress under the capex. 68% gross margin, 47% EBIT margin (above its 5-year mean), 40% net margin, 23% ROIC, net cash, and 21 straight years of dividend growth. The quality metrics survived the largest investment cycle in software history.

• A de-rated multiple. 23.9x forward earnings versus a 29.8x five-year mean (below -1 standard deviation), P/Fwd Sales of 8.8x below its own -1 SD band, PEG of 1.54 versus a 2.23 mean. With a +21% average analyst target.

Disadvantages

• The capex supercycle is eating FCF. Capex at 34.9% of revenue versus an 18.3% average cuts the FCF margin from ~29% to ~20%. P/FCF at 50.6x is above its +1 standard deviation band. If AI revenue disappoints, this spending becomes the largest misallocation in tech history. Depreciation from it will weigh on margins for years.

• Returns on capital are drifting down. ROIC fell from ~30% to 22.7% and ROE from ~49% to 34% since 2022, as the denominator fills with data centers faster than AI revenue arrives. The bet is that this reverses. It has not reversed yet.

• Earnings quality needs adjusting. FY2026 GAAP EPS of $17.95 includes $4.96B of OpenAI investment gains and a $3.2B Anthropic gain in Q4. The cleaner non-GAAP figure is $17.28. The concentration runs deeper than the income statement: the $678B RPO grew 25% excluding OpenAI. Simple arithmetic puts roughly $220B, about a third of the entire backlog, on one counterparty that does not yet fund itself from profits. OpenAI is simultaneously Microsoft’s largest AI customer, partner, and mark-to-market position. If its financing falters, that backlog converts slower or not at all.

• Cloud gross margin is compressing. Microsoft Cloud gross margin fell from 68% to 65% in a year on the Azure mix shift and AI infrastructure scaling. Modest so far, but the direction matters when consensus assumes margin stability.

• The legacy tail is shrinking. More Personal Computing declined 4% in Q4 with Windows OEM guided to decline high-teens in FY27 and XBOX taking impairment charges. Roughly 16% of revenue no longer grows, and the market caps the multiple accordingly.

What the bears get right: FCF is falling while the stock de-rates, and nobody (including Microsoft) knows the return profile on $175B of annual capex. If Azure decelerates before the spending peaks, EPS estimates (+11% -> +18% -> +20% for FY27-FY29) get cut, and today's 24x stops looking cheap. I do not dismiss that scenario.

Competitors

In cloud, AWS remains the largest provider. But Azure is closing the gap while growing faster. Google Cloud is third with its own vertically-integrated silicon and models. Oracle has emerged as the aggressive AI-infrastructure challenger, signing frontier-lab capacity deals.

In productivity software, Google Workspace remains the only real alternative. Enterprise share has barely moved in a decade.

The newer front is the model labs themselves. OpenAI and Anthropic increasingly sell directly to enterprises, which makes them partners, customers, and long-term competitors of Microsoft at the same time.

The charts below compare them side by side. On the multiple, Microsoft’s 23.8x forward earnings is cheaper than Amazon (28.6x) and Alphabet (25.1x). Only Oracle looks cheaper at 15.8x with the fastest expected growth (28.5% a year) and a PEG of 0.55. But the quality columns explain the discount: Microsoft’s ROIC of 22.7% is more than double Alphabet’s 13.9% and roughly 2.3x Oracle’s and Amazon’s 9.7%. And its debt-to-equity of 12.8% compares with 53% at Amazon and 389% at Oracle. The “cheap” AI cloud is running its buildout on a leveraged balance sheet, while Microsoft funds a larger one from cash flow.

The premium P/Fwd Sales (8.8x vs 3.0-7.4x) simply reflects the highest margins in the group.

Past

• FY2026 results: revenue $331.8B, up 18%; operating income $155.2B, up 21%; GAAP diluted EPS $17.95, up 32% (non-GAAP, excluding OpenAI gains: $17.28, up 22%); operating margin 46.8%.

• Q4 FY2026: revenue $90.0B (+18%), Microsoft Cloud $59.3B (+27%), Azure +43%, EPS $4.81 - and the $0.27 of one-time benefits still leaves a clean beat versus guidance.

• Total return CAGR: 24.7% over ten years, versus 14.9% for the S&P 500; over five years, 10.8% versus 12.6%. The stock has gone essentially sideways since 2024 while earnings compounded, which is exactly how multiples compress.

The year-by-year picture makes the setup visible: +57.0% in 2023, then +13.7%, +15.5%, and -4.6% year-to-date. Three years of a flat-to-down price against double-digit earnings growth. The worst drawdown year in recent history is 2022 at -28.4% - that is what "risk" has historically meant for this stock, versus -66% for NVIDIA in the same year.

Future

• Consensus revenue: FY2027 $384.9B (+16.0%), FY2028 $456.2B (+18.5%), FY2029 $538.8B (+18.1%).

• Consensus EPS: $19.18 -> $22.59 -> $27.04 (FY2027-FY2029). Growth accelerating from +11% to +19.7% as the capex depreciation wave gets absorbed. Five-year forward EPS CAGR: 15.4%.

• Guidance for the next quarter: revenue of $89.85-90.95B (midpoint above the $89.7B consensus), Azure growth of ~45% in constant currency, operating margin roughly flat.

• FY2027 guidance: double-digit revenue and operating income growth, operating margin down less than a point, capex growing year-over-year.

• Analysts: Strong Buy (13 Strong Buy/40 Buy/3 Hold/0 Sell, 56 covering), average target $555.77, +21.4% upside. The lowest target on the street is $400.

Current Valuation

Price/Fwd Earnings: 23.9x vs 29.8x 5Y average
Price/Fwd Sales: 8.8x vs 10.4x
Price/FCF: 50.6x vs 40.7x (this is the capex distortion, not a rich price)
Price/Book: 7.7x vs 11.6x
PEG: 1.54 vs 2.23
Fwd Earnings Yield: 4.20% vs 3.05%

The stock is down 11% in a year while the forward EPS estimate rose 23%, so the multiple compressed from the mid-30s to 23.9x. On earnings, sales, book, and growth, Microsoft trades below its own -1 standard deviation band. The single expensive-looking metric, P/FCF, measures the capex cycle rather than the valuation.

The market currently pays roughly an S&P-level multiple for a business with 40% net margins and $678B of contracted demand.

One chart frames this whole section. The corridor below takes the consensus forward EPS estimate and multiplies it by the three exit multiples from my fair price model: 20x - the July 2026 trough multiple (a market that never re-rates); 25x - roughly the -1 standard deviation band of recent years; and 30x - roughly Microsoft's own five-year average. Today that corridor runs from ~$386 to ~$579, and the price, at ~$457, is in the lower half of it, below the -1 SD line.

Fair Price

Consensus five-year EPS growth is 15.4% - I use 15%. I am not paying for a hypergrowth story - I am paying for visibility. With 0.8% from dividends, total expected growth is 15.8% a year, turning FY2027 consensus EPS of $19.18 into ~$39.94 by 2031.

The exit multiples are 20x/25x/30x: 20x is the trough the stock actually touched in July 2026 (the market never re-rates from the bottom), 25x is roughly the -1 standard deviation band of recent years (25.6x), and 30x is roughly Microsoft’s own five-year average (29.8x). Future prices are discounted at 12% a year, with a 30% margin of safety on top.

Bear case (exit P/E 20x): fair price $453 - MoS price $317
Base case (exit P/E 25x): fair price $567 - MoS price $397
Bull case (exit P/E 30x): fair price $680 - MoS price $476

At ~$457, the stock trades right at the bear case fair price of $453 - the scenario where the multiple never recovers from its five-year trough is already fully in the price. Against the base case, the stock has ~24% upside to fair value. The accumulation zone is $317-453, and today's price is a bit above its upper edge. Consider buying the dip below $453.

Verdict: Microsoft belongs in a long-term portfolio as a core quality compounder. The lowest-risk way to own the AI buildout: the monetization runs through subscription software rather than cyclical hardware, and with a beta of 1.13, this is the calmest AI position in mega-cap tech. NVIDIA, for comparison, is at a beta of 2.21. I have been accumulating for years, and I would size up meaningfully toward the base case MoS price of $397, which roughly coincides with the lowest analyst target on the street ($400). The realistic bear case here is not a crash but dead money: two more years of multiple compression while capex digests. Buy with a multi-year horizon, and judge the thesis on Azure growth and Microsoft Cloud gross margin, not the share price.

Checklist

Profitability:

• Gross margin at least 40%: 67.9%

• Net margin at least 10%: 40.3%

• FCF margin at least 10%: 20.2%

• Management (ROIC, ROE, ROA): Yes (all above 10%: 22.7%/34.0%/14.1%)

• Piotroski F-Score: 6 of 9

• Revenue surprises in last 5Y: No (2022; Based on TradingView’s data)

• EPS surprises in last 5Y: No (2022 and 2024; Based on TradingView’s data)

• EPS growth YoY 5Y in a row: Yes

Valuation and Advantage:

• Valuation below its 5Y averages: Yes (P/E, P/S, P/B, and PEG all below their -1 standard deviation bands; the one exception is P/FCF, distorted by the capex cycle)

• Valuation below the industry: Mostly (P/Fwd E 23.8x vs AMZN 28.6x and GOOGL 25.1x; only ORCL is cheaper at 15.8x on 389% debt-to-equity)

• Does it have a moat: Yes (wide)

• Outperformed the S&P 500 10Y CAGR: Yes (24.7% vs 14.9%)

Shares:

• Insider ownership at least 5%: No (under 1%, but normal for a 50-year-old mega cap)

• Fewer shares outstanding YoY: Yes

• Insider buys last six months: Yes (Feb; Based on FinViz’s data)

Price:

• 1Y stock price forecast is above 10%: +21.4%

• Next 5Y EPS growth estimates (CAGR) is above 10%: Yes (15.4%)

• Short Interest below 5%: Yes (1.20%)

Due Diligence

Profitability (10 of 12):

• Positive Gross Profit: $225.47B

• Positive Operating Income: $155.24B

• Positive Net Income: $133.75B

• Positive FCF: $67.0B

• Exceptional 1Y Revenue Growth: 18% (FY2026)

• Exceptional 3Y Revenue Growth: ~16% (per year for the last 3 years: $211.9B -> $331.8B)

• Exceptional Revenue Growth Forecast: ~17.5% (per year over the next 3 years, consensus)

• Exceptional ROE: 34.0%

• Exceptional 5Y Average ROE: 39.0%

• ROE is increasing: No (~49% in 2022 -> 34% - equity is growing faster than income during the buildout)

• Exceptional ROIC: 22.7%

• ROIC is increasing: No (~30% in 2022 -> 22.7% - the data center denominator effect)

Solvency (6 of 6):

• Total assets ($758.4B) exceed total liabilities ($316.0B) by 2.4x

• Negative Net Debt: -$20.0B (cash and short-term investments of $76.8B against $56.8B of total debt, leases included)

• Low Debt-to-Equity Ratio: 0.13

• Debt-to-Capital: 11.4% (5Y mean: 27.7%; the balance sheet keeps getting cleaner even through the capex cycle)

• Interest coverage: ~51x ($155.2B of operating income against $3.1B of interest expense and interest income of $3.3B covers the bill on its own)

• Altman Z-Score: ~8.2 (anything above 3 is the safe zone)

Watchlist Note

Enterprise software monopoly + fastest-growing hyperscaler (Azure $100B+, +43%). 23.9x fwd P/E vs 29.8x 5Y avg (below -1 SD) for 15.4% consensus EPS growth; PEG 1.54. 68% GM, 40% NI margin, 23% ROIC, net cash; RPO $678B +84%. Capex 35% of revenue crushes FCF - the bear case. Fair price: bear $453 / base $567 / bull $680 (15% growth). Accumulation zone $317-453. Watch: Azure >=35%, Cloud GM >=60%, capex/revenue turn. Earnings: late Oct.

One-Pager

This is not a financial or investing recommendation. It is solely for educational purposes.

TL;DR: NVIDIA is the dominant AI compute platform (GPUs + CUDA + networking) and the fastest-growing mega cap in the market. At ~21x forward earnings (vs a 37.7x five-year average) and a PEG of 0.47, this is the cheapest the stock has been in five years. Quality stays extreme: 74% gross margin, 70% ROIC, ~$119B of free cash flow. Verdict: I am starting to accumulate at these levels as a core Long-Term Pick position. My full fair price math and the accumulation zone are at the end of this post.

Investment Thesis

I first covered NVIDIA in February 2025, with a fair price of $187, back when the market was arguing about whether AI demand had already peaked. Since then, earnings have more than doubled, the stock has climbed to ~$200 - and the multiples have decreased anyway. I see that combination as rare, and it is the reason for this update.

• NVIDIA is the fastest-growing mega cap in the market at its lowest relative valuation in five years: ~21x forward earnings against a 37.7x five-year average - below even the -1 standard deviation band.

• Consensus expects ~44% annual EPS growth for the next five years. That puts the PEG at 0.47 - either consensus is badly wrong, or the stock is cheap.

• Quality is not the debate: 74% gross margin, 63% net margin, 70% ROIC, ~$119B of free cash flow, ~$40B of net cash, and management sees more than $1 trillion of cumulative Blackwell and Rubin revenue through 2027.

• The risks are real and partly binary: customer concentration, custom silicon, China closed off, Taiwan, peak margins. My model cuts consensus growth by more than half to absorb them.

• Even with that cut, the stock trades below my Bear Case fair price. The full math is in the Fair Price section below.

Company Overview

Next Earnings Date: Aug 26, 2026 (after-market, confirmed)
Market Cap: ~$5T
Sector: Information Technology | Industry: Semiconductors
Type: Mega Growth
Beta (5Y Monthly): 2.21 | Short Interest: 1.30%
Dividend Yield: 0.49%
Employees: ~42,000

NVIDIA is the leading designer of GPUs and the dominant supplier of AI computing infrastructure. GPUs process data in parallel across thousands of cores, which turned out to be exactly what training and running large AI models requires. And NVIDIA owns that market. The company has evolved from a chip vendor into a full-stack platform: it designs the processors, the networking, the systems, and the software layer on which most of the world’s AI is built, while TSMC handles manufacturing.

The product stack today: Data Center accelerators (Hopper, Blackwell, Blackwell Ultra GB300, and the Vera Rubin rack-scale systems now in full production); the new Vera CPU; networking that ties clusters together (NVLink, InfiniBand, Spectrum Ethernet, ConnectX NICs, BlueField DPUs); the CUDA platform and CUDA-X libraries with 7,000+ accelerated applications, models, and libraries; GeForce RTX graphics cards for gaming; the RTX Spark superchip taking NVIDIA into the PC processor market against Intel and AMD; DGX systems and DGX Cloud; professional visualization; automotive (DRIVE platform and the Alpamayo reasoning model for autonomous vehicles); Jetson Thor modules for robotics and edge AI; Omniverse and Cosmos for simulation and world models; the Isaac GR00T humanoid-robot platform; and the Nemotron family of open models.

Market Overview

The core question for NVIDIA is not market share - it is whether the AI infrastructure buildout continues. The freshest data say demand is accelerating, not cooling:

• Token generation is growing exponentially: from ~2 trillion tokens per month in January 2025 to ~123 trillion in April 2026 on OpenRouter alone - roughly 60x in fifteen months.

• GitHub shows what those tokens produce: 1.4B commits in early 2026 (nearly 3x the prior pace), 90M pull requests merged, 20M new repositories per month. The weekly rate is even steeper: ~275M commits per week versus ~20M a year ago - a 14x jump, as AI agents flood the platform with code. AI is now useful, and tokens are profitable, which is what stimulates operators to keep expanding compute.

• The economics behind that: ~30-40M software developers earning ~$3T in annual salaries are producing roughly 3x the output. And companies are hiring more engineers, not fewer. When the tool visibly pays for itself, the infrastructure spending behind it is not discretionary.

• Cloud GPU pricing is rising, not falling: per SemiAnalysis data, H100 rental prices are up ~21% over the trailing six months, and even the six-year-old A100 is up ~5%. Supply is still not catching up with demand.

• Management projects $3-4 trillion of annual AI infrastructure spending by 2030.

The bear narrative says AI capex is a bubble about to deflate. The market data says the opposite: used-generation GPUs are getting more expensive to rent. Bubbles do not usually raise the price of four-year-old hardware.

Economic Moat

NVIDIA has a wide moat built on two reinforcing layers. The first is CUDA: virtually all AI development of the last 15 years is built on it, and millions of developers create switching costs so high that even hyperscalers designing their own chips keep buying NVIDIA at scale. The second is extreme co-design: NVIDIA engineers the CPU, GPU, DPU, networking fabric, systems, and software as one integrated platform on a one-year cadence - a bar that a single-chip competitor cannot clear.

In the industry-standard MLPerf round, NVIDIA won every training benchmark and was the only platform submitting on all tests. In SemiAnalysis’s independent InferenceX testing, the GB300 NVL72 delivers up to 35x lower cost per token and 50x higher throughput per megawatt versus Hopper; Artificial Analysis measured 20x more coding agents per megawatt than H200. The reasoning-throughput record of 2.5M tokens per second is company-reported. In a power-limited data center, tokens per watt decide the economics - and that is exactly where the integrated platform wins.

More developers build on NVIDIA -> more models are optimized for it -> more workloads run on it -> more clouds deploy it -> which funds the next platform generation. Each turn of the wheel raises the cost of leaving.

Business Strategy

Three things stand out in the current strategy. First, the relentless roadmap cadence: Blackwell (2024) -> Rubin with HBM4 (2026, in full production) -> Rubin Ultra -> Feynman (2028), with networking, CPUs, and DPUs advancing in lockstep.

Second, demand diversification. AI Clouds, Industrial, and Enterprise customers now account for ~50% of Data Center revenue - the half that is not hyperscalers - and management expects it to outgrow hyperscale long-term. Data Center revenue itself went from $48B (FY2024) -> $115B (FY2025) -> $194B (FY2026) -> $75B in Q1 FY2027 alone. The customer list now spans AI labs (Anthropic, OpenAI), AI natives (Cursor, Perplexity), and enterprises (Eli Lilly, Samsung, Tesla, TSMC itself).

Third, new fronts beyond the GPU: the standalone Vera CPU (1.5x the performance of the latest 128-core x86 in company-cited benchmarks), RTX Spark in PCs, and physical AI - robotics, autonomous vehicles, and industrial automation, which management sizes at $50 trillion; treat that number as ambition, not addressable market. NVIDIA is the only company offering all three computers physical AI needs: train (DGX), simulate (Omniverse + Cosmos), run (Jetson Thor).

Capital Allocation

The numbers here barely need commentary: operating cash flow grew from $64B in FY2025 to $103B in FY2026, and hit $50B in Q1 FY2027 alone. Free cash flow went from $61B -> $97B, with $49B in the first quarter of FY2027.

Cash and short-term investments of $53B against $12.81B of debt - ~$40B of net cash after adjustments, with interest coverage above 400x.

Capital returns are stepping up to match: a new $80B buyback authorization on top of ~$39B remaining, ~$20B returned in Q1 FY2027 alone, and a commitment to return 50% or more of free cash flow to shareholders going forward.

The May 2026 dividend raise from $0.01 to $0.25 per quarter - a 25-fold jump; the trailing payout ratio is still just 0.61%, and even the new run-rate is ~15% of earnings. It is a statement of confidence, not an income proposition; buybacks remain the main channel, and the share count is already reducing.

Unlike its hyperscaler customers, NVIDIA carries almost no capex burden: 2.59% of revenue, versus 3.60% on average over five years. It designs chips; TSMC builds them. That is why a 47% FCF margin is even possible.

Advantages

• De facto monopoly on AI compute. GPUs plus CUDA form the default platform for training and inference; switching costs are so high that even customers building their own silicon keep buying NVIDIA at scale. Networking deepens the lock-in at cluster level, and new fronts (PCs, robotics, CPUs) keep opening.

• Unmatched financial profile. 74% gross margin, 63% net margin, 70% ROIC, 114% ROE, ~$119B of free cash flow, ~$40B net cash - every margin and return metric is far above its own five-year average. Nobody else converts demand into cash this efficiently.

• Hypergrowth priced as a mature company. ~21x forward earnings (vs a 37.7x average) for ~44% expected EPS growth is a PEG of 0.47 - the cheapest the stock has been relative to its growth in five years, with a +51% average analyst target on top.

• Demand is diversifying. AI Clouds, Industrial, and Enterprise now generate ~50% of Data Center revenue - the non-hyperscaler half - and management sees $1T+ of cumulative Blackwell and Rubin revenue through 2027. The single-customer-type risk of 2023-2024 is fading quarter by quarter.

• Capital returns. A new $80B buyback on top of $39B remaining, the dividend raised 25-fold, a commitment to return 50%+ of free cash flow, a decreasing share count - and stock-based compensation of only 2.7% of revenue, rare for a company this dominant.

Disadvantages

• Customer concentration and cyclicality. A handful of hyperscalers and AI labs generate most Data Center revenue, and every one of them is developing in-house silicon. If AI capex pauses because returns disappoint, the same operating leverage that created 63% margins works brutally in reverse.

• Geopolitics. Export controls have largely cut NVIDIA off from the Chinese AI market, Taiwan concentration is a single point of manufacturing failure, and the AI supply chain is in the middle of US-China tensions. These risks are real, partly binary, and outside management’s control.

• Peak-margin risk. Consensus extrapolates 74% gross margins years forward. AMD’s accelerators, Broadcom’s custom ASICs, and customers’ own chips all target exactly this profit pool; even modest erosion would compound with decelerating growth (+82% -> +23% by FY2029) to make today’s multiples look less cheap.

• Volatility. Beta of 2.21, and NVDA’s own history includes 50%+ drawdowns inside secular uptrends (2022: -51.48%). Whatever the fair price says, the market will periodically offer this stock 30-40% cheaper - and holding through that is part of the price of admission.

• The supply chain. Leading-edge manufacturing is concentrated at TSMC, advanced packaging capacity is scarce, and HBM memory comes from a handful of suppliers - any bottleneck there caps growth regardless of demand. Add the key factor: no mega cap is more identified with one founder-CEO than NVIDIA with Jensen Huang.

What the bears get right: nobody, including NVIDIA, actually knows the demand curve past 2027. Semiconductor consensus has been wrong at every cycle turn in history, and it is currently extrapolating the greatest earnings boom ever recorded. AI monetization might face a slowdown. Today's "cheap" 21x becomes an expensive 40x retroactively, because the E collapses faster than the P. I do not dismiss that scenario - I price it: the 20% growth cap, the 20x bear exit multiple, and a 30% margin of safety on top exist precisely because the bears have a point.

Competitors

AMD is the only merchant GPU alternative with a credible roadmap; Broadcom builds custom AI ASICs for hyperscalers; Intel remains behind in both.

The charts below answer the obvious question. Yes, consensus expects AMD to grow EPS faster over the next five years - 54.42% a year versus NVIDIA’s 44.36%, but that is the only line AMD wins, and the market has already paid for it: ~56x forward earnings versus ~21x, and a PEG of 1.03 versus 0.47.

Everything else is not close. NVIDIA turns a dollar of revenue into 63 cents of profit, AMD into 13; NVIDIA’s ROIC is 70%, AMD’s is 6.5%. Both balance sheets are clean, so leverage decides nothing here. I like AMD as a business, and competition keeps NVIDIA honest - but at these prices the “cheaper alternative” is the expensive one.

The real long-term competition is the customers themselves - Google’s TPU is the only non-NVIDIA platform deployed at serious scale, with Amazon’s and Microsoft’s chips further back. So far the pattern has held for a decade: in-house chips absorb some internal workloads, while external demand and frontier work stay on NVIDIA.

Past

• FY2026 results: revenue $215.9B, up 65%; GAAP operating income $130.4B, up 60%; diluted EPS $4.90, up 67%; GAAP gross margin 71.1%, down 3.9 points on the Blackwell ramp (LTM gross margin has since recovered to 74%).

• LTM: revenue $253.5B, net income $159.7B, free cash flow $119.1B.

• Total return CAGR: 57.6% over five years and 66.9% over ten, versus 12.98% and 15.25% for the S&P 500.

Future

• Consensus revenue: FY2027 $393.6B (+82%), FY2028 $560.8B (+42%), FY2029 $688.0B (+23%).

• Consensus EPS: $8.99 -> $12.87 -> $15.98 (FY2027-FY2029). Five-year forward EPS CAGR: ~44%, versus a ~29% five-year mean of that same estimate.

• Next report Aug 26: revenue estimate $91.8B, EPS estimate $2.08.

• Analysts: Strong Buy (10 Strong Buy/48 Buy/2 Hold/1 Sell, 61 covering), average target $302.83, +51% upside; the lowest target on the street is $180.

Growth is decelerating in percentage terms, which is normal at this size.

Current Valuation

Price/Fwd Earnings: 20.8x vs 37.7x 5Y average
Price/Fwd Sales: 11.1x vs 17.5x
Price/FCF: 40.7x vs 81.3x
Price/Book: 24.8x vs 36.0x
PEG: 0.47 vs 1.27
Fwd Earnings Yield: 4.81% vs 2.65%

Every multiple is far below its five-year average, several below their -1 standard deviation bands. Earnings simply grew much faster than the price. The forward EPS estimate has now overtaken what the market is willing to pay for it - the market pays less than half a unit of valuation per unit of expected growth.

One chart frames this whole section. The corridor below takes the consensus forward EPS estimate and multiplies it by the three exit multiples from my fair price model: 20x - roughly today’s multiple, a market that never re-rates; 28x - roughly the -1 standard deviation band of recent years; and 38x - NVIDIA’s own five-year average. Today that corridor runs from ~$199 to ~$378, and the price, at ~$200, is pressed against the very bottom of it. The market is pricing NVIDIA as if today’s skepticism is permanent - every re-rate scenario is upside.

Fair Price

I use 20% annual EPS growth - and that is not a forecast; it is a rule: 20% is the maximum growth rate I ever plug into this model, no matter what the estimates say. Consensus expects ~44%; the FY2026-FY2028 estimates imply ~34% a year. NVIDIA’s estimates are also the most fragile in mega-cap tech, so the model should not need them to be right. With ~0.4% from dividends, total expected growth is 20.4% a year, turning FY2026 EPS of $8.99 into ~$22.75 by 2031.

The exit multiples are 20x/28x/38x: 20x is roughly today’s forward multiple (the market never re-rates), 28x is roughly at the -1 standard deviation band of recent years, and 38x is simply NVIDIA’s own five-year average.

Bear case (exit P/E 20x): fair price $258 - MoS price $181
Base case (exit P/E 28x): fair price $361 - MoS price $253
Bull case (exit P/E 38x): fair price $490 - MoS price $343

At ~$200, the stock trades below the Base Case MoS price of $253 - the full 30% margin of safety is already in the price even though the model cuts consensus growth by more than half. Even against the bear case, the stock is ~24% below a $258 fair price, and the Bear Case MoS price of $181 is almost exactly the lowest analyst target on the street. The accumulation zone is $181-258, and today’s price is inside it.

For the track record: my February 2025 fair price was $187. The stock passed it, and instead of becoming expensive, it became cheaper - because earnings more than doubled while the multiple compressed. That is the update in one sentence.

Verdict: NVIDIA belongs in a long-term portfolio as a core AI-infrastructure position - sized for its volatility, not its quality. I am personally starting to accumulate at these levels, inside the $181-258 zone, and I treat NVDA as one of the core companies of my Long-Term Pick portfolio going forward. The realistic bear case (capex digestion, margin normalization) hits the multiple and the estimates at once, so drawdowns of 30-40% are a feature of the position, not a broken thesis. Buy with a multi-year horizon, and judge the thesis on hyperscaler capex guidance and the Rubin ramp, not the share price.

Checklist

Profitability:

• Gross margin at least 40%: 74.2%

• Net margin at least 10%: 63.0%

• FCF margin at least 10%: 47.0%

• Management (ROIC, ROE, ROA): Yes (all far above 10%: 70%/114%/53%)

• Piotroski F-Score: 8 of 9

• Revenue surprises in last 5 years: Yes (Based on TradingView’s data)

• EPS surprises in last 5 years: Yes (Based on TradingView’s data)

• EPS growth YoY 5 years in a row: No (the 2022 downcycle; Based on TradingView’s data)

Valuation and Advantage:

• Valuation below its 5Y averages: Yes (every multiple, several below -1 standard deviation)

• Valuation below the industry: Yes (P/Fwd E 20.8x vs SOXX at 23.3x)

• Does it have a moat: Yes (wide)

• Outperformed the S&P 500 10-year CAGR: Yes (66.9% vs 15.3%)

Shares:

• Insider ownership at least 5%: No (~4%, mostly Jensen Huang)

• Fewer shares outstanding YoY: Yes

• Insider buys last six months: No (Based on FinViz’s data)

Price:

• 1Y price forecast is above 10%: +51%

• Next 5Y EPS growth estimate (CAGR) is above 10%: Yes (~44%)

• DCF Value: ~$213; undervalued by ~8% (5 years, revenue CAGR ~29% - below consensus, discount rate: 10%, terminal growth: 3%, equity model: FCFF)

• Short Interest below 5%: Yes (1.30%)

Due Diligence

Profitability (12 of 12):

• Positive Gross Profit: $187.95B (for the last twelve months)

• Positive Operating Income: $162.28B (for the last twelve months)

• Positive Net Income: $159.71B (for the last twelve months)

• Positive Free Cash Flow: $119.08B (for the last twelve months)

• Exceptional 1-Year Revenue Growth: 65% (FY2026)

• Exceptional 3-Year Revenue Growth: ~100% (per year for the last 3 years: $26.97B -> $215.94B)

• Exceptional Revenue Growth Forecast: ~47% (per year over the next 3 years, consensus)

• Exceptional ROE: 114% (for the past 12 months)

• Exceptional 5-Year Average ROE: 75%

• ROE is increasing: ~20% -> 114% (in the last 3 years)

• Exceptional ROIC: 70% (for the past 12 months)

• ROIC is increasing: ~14% -> 70% (in the last 3 years)

Solvency (6 of 6):

• Total assets ($259.47B) exceed total liabilities ($64.00B) by 4x

• Negative Net Debt: -$40.36B (cash and short-term investments of $53.17B against $12.81B of debt)

• Low Debt-to-Equity Ratio: 0.07

• Debt-to-Capital: 6.2% (5-year mean: 22.2%; the balance sheet keeps getting cleaner)

• Interest coverage (FFO): 421.6x

• Altman Z-Score: 51.13

Watchlist Note

Dominant AI compute platform (GPUs + CUDA + networking). ~21x fwd P/E vs 37.7x 5Y avg for ~44% consensus EPS growth; PEG 0.47. 74% GM, 63% NI margin, 70% ROIC, ~$119B FCF, ~$40B net cash. Fair price: bear $258 / base $361 / bull $490 (20% growth cap). Accumulation zone: $181-258. Watch: hyperscaler capex guidance, GM >=70%, Rubin ramp. Earnings: Aug 26.

One-Pager

The whole analysis on one page - save it or share it with someone who still thinks NVDA is expensive.

This is not a financial or investing recommendation. It is solely for educational purposes.