AMD’s latest results show a company changing scale: quarterly revenue reached $11.5 billion and Data Center revenue more than doubled year over year. The harder question is whether that growth will become durable free cash flow per diluted share—and whether the return on the capital and research required to produce it can justify the expectations already attached to the business.
This analysis uses reported information through June 27, 2026, company guidance published August 4, and demand observations collected through September 12. Popularity is treated only as a research-discovery signal, never as a buy signal.
Reviewed by: Stock Metric Lab (source and calculation review) Data cutoff: September 13, 2026
Key takeaways
- AMD reported Q2 2026 revenue of $11.536 billion, up 50% year over year. Data Center supplied 58% of total revenue and grew 107%.
- A mechanical trailing-12-month calculation produces approximately $41.3 billion of revenue and $7.7 billion of free cash flow. This is our calculation, not company guidance, and it needs normalization before valuation.
- Balance-sheet liquidity is a strength: cash and short-term investments totaled $13.1 billion at quarter-end against approximately $3.2 billion of debt.
- The quality-of-growth question is more important than the headline growth rate. Investors should separate operating progress from the prior-year export-control charge, unrealized investment gains, stock-based compensation, acquisitions, and working-capital movements.
- AMD’s customer partnerships can expand demand, but customer warrants covering as many as 320 million shares create potentially material dilution if their vesting conditions are met.
- Our illustrative cash-flow valuation range is deliberately wide. Even aggressive assumptions demonstrate how much sustained cash-flow growth and margin conversion matter when valuing a fast-growing semiconductor company.
Why investor attention increased
Advanced Micro Devices designs processors rather than manufacturing them in its own leading-edge factories. Its principal products include EPYC server CPUs, Instinct AI accelerators, Ryzen PC processors, Radeon graphics products, and adaptive and embedded computing products acquired with Xilinx. Third-party foundries—most importantly TSMC—manufacture its advanced chips. That “fabless” model limits factory spending but leaves AMD dependent on foundry capacity, advanced packaging, and memory supply.
Public attention around AMD has not moved in a straight line. English Wikipedia pageviews were 753 in the seven days through September 12 versus 920 in the preceding seven days, a decline of about 18%. AMD also did not appear in Monex’s top-10 U.S. stock trading-value ranking for August 31 through September 4. Those are useful counterweights to the temptation to equate company announcements with broad investor demand.
The fundamental catalyst is clearer. AMD’s August 4 results showed record revenue and a 107% increase in Data Center revenue. Management guided Q3 revenue to approximately $13.0 billion, plus or minus $300 million, with the midpoint implying about 41% year-over-year growth. It also expected non-GAAP gross margin of roughly 56%.
This combination—rapid fundamental growth but mixed audience-demand signals—is precisely why the article should focus on economics rather than excitement.
How AMD connects to Stock Metric Lab’s framework
AMD sits at the intersection of several existing Stock Metric Lab research clusters.
First, our 2026 AI CapEx comparison examines spending by large cloud platforms. AMD is a supplier-side test of whether that spending becomes revenue, operating profit, and ultimately free cash flow rather than remaining a technology narrative.
Second, AMD can be read alongside the site’s NVIDIA analysis, Broadcom analysis, and TSMC analysis. NVIDIA is the most direct accelerator and software-ecosystem comparison; Broadcom helps frame custom silicon and networking economics; TSMC highlights AMD’s dependence on external leading-edge manufacturing and advanced packaging.
Third, AMD makes the site’s analytical tools concrete. The FCF Yield Calculator can test normalized rather than peak cash flow. The DCF Calculator shows how valuation changes with FCFF growth, WACC, and terminal growth. The Reverse DCF Calculator is especially useful for asking what growth a current market price requires without turning the article into a price target.
Readers should also use the free-cash-flow analysis guide and ROIC versus WACC guide when judging whether AMD’s investment is creating value.
Business model and economic drivers
AMD reports three operating segments: Data Center, Client and Gaming, and Embedded. The mix now matters as much as total revenue.
Data Center produced Q2 revenue of $6.718 billion and operating income of $2.103 billion. Its 31% segment operating margin made it both the largest growth engine and the largest identified contributor to segment profit. Demand comes from EPYC server CPUs, Instinct accelerators, and related infrastructure. The economic variables to watch are deployment scale, software adoption, supply availability, selling prices, and the share of systems economics AMD can retain.
Client and Gaming generated $3.841 billion of revenue and $582 million of operating income. Client revenue rose 23% to $3.062 billion, while Gaming fell 31% to $779 million. Combining the businesses in one reported segment can obscure opposing cycles, so both subcategories deserve separate monitoring.
Embedded revenue was $977 million with $386 million of operating income, a 40% segment margin. This remains a profitable franchise, although its end markets and recovery cycle differ from AI accelerators.
Finally, corporate and acquisition-related costs are substantial. AMD reported an All Other operating loss of $1.081 billion in Q2. Segment profit should therefore never be mistaken for consolidated operating profit.
Latest reported results
| Metric | Q2 2026 | Q2 2025 | Change |
|---|---|---|---|
| Revenue | $11.536bn | $7.685bn | +50% |
| GAAP gross margin | 54% | 40% | +14 pts |
| GAAP operating income | $1.990bn | $(0.134)bn | NM |
| GAAP net income | $2.297bn | $0.872bn | +163% |
| Diluted EPS | $1.38 | $0.54 | +156% |
The comparison needs an important qualification: Q2 2025 included $800 million of inventory and related charges caused by U.S. export controls on MI308 products. Part of the reported year-over-year margin expansion is therefore the non-recurrence of a prior charge, not solely current operating leverage.
Another adjustment sits below operating income. Other income was $598 million in Q2 and $763 million in the first half, driven mainly by an unrealized gain after a privately held investment became publicly traded. That benefit increased net income but does not demonstrate semiconductor operating performance.
AMD’s full-year 2025 revenue was $34.639 billion, GAAP operating income was $3.694 billion, and GAAP net income was $4.335 billion. A mechanical trailing-12-month bridge—FY2025 plus H1 2026 minus H1 2025—produces revenue of about $41.305 billion and operating income of about $6.488 billion. These figures are calculated from filings and are not company-defined non-GAAP metrics.
Cash flow, reinvestment, and balance-sheet risk
For the first half of 2026, AMD’s Form 10-Q reports continuing-operations cash from operations of $5.321 billion and capital expenditure of $1.197 billion. The simple difference is $4.124 billion, consistent with AMD’s non-GAAP free-cash-flow presentation for the period.
On the same mechanical trailing-12-month basis, continuing-operations operating cash flow was approximately $9.413 billion and capital expenditure approximately $1.677 billion, producing $7.736 billion of simple free cash flow. This is a starting point, not automatically sustainable owner earnings.
Three normalization questions deserve attention:
- Working capital: rapid growth can temporarily raise or release cash through receivables, inventory, and payables.
- Stock-based compensation: first-half stock-based compensation was approximately $990 million. It is non-cash in the cash-flow statement but can dilute shareholders and is an economic cost.
- Acquisitions and intangibles: AMD’s asset-light manufacturing model keeps factory capital expenditure below that of an integrated manufacturer, but acquisitions, acquired technology, and amortization still represent real reinvestment choices.
At June 27, AMD held $5.086 billion of cash and $8.025 billion of short-term investments, versus approximately $3.226 billion of current and long-term debt. Net cash and investments were therefore about $9.885 billion by simple subtraction. That cushion reduces financing risk, but it does not eliminate valuation risk.
ROIC: a useful question with an accounting trap
AMD spent heavily on research and development to create products whose benefits can last for years. GAAP accounting expenses R&D as incurred. A conventional ROIC calculation can therefore understate invested capital during periods of heavy development and overstate returns later if prior research produces current revenue.
A better analytical range includes both reported and R&D-capitalized versions of NOPAT and invested capital. The objective is not to manufacture a flattering number; it is to test whether returns stay above an appropriate cost of capital after treating research consistently.
Investors should monitor consolidated operating margin, incremental operating margin, and normalized FCFF alongside ROIC. If Data Center revenue expands but consolidated returns do not, the value captured may be smaller than the market opportunity suggests.
Valuation questions and scenario assumptions
The table below is an illustrative 10-year equity cash-flow exercise, not a price target or a full valuation model. It starts with the $7.736 billion mechanical trailing-12-month operating-cash-flow-minus-capex estimate and divides by the 1.632 billion shares outstanding reported as of July 29, 2026. Because this cash-flow measure is after interest, the calculation does not add net cash or subtract debt. Treating it as unlevered FCFF and then adding net cash would mix enterprise- and equity-value conventions. The model assumes constant annual cash-flow growth for ten years; that deliberately demanding simplification makes the sensitivity visible but does not capture a normal fade in growth.
| Scenario | 10-year cash-flow growth | Equity discount rate | Terminal growth | Implied value/share |
|---|---|---|---|---|
| Downside | 12% | 11% | 3.0% | $117 |
| Base | 18% | 10% | 3.5% | $223 |
| Upside | 25% | 9% | 4.0% | $496 |
The spread is the lesson. A high-growth semiconductor valuation is extremely sensitive to how long growth persists, how quickly margins convert to cash, and whether dilution offsets enterprise-value creation. At the September 11, 2026 closing price of $516.13, this same simplified framework requires about 29.5% annual cash-flow growth for ten years when using a 10% equity discount rate and 3.5% terminal growth. This is a model output, not a forecast; changing the starting cash flow, fade pattern, discount rate, terminal growth, or diluted share count changes the answer materially. Readers can reproduce the mechanics in the DCF Calculator, then use the Reverse DCF Calculator.
A more complete model should use a multi-stage forecast and test diluted share counts that include a probability-weighted warrant outcome rather than relying only on currently outstanding shares. The market-price observation above is dated rather than described as real time so readers can distinguish reported financial data from a changing quote.
Dilution deserves its own scenario
AMD disclosed customer warrants for up to 160 million shares each in arrangements with OpenAI and Meta, or as many as 320 million shares in total. At June 27, none had vested. The maximum represents approximately 19.6% of the 1.632 billion shares then outstanding by our calculation.
That is not a prediction of full dilution. Vesting depends on contractual milestones, and successful vesting could accompany substantial revenue growth. The correct treatment is scenario analysis: estimate the incremental revenue, margin, capital needs, and probability-weighted share issuance together. Counting the revenue opportunity while ignoring the associated dilution would be internally inconsistent.
Risks and disconfirming evidence
- Competition and software: NVIDIA’s ecosystem, Intel, custom chips, and customer-owned designs can pressure adoption, pricing, or margins.
- Manufacturing concentration: AMD relies on TSMC for all 7-nanometer-and-below CPU and GPU wafers. Advanced packaging, memory, and other supply constraints can limit shipments.
- Export controls: restrictions on advanced AI products can reduce addressable demand and cause inventory charges or product redesigns.
- Cycle risk: PCs, gaming, embedded products, and data-center infrastructure do not move in lockstep. Forecast errors can create inventory and purchase-commitment risk.
- Earnings quality: unrealized investment gains and the prior-year export-control charge complicate headline comparisons.
- Dilution: stock-based compensation and customer warrants can make per-share value grow more slowly than enterprise value.
- Attention signal: the latest sampled Wikipedia and Monex observations do not show a broad demand surge. This weakens the case for treating the article as urgent and supports keeping it in editorial review.
The strongest counterargument to a cautious valuation framework is that a new computing platform can grow far faster and longer than historical semiconductor cycles suggest. Evidence that would support that view includes sustained Data Center growth, rising consolidated GAAP margins, cash conversion that remains strong after working-capital normalization, expanding ROCm adoption, and limited dilution per dollar of incremental FCFF.
What to monitor next
- Data Center revenue growth and segment operating margin.
- Consolidated GAAP operating margin versus non-GAAP margin.
- Operating cash flow, capital expenditure, and working-capital movements.
- Stock-based compensation and diluted weighted-average shares.
- Warrant vesting milestones and the economics attached to them.
- MI-series deployment disclosures and ROCm adoption indicators.
- TSMC capacity, advanced-packaging availability, and memory supply.
- Export-control changes and any related inventory or compliance charge.
- Client, Gaming, and Embedded revenue separately, not only the combined segment.
Primary sources and methodology
- AMD Q2 2026 earnings release, August 4, 2026.
- AMD Form 10-Q for the quarter ended June 27, 2026.
- AMD Form 10-K for the year ended December 27, 2025.
- Nasdaq AMD historical quotes, used for the September 11, 2026 market-price reference.
- Nasdaq AMD historical-data page, market-price reference dated September 11, 2026.
- Monex U.S. stock trading-value ranking, week of August 31 through September 4, 2026.
- Wikimedia Pageviews API for the English-language AMD article, sampled August 29 through September 12, 2026.
Reported figures are identified as such. Trailing-12-month FCF, net cash, dilution percentages, valuation outputs, and the reverse-growth calculation are Stock Metric Lab calculations from the cited filings and dated market data. The September 11 market price is a snapshot, not a live quote; market capitalization is deliberately omitted.
*This article is for educational and informational purposes only. It is not investment, financial, tax, legal, or accounting advice, and it is not a recommendation to buy, sell, or hold any security. Verify current filings and market information before making an investment decision.*
See the NVIDIA vs. AMD vs. Broadcom comparison for a side-by-side test of AMD’s margin, free-cash-flow, and ROIC path against the two larger AI infrastructure models.