Microsoft’s Anthropic investment closed in November 2025 at $5 billion. Eight months later, the Q4 FY26 disclosure shows the position generated a $3.2 billion gain against that basis. For context, Microsoft’s much older and much larger OpenAI investment added $5 billion in gains for the entire fiscal year. The Anthropic stake produced 64% of the OpenAI full-year figure in a single quarter, on roughly one-tenth of the original commitment size.

What this reflects is not just Anthropic’s valuation trajectory — it is Microsoft’s timing on entry. The November 2025 investment came at a valuation multiple that has since expanded significantly as Anthropic’s Claude Fable and later Opus 4.7 models drove enterprise adoption. Meanwhile OpenAI’s valuation has been more volatile through FY26, with several markup and markdown cycles as the market has re-rated its economics. The mark-to-market difference between the two positions is what produced the divergent quarterly numbers.

The unusual thing about the disclosure is that Microsoft doesn’t normally break out Anthropic every quarter. Per TechCrunch’s read of the release, Microsoft discusses its OpenAI investment quarterly but treats Anthropic as a routine equity holding that only gets disclosed when it moves the earnings needle. The fact that Q4 FY26 was the quarter that changed indicates the position has grown to a size where investors need to see it separately.

The Microsoft Anthropic transaction in November 2025 was not a straight equity investment. It was a circular capital arrangement. Microsoft put $5 billion in equity into Anthropic. Anthropic agreed to spend $30 billion buying Azure services over a multi-year commitment window. Money moves in a loop: Microsoft cash → Anthropic equity → Azure revenue → back to Microsoft.

Circular deals like this are common in the AI infrastructure market right now. Nvidia has similar arrangements with its largest cloud and lab customers. OpenAI’s original Microsoft partnership had a circular element as well. What makes the Microsoft Anthropic version distinctive is the ratio — 6x return commitment on the equity stake, structured as multi-year Azure consumption rather than a one-time revenue recognition event.

The accounting treatment matters. Microsoft records the equity investment on its balance sheet and marks it to market. The $30B Azure commitment gets recognized as revenue as Anthropic actually consumes the compute. So Microsoft is booking both the mark-to-market gain on the equity position and the operational revenue from the Azure services, effectively getting two lines of upside from one strategic transaction.

The Anthropic story would be interesting in isolation. Set next to what happened to Microsoft’s OpenAI position in the same quarter, it becomes structural. Microsoft marked its OpenAI investment down by roughly $600 million in Q4 FY26, subtracting $0.07 from diluted EPS. On the full-year basis OpenAI still generated a $5B gain and $0.67 EPS boost, but the quarterly direction reversed.

What drove the OpenAI markdown is not fully broken out in the release, but two factors are visible in the wider market. First, competitive pressure — Anthropic’s Claude family and Google’s Gemini have both taken enterprise share from OpenAI over the last two quarters. Second, OpenAI’s own economics — the company has been aggressively repricing its API tiers and adding new pricing complexity that has slowed monetization at the enterprise level.

The FY25 comparable makes the reversal more striking. In fiscal 2025, Microsoft’s OpenAI investment produced a $3.6 billion loss and subtracted $0.49 from EPS. FY26 flipped that to +$5B and +$0.67. The volatility in Microsoft’s OpenAI mark is now a defining feature of the quarterly earnings pattern. Adding a second, uncorrelated AI lab position — Anthropic — dampens that volatility at the portfolio level.

Alongside the Microsoft Anthropic disclosure, Q4 FY26 confirmed Azure crossed $100 billion in annual revenue for the first time. Q4 Azure growth was 43%, above the 39-40% Amy Hood guided in April and well above Wall Street expectations. Intelligent Cloud revenue was $39.3 billion, up 32%. Microsoft Cloud in aggregate was $59.3 billion, up 27%.

The Azure numbers are the operational backdrop for why the Microsoft Anthropic hedge matters strategically. Google Cloud grew 82% last quarter to $24.8 billion — nearly double Azure’s growth rate. Microsoft needs a second AI franchise to complement the Azure/OpenAI stack because the Google Cloud/Gemini stack is compounding faster on the top line. Anthropic, distributed largely on AWS but with a $30 billion Azure commitment, is Microsoft’s way of ensuring Azure captures Anthropic’s growth even if the model itself runs partially on competitor clouds.

Commercial remaining performance obligations grew 84% to $678 billion. That backlog is the visibility signal enterprise investors care about. Google Cloud’s comparable RPO figure is $514 billion. Azure has more total backlog by dollar amount, though Google Cloud’s percentage growth of RPO is higher. The Microsoft Anthropic deal contributes to Azure’s RPO figure through Anthropic’s multi-year commitment — the $30 billion converts to backlog and then to recognized revenue as consumption occurs.

Microsoft has been deliberately opaque about Copilot seat counts for two years. Q4 FY26 changed that. Satya Nadella disclosed that Microsoft 365 Copilot has passed 30 million paid seats. That number contextualizes the enterprise AI adoption story: 30 million paid Copilot seats represents about 6.7% of the estimated 450 million Microsoft 365 commercial customer base.

The Register noted that 6.7% is “not an overwhelming vote of confidence” given Microsoft’s aggressive push and the two years since Copilot launched. Fair. But it is a meaningful number in absolute terms. GitHub Copilot alone was well under 3 million paid seats a year ago. Google’s Antigravity has 2.4 million weekly active users per last week’s Alphabet disclosure. 30 million is now the benchmark other AI productivity tools get compared against — Microsoft has effectively set the ceiling.

The pricing model shift is worth noting too. Microsoft imposed usage-based billing on top of seat-based charges earlier this year, meaning the 30 million paid seats produce variable revenue depending on token consumption rather than pure per-seat economics. This is the same structural repricing that has been eating traditional per-seat SaaS vendors. Microsoft is doing the pivot on its own terms with 30 million seats already committed.

Microsoft stock jumped 7% in extended trading on July 29 after the earnings release, per CNBC. That is a meaningful move for a stock that had been down 19% YTD heading into the print — the biggest laggard among the mega-cap software names inside this year’s Software Selloff. The rebound came before the market has fully digested the Microsoft Anthropic disclosure, and after the operating beats and Azure milestone drove immediate coverage.

The market’s read is that Microsoft delivered a rare “clean” print in a quarter where investors had assumed the worst. Azure growth beat guidance. Copilot seats confirmed the enterprise attach. Anthropic gains gave the EPS an unexpected lift. Capex plan was reiterated at $175 billion for FY27 with a mechanical accounting change (data center useful life extending from 15 to 25 years) that reduces reported depreciation without changing actual spending. Free cash flow was down 23% to $19.6 billion, but Amy Hood guided to positive FCF for FY27.

The 7% pop puts Microsoft back near $420 in extended trading, above the $400 breakpoint that dominated the Software Selloff coverage last week. Whether the rebound holds through today’s session depends on how the analyst desks parse the useful-life accounting change and the ongoing $175B capex line. But the Microsoft Anthropic disclosure specifically has changed the AI-portfolio narrative around Microsoft in a way that will persist beyond one earnings cycle.