For years, the Microsoft-OpenAI partnership ran on press releases and guesswork. On 28 October 2026, the guesswork ended: in a regulatory filing, Microsoft disclosed Microsoft OpenAI revenue of $10 billion for the quarter ended 30 September 2026, counting every commercial arrangement between the two companies, revenue-sharing payments included.
That single line is nearly 11% of Microsoft’s $92.3 billion quarterly total, and it reframes everything written about “divorcing” tech’s most famous alliance. The two companies compete openly for the same customers – and money flows between them at a scale neither has ever admitted on the record. Here is what the Microsoft OpenAI revenue disclosure actually says, where the money comes from, and why it matters even if you never buy either company’s products.
What The Numbers Actually Say
The number arrived in Microsoft’s quarterly SEC filing, the form 10-Q covering the period that ends with its first fiscal quarter of 2027. Microsoft has reported Azure growth for years, but this is among the clearest admissions yet that one customer relationship now moves a visible fraction of the company’s income. A Microsoft OpenAI revenue figure of $10 billion in ninety days is larger than the annual revenues of most Fortune 500 constituents.
Read the disclosure carefully
The filing describes Microsoft OpenAI revenue as income from commercial arrangements with OpenAI, including revenue-sharing payments. That umbrella covers several distinct streams: OpenAI’s consumption of Azure compute, the revenue Microsoft recognises from selling ChatGPT subscriptions through its store channels, and other services sold between the two firms. Microsoft did not itemise the split in the Microsoft OpenAI revenue disclosure.
The comparison that anchors the number: total Microsoft revenue for the same quarter was $92.3 billion, and the company logged its tenth consecutive record-breaking quarter. So one counterparty – still partially owned by Microsoft itself – accounts for roughly one dollar in nine. Ten quarters ago, that relationship barely registered in disclosures at all.
One calendar footnote matters when you read the headlines. Microsoft’s fiscal year begins in July, so the quarter ended 30 September 2026 is Q1 of fiscal 2027, while the $24.1 billion annual figure covers fiscal 2026 (July 2025 through June 2026). Comparisons against OpenAI’s own calendar-year revenue need that offset applied – and so do claims about which company is growing faster.
One quarter, or a run rate?
Zoom out and the trajectory steepens. For its full fiscal 2026, which ended in June, Microsoft reported $24.1 billion from OpenAI-related arrangements. The quarter’s $10 billion is more than 40% of that entire year. If the September quarter pace simply held, annual Microsoft OpenAI revenue would reach about $40 billion – roughly double the fiscal 2026 figure – before any new deal terms.
The disclosed figures show acceleration, not plateau: one quarter now equals nearly half of the full prior fiscal year.
What about the equity stake?
Money flowing to Microsoft is not the same as control fading away quietly. Microsoft’s ownership of OpenAI slipped from about 25% in June to 24% in September, continuing a gradual slide from roughly 27% after OpenAI’s October 2025 recapitalisation. Microsoft did not explain what caused the latest dilution. The shape of the relationship is now unusual in corporate life: a shrinking shareholder that is simultaneously the counterparty’s biggest supplier, distributor, and – on paper – its most dangerous rival. The Microsoft OpenAI revenue line, meanwhile, keeps growing even as the equity stake shrinks – proof that the commercial tie has outgrown the ownership tie.
Where Microsoft OpenAI Revenue Actually Comes From
Ask where Microsoft OpenAI revenue comes from, and three candidates appear: cloud computing consumed by OpenAI, revenue-sharing payments owed under the investment deals, and smaller commercial channels. The filing bundles them, but the surrounding numbers tell a story about proportions.
Azure: the cloud that trains the models
OpenAI runs its frontier training and most ChatGPT inference on Azure. When OpenAI committed to buying an additional $250 billion of Azure services as part of the October 2025 agreement, it effectively pre-purchased the bulk of what Microsoft OpenAI revenue would look like for years. Training runs on tens of thousands of GPUs for weeks, and ChatGPT’s hundreds of millions of weekly users consume inference every hour of every day. That consumption, billed monthly, is almost certainly the heavyweight inside the Microsoft OpenAI revenue figure – the computing bill of the largest AI company on Earth, paid to one landlord.
Revenue share: the 20% question
The second stream is the famous revenue share. Reuters reported in May 2025 – and neither company has publicly confirmed the figure – that OpenAI shares around 20% of its revenue with Microsoft, the consideration for more than $13 billion of early investment. In November 2025, TechCrunch reported citing leaked documents that Microsoft had received $865.8 million in net revenue-sharing payments during the first nine months of 2025.
Put those figures side by side and the arithmetic is blunt. Even generously annualised, revenue-share payments ran well below $2 billion a year in 2025. A single quarter of total arrangements at $10 billion means the lion’s share of Microsoft OpenAI revenue must be compute and services – OpenAI buying Microsoft’s cloud – not Microsoft collecting its cut.
The streams compared
| Stream | Who pays whom | Scale in the disclosure | Confirmed? |
|---|---|---|---|
| Azure compute consumption | OpenAI → Microsoft | The largest share of the $10B quarter | Partly – bundled figure |
| Revenue-sharing payments | OpenAI → Microsoft | <$2B/yr pace per 2025 leaks | Percentages never confirmed |
| Store & channel fees | OpenAI → Microsoft | Small but growing (app-store cuts) | Reported, not itemised |
| Equity gains | Not counted here | 24% stake, value swings | Separate from revenue line |
How The Deal Between Microsoft And OpenAI Works Now
The disclosure is only legible against the deal architecture underneath it, which has been rebuilt twice in a year – and each rebuild changed what Microsoft OpenAI revenue would even mean. The 2019-era arrangement looks nothing like the current one, and the current one will not survive 2030 unchanged.
From exclusivity to an open door
Through most of the partnership, OpenAI could not run its models on another cloud. That ended in October 2025, when Microsoft relinquished its exclusive claim as part of OpenAI’s recapitalisation into a public benefit structure. The trade: OpenAI won the freedom to buy compute from Amazon, Google, Oracle or anyone else, and Microsoft received a commitment to purchase an additional $250 billion of Azure services – the largest cloud contract ever reported.
The April 2026 agreement
The next revision came in April 2026. Under its terms, OpenAI remains on the hook to keep sharing revenue with Microsoft through 2030, subject to a cap, even while it spreads workloads across competing providers. Satya Nadella said at the time he was ready to “exploit” the new agreement. Read the fine print together and the design intent is clear: Microsoft gave up the lock on OpenAI’s workloads but kept the lock on OpenAI’s money machine, converting exclusivity into a long annuity.
That annuity is what keeps Microsoft OpenAI revenue flowing even as Azure’s share of OpenAI’s actual computing quietly falls – and it is why the two words everyone misreads, “through 2030,” deserve their own paragraph inside any Microsoft OpenAI revenue analysis.
What the $250 billion pledge really is
Cloud commitments of that size are negotiated like sovereign debt: multi-year, tiered, with carve-outs for capacity Microsoft must reserve. The $250 billion is a floor on what OpenAI plans to spend on Azure, not a spending forecast it is obliged to overshoot. For Microsoft investors, it converts volatile AI demand into something closer to a contracted utility bill – and utility bills are exactly the kind of income that sustains Microsoft OpenAI revenue through model-price crashes and funding winters alike.
Deal terms as they stand
| Term | Before Oct 2025 | After Apr 2026 |
|---|---|---|
| Cloud exclusivity | Azure only | OpenAI free to use any provider |
| Committed Azure spend | No fixed pledge | +$250B additional commitment |
| Revenue share to Microsoft | ~20% of OpenAI revenue (reported) | Continues through 2030, capped |
| Microsoft stake | Up to ~27% over time | ~24% after dilution |
| AGI clause | Microsoft claim ended Oct 2023 review | Settled in favour of OpenAI |
Partner And Rival At Once
The strangest feature of the Microsoft OpenAI revenue disclosure is that Microsoft’s largest growth engine is also its declared enemy. In July 2026, TechCrunch reported that Microsoft was training salespeople to talk down OpenAI and Anthropic, and that the company was openly competing with the labs more than ever – pitching its own models, its own agents, its own Copilot everything – while collecting billions from the labs it discourages.
Why sell your rival’s best friend?
Because the money does not care about the rivalry. Enterprises that want OpenAI models still want them inside Microsoft’s procurement, compliance, and security envelope. Microsoft sells the building; OpenAI rents the floor; both charge the tenant. From Microsoft’s side, hosting the competition monetises demand it might lose anyway – and keeps pricing, capacity, and enterprise relationships under Microsoft’s roof.
Azure now hosts Anthropic’s Claude too, on terms announced in late 2025, which turns the cloud into a neutral ground where Microsoft collects from three frontier labs at once. The rivalry is real: Copilot, Microsoft’s own models, and its agent stack compete for the same enterprise budgets. The revenue is real too, and it keeps flowing precisely because enterprise buyers refuse to bet on a single lab – so Microsoft OpenAI revenue is partly a tax on other companies’ choices.
Why investors asked for this number at all
Disclosure rules did not force Microsoft to isolate one customer. Accounting standards only require flagging customers above 10% of revenue, and OpenAI-related arrangements now clear that bar, so the line appears whether Microsoft wanted prominence or not. That is the quiet significance: Microsoft OpenAI revenue has crossed the threshold where secrecy becomes optional and transparency becomes mandatory, the same threshold that once marked Apple at Foxconn or Walmart at its biggest suppliers.
The concentration question
Investors noticed. A relationship producing $40 billion annualised Microsoft OpenAI revenue is also a single point of failure worth $40 billion. If OpenAI loses funding, deters customers, or migrates workloads faster than expected, the growth line in Microsoft’s Intelligent Cloud segment moves with it. Microsoft’s own risk sections have long acknowledged this dependency; the numbers now make it impossible to analyse the company without it.
The Money Trail Beyond Microsoft
Microsoft is one node in a web of reciprocal deals that has come to define AI’s industrial economics, and Microsoft OpenAI revenue is now its most fully measured strand. Nvidia announced plans to invest up to $100 billion in OpenAI, then OpenAI agreed to buy Nvidia chips; Amazon reportedly discussed a $10 billion investment while selling OpenAI AWS capacity; OpenAI signs multi-billion-dollar commitments with Oracle and AMD, often taking equity or warrants in return.
Circular financing, questioned loudly
Critics call this circular financing: money that loops through suppliers, customers, and shareholders until everyone’s revenue looks healthier than end-user demand alone would produce. Reporting on the financing of the AI buildout has flagged systemic-risk concerns precisely because the loops concentrate failure. The Microsoft OpenAI revenue disclosure is the benign case of the same phenomenon – a real, contracted, disclosed flow – but it is still one loop in a machine critics are asking regulators to inspect.
How big the loop actually is
Lay the disclosed and reported commitments end to end and the arithmetic staggers: Nvidia’s up-to-$100 billion investment paired with OpenAI chip purchases, Amazon’s reported $10 billion talks alongside AWS capacity sold to the same company, Oracle and AMD cloud deals with equity features, and Microsoft’s $250 billion Azure commitment generating the $10 billion a quarter now on the record. None of these arrangements is fraudulent; several are reciprocal enough that each party books the other’s spending as growth.
The OpenAI-related revenue line is the first piece of that web forced into daylight, which is exactly why analysts are measuring every other piece against it. Our breakdown of the Anthropic S-1 filing shows the same disclosure pressure arriving at the next lab, and the rise of the neoclouds explains who else is competing for that spend.
Receivables are not revenue
One number in the filing is routinely misread. Microsoft reported $7.2 billion in outstanding receivables from OpenAI at the end of September, up from $6 billion at the end of June. That is money OpenAI already owes for transactions that happened – it sits on the balance sheet, not the income statement, and it is not additional Microsoft OpenAI revenue. It does tell you something useful: the pace of billing between the two firms is outrunning payment, and the scale of trade large enough for a $1.2 billion swing in ninety days.
Receivables also hint at OpenAI’s own cash position. A company turning over hundreds of billions in future commitments can still prefer to defer payments while it hoards capital for data-centre construction. Watch that line in future filings: a falling balance would suggest OpenAI tightening its belt, while a rising one suggests the spending race is still absorbing everything it earns.
What Comes After 2030
If the September quarter pace held, annual Microsoft OpenAI revenue would land near $40 billion in fiscal 2027. But the year 2030 now functions as the visible horizon of the arrangement, and several forces bend the Microsoft OpenAI revenue curve in both directions.
The cap is the countdown
The revenue-share component ends, or hits its cap, in 2030 by design. Everything after that year has to be earned as ordinary commerce: compute, enterprise integrations, channels. Microsoft is hedging by building its own frontier stack, by Copilot’s bundling into every seat it sells, and by signing OpenAI’s competitors to their own cloud deals. The annuity behind roughly a fifth of Microsoft OpenAI revenue‘s growth story has an expiry date written into it; Microsoft has spent 2026 building replacements.
What could shrink the number
Three plausible paths compress the line. OpenAI diversifying training and inference to AWS, Google Cloud, Oracle, CoreWeave, and its own data centres, as its Stargate buildout matures, shrinks Azure’s slice. A funding squeeze at OpenAI reduces compute demand across the industry. And model pricing falling fast means the same tokens bill for less each year. None of these ends the relationship; all of them make the phrase record-breaking harder to repeat – a point Wall Street’s latest earnings flood already put on display.
What could grow it
The bull path is equally concrete: enterprise agent workloads are still tiny and rising, ChatGPT’s consumer subscriptions keep crossing into commerce, and the $250 billion Azure pledge is a floor under OpenAI’s spending for years. If agentic computing becomes the default way software works, Microsoft OpenAI revenue has a lane to keep climbing well past 2030 – with the share component fading and the compute component carrying the line.
There is also a slower current to watch: precedent. Microsoft is the first hyperscaler forced to disclose a single AI-lab relationship this explicitly. If Google’s Anthropic dealings or Amazon’s OpenAI arrangements ever mature into similar revenue lines, expect the same 10% disclosure trigger to drag them into the open within a year or two. The era of invisible AI economics is closing quarter by quarter.
Why Microsoft OpenAI Revenue Matters For You
Even if you do not own a single Microsoft or OpenAI stock, Microsoft OpenAI revenue shapes the market you buy AI in – it sets prices, model availability, and how much any single lab can demand. Here is the practical reading of the Microsoft OpenAI revenue disclosure for each seat at the table.
If you buy AI from Microsoft
You are renting a platform whose economics depend heavily on one anchor tenant. Expect Microsoft to keep bundling its own models aggressively to diversify that dependence – good for prices, ambiguous for neutrality when you compare Copilot’s default model against a first-party rival. Also expect enterprise-grade continuity: Microsoft has every financial reason to keep the Azure AI platform stable, invest in it, and never let service levels wobble for OpenAI’s sake.
If you build on AI
Two $10-billion-scale players competing and cooperating at once creates unusual leverage. Microsoft courted Anthropic onto Azure; Google sells Gemini from inside its own cloud. Compute buyers today can play host and tenant against each other. The practical move: architect for portability now, while prices are falling and every provider wants your workload, because the bargaining chips are cheapest exactly when the market is this contested.
Watch model availability too. When Microsoft OpenAI revenue depends on OpenAI staying healthy on Azure, Microsoft has a conflicting incentive: keep OpenAI’s economics strong enough to keep paying the bills, while selling Copilot against it in the same breath. In practice that means your multi-model option inside Microsoft’s tools is protected by Microsoft’s own profit-and-loss statement – a strange but comforting alignment inside the Microsoft OpenAI revenue machine.
If you just watch the AI economy
Treat the disclosure as the moment the AI economy got a public thermometer. When a single bilateral flow reaches 11% of a trillion-dollar company’s quarterly revenue, concentration risk is no longer hypothetical – it is reportable, auditable, and interesting to regulators. The roots of the AI boom are now visible, tangled, and shared between a handful of firms. What happens to that one line in future filings is a fair proxy for what happens to the entire buildout.
Frequently Asked Questions
How much is Microsoft OpenAI revenue per quarter?
Microsoft disclosed $10 billion for the quarter ended 30 September 2026, its fiscal Q1 2027, in its SEC filing. For the full fiscal 2026 year the figure was $24.1 billion, and at the September quarter’s pace the annual Microsoft OpenAI revenue run rate is about $40 billion.
Is the $10 billion a revenue share from OpenAI?
No. It bundles all commercial arrangements: primarily OpenAI’s purchases of Azure compute, plus revenue-sharing payments and smaller channels. Leaked 2025 documents put net revenue-share payments at $865.8 million over nine months – far too small to explain a quarter of Microsoft OpenAI revenue on their own.
Does Microsoft still own OpenAI?
Yes, partially. Microsoft’s stake was about 24% as of September 2026, down from roughly 27% after the October 2025 recapitalisation. The stake’s financial value is separate from the Microsoft OpenAI revenue line, which is ordinary income from trade between the two companies, not an investment return.
When does OpenAI stop paying Microsoft?
The revenue-sharing obligation continues through 2030 under the April 2026 agreement, subject to a cap. Compute purchases are separate and are governed by OpenAI’s $250 billion Azure commitment, which extends well beyond that date.
Is this partnership or competition?
Both, simultaneously. Microsoft trains its sales force to argue against OpenAI models while selling OpenAI’s workloads on Azure and sharing their revenue back. The disclosure is the clearest proof yet that the rivalry and the revenue are the same deal seen from two ends.
References
Microsoft Form 10-Q, quarter ended September 30, 2026 (SEC EDGAR)
TechCrunch: Microsoft recorded $10 billion in revenue from OpenAI
TechCrunch: Microsoft hits 10 quarters of record-breaking revenue in a row
TechCrunch: Leaked documents shed light on how much OpenAI pays Microsoft
Reuters: OpenAI plans to slash revenue share with Microsoft
TechCrunch: Satya Nadella says he is ready to exploit the new OpenAI deal
TechCrunch: Microsoft is openly competing with OpenAI and Anthropic more than ever
TechCrunch: The billion-dollar infrastructure deals powering the AI boom
TechCrunch: OpenAI completes its for-profit recapitalization
