Microsoft‑OpenAI‑NVIDIA $100B loop

Microsoft, OpenAI and Nvidia are increasingly tied together through contracts that span cloud capacity, model distribution and chip supply. The arrangement has been described by analysts as a circular AI economy because the same companies often appear on both sides of the transaction flow. Public disclosures and company statements show the links are real, even if the exact size of the loop depends on which commitments are counted. The result is a market in which a small group of companies controls large parts of compute, model access and enterprise distribution. ### Which public facts support the idea of a Microsoft-OpenAI-Nvidia loop? Microsoft has invested more than $13 billion in OpenAI since 2019, according to the companies’ April 27, 2026 announcement on their amended partnership. Under that deal, Microsoft remains OpenAI’s primary cloud provider, OpenAI products are still slated to ship first on Azure unless Microsoft declines, and OpenAI continues paying Microsoft a 20% revenue share through 2030, subject to a total cap. Nvidia and OpenAI added a second major leg to that structure on September 22, 2025, when they announced a strategic partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI infrastructure. Nvidia said it intends to invest up to $100 billion in OpenAI as each gigawatt is deployed, with the first phase targeted for the second half of 2026 on the Vera Rubin platform. (cnbc.com) Microsoft and Nvidia have also deepened their direct commercial ties. Microsoft said on March 18, 2025 that Azure was integrating Nvidia Blackwell systems and Nvidia NIM microservices into Azure AI Foundry, extending a collaboration that Microsoft said already underpins ChatGPT in Azure OpenAI Service and Microsoft Copilot. (openai.com) ### Where does the “$100 billion” figure come from? FourWeekMBA’s May 2026 analysis uses “$100 billion” as a framing device for the circular economy argument. The clearest public number near that scale is Nvidia’s stated plan to invest up to $100 billion in OpenAI infrastructure, announced jointly by the two companies in September 2025. (azure.microsoft.com) That number does not mean every part of the Microsoft-OpenAI-Nvidia relationship totals exactly $100 billion today. It is better read as a shorthand for the scale of overlapping commitments now tying together the chip supplier, the model company and the cloud distributor. That is an inference based on the companies’ disclosed agreements and public commentary. (fourweekmba.com) ### How does money move around this system? OpenAI’s January 2026 revenue update showed the company’s annualized revenue had passed $20 billion in 2025, up from $6 billion in 2024, while computing capacity rose to 1.9 gigawatts from 0.6 gigawatts. Chief Financial Officer Sarah Friar said growth tracked an expansion in computing capacity and said OpenAI was keeping a “light” balance sheet by partnering and keeping flexibility across providers and hardware types. (cnbc.com) Under the April 2026 Microsoft agreement, OpenAI still pays Microsoft a revenue share, while Microsoft no longer pays one to OpenAI. At the same time, Microsoft distributes OpenAI products through Azure, and Azure itself relies heavily on Nvidia hardware for AI infrastructure. ### Why are product teams paying attention to this? (finance.yahoo.com) Enterprise buyers have more reason to watch concentration risk when the same vendors appear repeatedly across chips, cloud and model access. OpenAI’s revised agreement with Microsoft loosened exclusivity by allowing OpenAI to serve products across other cloud providers, but Microsoft remains the primary cloud partner and retains a non-exclusive IP license through 2032. (cnbc.com) For engineering teams, that structure makes vendor-agnostic orchestration, neutral telemetry and portable evaluation systems more valuable if they want leverage across providers. That conclusion is an inference from the concentration of supply and distribution visible in the public agreements, rather than a claim made in a single company filing. (cnbc.com) ### What comes next that can be checked? The second half of 2026 is the next concrete milestone in the chain. OpenAI and Nvidia said the first gigawatt of Nvidia systems is targeted to come online then, and that deployment will provide the next public test of how much of this AI infrastructure buildout moves from announced commitments to operating capacity. (openai.com) (cnbc.com)