Amazon has completed its $50 billion investment in OpenAI, revealing in a July 31 U.S. Securities and Exchange Commission filing that it wired the remaining $35 billion of its February commitment to the ChatGPT maker in the second and third quarters. The disclosure closes one of the largest single equity checks written in the artificial intelligence race and cements Amazon Web Services as OpenAI’s exclusive third-party cloud provider for its Frontier program.
What Amazon disclosed
According to the filing, Amazon put in $15 billion during the first quarter after signing the deal on February 27, added another $13.7 billion in the second quarter, and delivered the final $21.3 billion after June 30. Amazon did not specify why it made the remaining investment, but the original agreement made the second tranche contingent on OpenAI hitting undisclosed milestones — reportedly tied to an eventual initial public offering or progress toward artificial general intelligence. The AI lab has not yet gone public.
AWS locked in as OpenAI’s Frontier cloud
The February agreement pairs the equity investment with an expanded infrastructure deal that could reach $100 billion over eight years, making AWS the exclusive third-party cloud for OpenAI’s Frontier program. That commitment slots alongside the model developer’s Stargate build-out — including its multi-gigawatt Project Camellia campus in Georgia — and its infrastructure agreements with Oracle and Microsoft, giving Sam Altman a diversified capacity backbone for the next wave of model training.
Why it matters for physical AI and robotics
Amazon has spent 2026 sharpening its focus on infrastructure and enterprise AI, even as it watches rivals such as Anthropic race ahead in agent safety. The completed OpenAI stake gives AWS the flagship reference workload it needs to compete with Microsoft Azure for hosting frontier models — the same GPU capacity that increasingly powers embodied AI, humanoid robots and autonomous systems. ChatGPT is approaching one billion weekly active users, the fastest scale-up in consumer software history, and OpenAI’s pricing cuts to its GPT-5.6 models late last week underscore how the economics of model training now hinge on hyperscale infrastructure deals like this one.
Reporting based on coverage from PYMNTS, The Information and Amazon’s July 31 SEC filing.