OpenAI will sell its frontier models and its Codex coding agent through Google Cloud in the coming weeks, the companies said on Wednesday, in a deal reported by Reuters on October 21. The arrangement brings models such as GPT-6 Astra to Google Cloud’s Model Garden and lets customers buy them through Google Cloud Marketplace, paying with their existing cloud spending commitments.
Google becomes the latest hyperscaler to distribute OpenAI’s models, after OpenAI’s exclusive distribution arrangement with Microsoft expired in April and OpenAI expanded distribution to Amazon Web Services. The two companies compete directly in AI products and search, even as the economics of building and deploying large models push them toward cooperation.
What Google Cloud Customers Get
The partnership puts OpenAI’s latest models inside Google Cloud’s Model Garden catalogue, purchasable through the Marketplace against commitments customers have already made — a meaningful detail for enterprises, because it turns a separate AI contract into an extension of an existing cloud bill. OpenAI is using the channel to sell its models to more enterprise customers, the companies said.
The deal also targets regulated industries specifically. Customers in those sectors will be able to access OpenAI models through Google Cloud’s access controls, data governance tools and infrastructure, and use them to build agentic AI workflows — the same enterprise push visible in moves such as OpenAI’s acquisition of Orchestra Bio for drug-development agents.
The End Of Single-Cloud Exclusivity
For a decade of cloud computing, platform exclusives were the prize: Microsoft’s early investment in OpenAI came with exclusive distribution that made Azure the only cloud selling OpenAI models. That arrangement expired in April, and OpenAI has since moved onto AWS and now Google Cloud — each a major hyperscaler competing for AI market share, and each now taking the cloud provider’s usual cut from sales through its platform.
Google Cloud chief executive Thomas Kurian has told Reuters in a previous interview that the platform is an open platform and welcomes any model popular with developers. Google had already struck a cloud deal with OpenAI last year to supply computing capacity; selling the models themselves is the next layer. Reuters also noted, from Anthropic’s IPO filing, that 47 per cent of Anthropic’s 2025 revenue came through Amazon’s and Google’s cloud platforms — evidence of how central hyperscaler distribution has become to frontier labs’ income, a dynamic also visible in Microsoft’s expanding sovereign cloud AI offers.
Why It Matters
OpenAI’s arrival on Google Cloud matters because it completes the shift of frontier AI distribution from exclusive partnerships to open cloud marketplaces. When the same GPT-6 Astra can be bought through Azure, AWS or Google Cloud against existing commitments, competition moves from who stocks the model to price, governance tooling and the surrounding platform — a contest that favours customers and squeezes the margin any single cloud can take. For OpenAI, multi-cloud reach is now its enterprise sales strategy, extending even to products like ChatGPT’s new locked chats on the consumer side. For Google, stocking its fiercest product rival’s models is a deliberate bet that being the neutral shelf matters more than protecting Gemini’s exclusivity.
Background
OpenAI and Microsoft’s partnership, built on Microsoft’s multibillion-dollar investments from 2019 onward, originally gave Azure exclusive rights to sell OpenAI’s models — a structure that shaped the first years of the commercial AI race. As OpenAI’s compute needs outgrew any single provider, the relationship loosened: OpenAI signed cloud capacity deals with Google and others in 2025, the Microsoft exclusivity lapsed in April 2026, and distribution through AWS followed. Wednesday’s Google Cloud deal makes OpenAI’s models available on all three major hyperscalers for the first time.
Source: Reuters via Channel NewsAsia
0 Comments