Two of the biggest names in AI have spent years competing for the same enterprise budgets, and on Wednesday they announced they will share a shelf. OpenAI said its frontier models and Codex agentic workflows are coming to Google Cloud, where businesses will be able to buy them through Google Cloud’s Model Garden and pay with the money they already spend on Google Cloud.
“OpenAI models are coming to Google Cloud. We’re bringing our frontier models and Codex agentic workflows to Google Cloud’s Model Garden, so enterprises can use our technology where they already build,” the company said in an X post on October 21, according to ANI.
For enterprise IT teams, the headline is less about the technology and more about the plumbing: customers will be able to draw down their existing Google Cloud spending commitments to pay for OpenAI inference, and the models will also land inside Antigravity, Google’s agent-driven coding tool, and the Gemini Agent. General availability is expected in the coming weeks, with a private preview already rolling out to select enterprise customers, Ars Technica reports.
The models Google Cloud gets
According to Ars Technica, the full current GPT-6.1 lineup will be included: Astra, Sol, and Luna. Google Cloud’s Model Garden already offers Google’s own Gemini models plus Anthropic’s Claude, which means Google can now pitch itself as the only large cloud provider hosting all three leading closed-source model families in one place, alongside open models like Mistral, Llama, and Gemma.
“We are focused on giving customers a broad selection of models, including those from OpenAI,” Google Cloud CEO Thomas Kurian said, per Tech Wire Asia.
Neither company disclosed the financial terms of the agreement or pricing for OpenAI models on Google Cloud. That leaves the interesting questions open: whether OpenAI charges the same rates through GCP as through its own API, and what revenue share Google takes on Marketplace transactions.
Microsoft is still the primary partner, just less exclusively
OpenAI was careful to state that Microsoft remains its primary cloud partner. But the word “primary” does a lot of work now. In June 2026, OpenAI began letting customers run its models and Codex through their Oracle cloud commitments. The Google deal extends the same playbook to a second rival of Azure. Taken together, the two moves end the era when OpenAI’s distribution ran almost entirely through Microsoft.
The pattern is familiar from Anthropic’s enterprise push, which has also leaned on multi-cloud availability rather than exclusive deals. The model providers have figured out what the cloud providers knew decades ago: procurement beats pricing. An enterprise buyer will not renegotiate vendor contracts, redo security reviews, and migrate data just to try a different model. But it will happily toggle on a new model inside a platform it already pays for.
Why the billing detail is the real story
Google Cloud’s Marketplace transactions draw down existing cloud commitments and count toward Flexible Savings Plans across model providers, which removes a friction point that has quietly killed multi-model strategies. As one Google Cloud executive put it on LinkedIn, procurement teams no longer have to choose between burning their commitments and trying a new vendor.
This is also the shape of what is coming for smaller builders. Nvidia’s open-weight agent models and Europe’s push for sovereign AI infrastructure, like OVHcloud’s planned proprietary model, all point to a market where the unit of competition is not the model but the platform it runs on. The model is the product; the invoice is the moat.
The awkward part nobody mentioned on stage
The partnership has raised eyebrows in developer circles. Selling a direct Gemini competitor on Google Cloud sits oddly with Google’s own AI ambitions, and some developers questioned how OpenAI plans to stop Google from learning from the traffic that passes through its infrastructure. Neither company has addressed distillation concerns publicly, and Microsoft’s own position as the jilted exclusive partner has not been addressed either.
Still, the strategic logic is hard to argue with. OpenAI gets distribution to every enterprise that standardized on GCP. Google gets a cut of OpenAI’s inference revenue and a stronger enterprise story. And enterprise buyers get what they have been asking for since the first model war: the ability to compare models side by side without migrating anything. The AI industry’s most bitter rivalry just became its most pragmatic partnership.
