OpenAI is bringing its AI models and Codex coding tools to Google Cloud, according to Axios.
The move extends OpenAI's distribution to customers of Microsoft Azure, Amazon Web Services and now Google Cloud, the three leading US cloud providers.
It marks a sharp shift from an arrangement that once tied OpenAI almost exclusively to Microsoft's Azure.
As recently as February, OpenAI and Microsoft had reaffirmed that Azure would remain the exclusive home for OpenAI's stateless application programming interfaces (APIs), the direct channels enterprises use to access its models.
That exclusivity was subsequently loosened, with a revised agreement letting OpenAI sell its services through any cloud provider, though Azure still gets first access when Microsoft has the capacity to host new products.
AWS became the first beneficiary of that shift, gaining OpenAI models, Codex and agent capabilities through channels including Amazon Bedrock.
Codex itself has evolved from a coding assistant into a cloud-based agent capable of running tasks remotely and continuing work while a developer is offline.
The expansion comes as OpenAI's infrastructure commitments balloon: industry analysis puts its disclosed cloud capacity commitments at more than $500 billion, including roughly $250 billion with Azure and about $138 billion with AWS.
For Google and AWS, hosting OpenAI's models offers a way to satisfy enterprise demand for a widely used AI supplier without forcing customers to migrate data, identity and compliance systems to a new provider.
Microsoft, meanwhile, retains its position as OpenAI's primary infrastructure partner even as the exclusivity that once defined the relationship continues to erode.
Read more about Cloud Infrastructure
