Google backs open-weight AI letter alongside OpenAI and xAI
Google, OpenAI, and xAI back an open letter urging US policymakers not to restrict open-weight AI models, joining 25 signatories including Nvidia
TL;DR
OpenAI announced on July 25, 2026 that its models and Codex will be sold directly to Google Cloud customers, joining AWS and Azure. The move broadens AI access across all major US clouds, reduces Microsoft’s exclusive edge, and reflects OpenAI’s $10 billion revenue scale and hardware diversification.
Key facts
- OpenAI announced on July 25, 2026 that its models and Codex will be available on Google Cloud.
- The rollout adds Google Cloud to the existing availability on Amazon Bedrock (April 28, 2026) and Microsoft Azure.
- OpenAI finalized a deal to use Google Cloud infrastructure in May 2025.
- Google Cloud officially joined OpenAI’s supplier list in July 2025 alongside Microsoft, Oracle and CoreWeave.
- OpenAI rents Google’s TPUs, indicating a shift from Nvidia GPUs.
- OpenAI reported an annualized revenue run rate of $10 billion as of June 2025.
On July 25, 2026 OpenAI announced that its flagship AI models and the Codex coding assistant will be offered directly to Google Cloud customers. The move places OpenAI’s tools on all three dominant U.S. cloud platforms—Google Cloud, Amazon Web Services and Microsoft Azure—allowing developers to access the technology without a separate integration. The rollout follows OpenAI’s earlier launch on AWS Bedrock in April 2026 and builds on a supplier relationship that began in May 2025 when OpenAI started renting Google’s TPU chips. OpenAI reported an annualized revenue run rate of $10 billion as of June 2025, underscoring the scale of its compute needs.
Background
OpenAI’s partnership history has been shaped by its need for massive compute capacity. For years the company relied heavily on an exclusive relationship with Microsoft, which hosted its models on Azure and served as the primary cloud partner. As usage surged, OpenAI sought additional capacity, leading to a diversification strategy. In May 2025 OpenAI finalized a deal to use Google Cloud infrastructure, and by July 2025 Google Cloud joined a roster of suppliers that already included Microsoft, Oracle and CoreWeave. The partnership extended beyond mere tenancy; OpenAI began renting Google’s custom Tensor Processing Units (TPUs), marking a notable shift away from the Nvidia GPUs that have powered much of the AI boom. This infrastructure diversification set the stage for the later product rollout, enabling OpenAI to offer its frontier models and Codex directly through Google’s platform. The earlier expansion onto Amazon’s managed AI service, Bedrock, on April 28, 2026, completed the coverage of the three major U.S. cloud providers, giving enterprise customers a choice of cloud environments for OpenAI’s services.
What happened
The July 25, 2026 announcement detailed that Google Cloud customers will now have direct access to OpenAI’s frontier models and the Codex coding tool, eliminating the need for a separate integration path. CryptoBriefing reported that this rollout mirrors OpenAI’s earlier deployment on Amazon Bedrock, which became available on April 28, 2026. The article explained that OpenAI’s models were already accessible on Microsoft Azure, a long‑standing home base for the company. The shift from a purely customer‑side relationship—where OpenAI purchased compute from Google—to a supplier relationship means Google’s own users can now purchase OpenAI’s AI products. The background to this change includes OpenAI’s May 2025 agreement to use Google Cloud infrastructure and the July 2025 addition of Google Cloud to its supplier list alongside Microsoft, Oracle and CoreWeave. OpenAI’s adoption of Google’s TPUs, noted in the source, illustrates its willingness to look beyond Nvidia hardware to meet capacity and cost demands. The announcement also highlighted that OpenAI reported an annualized revenue run rate of $10 billion as of June 2025, emphasizing the financial scale driving its compute diversification.
Market impact
OpenAI’s expansion onto Google Cloud broadens the availability of its AI models across the three dominant U.S. cloud platforms, potentially increasing demand for cloud compute services on Google’s infrastructure. The move may shift some enterprise workloads that previously relied on Azure or AWS to Google Cloud, influencing competitive dynamics among the providers. OpenAI’s reported $10 billion annualized revenue run rate, cited in the source, underscores the sizable market for AI services and suggests that the company’s compute needs will continue to drive significant spending on cloud resources. While the source does not provide specific price changes or transaction volumes, the broader access to OpenAI’s tools could stimulate higher adoption rates among developers, potentially translating into increased cloud usage fees for Google, Amazon and Microsoft alike. The diversification of hardware, with OpenAI renting Google’s TPUs, may also affect the market for AI accelerators, introducing competitive pressure on Nvidia’s dominance.
Why it matters
The integration of OpenAI’s models and Codex into Google Cloud completes a tri‑cloud coverage that gives enterprise customers the flexibility to choose their preferred platform without sacrificing access to leading AI capabilities. This reduces Microsoft’s exclusive advantage on Azure, potentially narrowing the competitive edge that Azure has enjoyed through its early partnership with OpenAI. By renting Google’s TPUs, OpenAI signals that large AI developers are willing to diversify hardware sources, which could encourage broader adoption of alternative accelerator technologies and affect pricing dynamics in the AI chip market. The $10 billion revenue figure indicates that OpenAI’s services are a major driver of cloud spend, meaning that any shift in provider relationships can have material financial implications for the cloud giants. Moreover, the expanded availability may accelerate AI integration across industries, influencing downstream markets such as fintech, biotech and software development, where OpenAI’s tools are increasingly embedded.
DeFi Intel’s take
Editorial analysis, not a statement of fact.
DeFi Intel's take: the rollout suggests that OpenAI is cementing its role as a platform‑agnostic AI provider, leveraging competition among the three U.S. cloud giants to secure the compute capacity it needs. We read this as a strategic move to dilute Microsoft’s early‑partner advantage while tapping into Google’s TPU ecosystem, which could reshape hardware supplier dynamics. The $10 billion revenue run rate underscores the financial muscle behind OpenAI’s expansion, hinting that cloud providers will vie for a larger slice of AI‑related spend. This development may also spur broader AI adoption in enterprise settings, potentially increasing demand for decentralized finance solutions that integrate advanced AI capabilities.
Sources used for this analysis
DeFi Intel wrote this analysis from the reporting above. Any figure or quotation can be checked against those pages.
Frequently asked questions
Which cloud platforms now offer OpenAI’s models and Codex?
OpenAI’s models and the Codex coding tool are now available on Google Cloud, Amazon Web Services through Bedrock, and Microsoft Azure. This gives developers access to OpenAI’s technology across all three major U.S. cloud providers.
When did OpenAI first start using Google Cloud infrastructure?
OpenAI finalized a deal to use Google Cloud infrastructure in May 2025, and by July 2025 Google Cloud officially joined OpenAI’s list of suppliers alongside Microsoft, Oracle and CoreWeave.
What hardware does OpenAI use from Google?
OpenAI rents Google’s Tensor Processing Units (TPUs), which are custom chips designed for AI workloads. This marks a shift from the Nvidia GPUs that have traditionally powered much of the AI boom.
Direct answers
Short answers, written to be quoted.
- Can I now use OpenAI’s Codex on Google Cloud?
- Yes, OpenAI’s Codex coding tool is now directly available to Google Cloud customers, allowing developers to integrate it without a separate integration path.
- Which cloud services host OpenAI’s frontier models?
- OpenAI’s frontier models are hosted on Google Cloud, Amazon Web Services through Bedrock, and Microsoft Azure, covering all three major U.S. cloud platforms.
- What hardware does OpenAI use from Google?
- OpenAI rents Google’s custom Tensor Processing Units (TPUs) for AI workloads, marking a diversification from its previous reliance on Nvidia GPUs.
DeFi Intel is an entity-graph aggregator: we curate, tag and link crypto news to a typed knowledge graph of protocols, tokens, people and incidents. We do not republish the full article body. Use the link above to read the original report at Crypto Briefing.
Entities in this story
Discussion
Comments are reviewed before publication.
Want the full article?
Continue reading on Crypto Briefing →
No comments yet.