OpenAI and AWS Finalize $38B Seven-Year Deal for Frontier Compute

• Historic $38 Billion Cloud Agreement:

  On November 3, 2025, OpenAI and AWS announced a $38 billion, multi-year partnership granting OpenAI access to AWS compute for training and deploying frontier models. OpenAI will run workloads on EC2 UltraServers with hundreds of thousands of Nvidia GB200/GB300 GPUs, with capacity to scale to tens of millions of CPUs. All targeted capacity is slated to be online by end-2026, with expansion room into 2027+.

• Massive Compute Scaling and Performance:

  AWS will provide low-latency GPU clustering across UltraServers, optimized for large-scale training and inference. AWS operates clusters exceeding 500,000 chips, enabling efficient ChatGPT inference and next-gen model training, and positioning AWS as a backbone for OpenAI’s expanding agentic AI workloads.

• Shift from Microsoft Exclusivity (Contract Update):

  From 2019–2023, OpenAI sourced compute exclusively from Microsoft under a contract requiring Microsoft approval for other cloud purchases. Over the past 18 months, Microsoft allowed separate deals with other clouds, and last week (late Oct 2025) Microsoft and OpenAI renegotiated terms, removing the approval requirement and formally freeing OpenAI to buy from any provider.

• Amazon’s Strategic Gain and Market Impact:

  Amazon stock rose 5% following the announcement. The OpenAI and AWS collaboration reinforces Amazon’s leadership in AI infrastructure despite its investments in rival Anthropic and its $11 billion Indiana data center project. By hosting OpenAI workloads, AWS demonstrates the ability to serve competing AI developers and dominate the global cloud race against Microsoft and Google.

• Multi-Cloud Diversification:

  OpenAI also committed an additional $250B to Microsoft Azure services. The strategy is explicitly multi-cloud, diversifying risk and securing access to broad compute ecosystems amid global GPU shortages.

• $1.4 Trillion AI Infrastructure Expansion:

  The OpenAI and AWS deal sits within OpenAI’s broader ~$1.4T build-out plan. Sam Altman has outlined goals for ~30 GW of compute capacity globally, spanning partnerships with Nvidia, Broadcom, Oracle, Google, and others. OpenAI is also pursuing new data-center projects with Oracle, SoftBank, the U.A.E., and partners beyond the U.S., while Big Tech peers Amazon, Google, Meta, and Microsoft have collectively committed >$360B in capex over the last year.

• Executive Statements:

  Sam Altman:

  “Scaling frontier AI requires massive, reliable compute,” emphasizing that “our partnership with AWS strengthens the broad compute ecosystem that will power this next era.”

  Matt Garman (AWS CEO):

  “The breadth and immediate availability of optimized compute demonstrates why AWS is uniquely positioned to support OpenAI’s vast AI workloads.”

• Corporate and Technical Integration:

  OpenAI’s foundation and open-weight models are already integrated into Amazon Bedrock, serving enterprise clients such as Peloton, Thomson Reuters, Comscore, Triomics, and Verana Health. The new agreement formalizes OpenAI as a direct AWS customer, enhancing accessibility, scalability, and collaboration between OpenAI and AWS technologies.

• Architectural Design and Efficiency:

  AWS’s interconnected GPU clusters are designed for maximum throughput with low-latency links, supporting both inference (e.g., ChatGPT) and training. With all targeted capacity deployed by end-2026, this ranks among the largest single AI infrastructure commitments to date.

• IPO Preparations and Market Context:

  The deal arrives as OpenAI finalizes its restructuring toward an IPO potentially valuing the company near $1 trillion. Despite rapid revenue growth, the company remains unprofitable, fueling discussions of a potential AI investment bubble. Analysts note that OpenAI and AWS’s massive cloud expansion underscores both firms’ long-term bet on AI infrastructure as a core economic driver.

Why This Matters:

The OpenAI and AWS partnership marks a transformative moment in the AI industry. It secures critical compute resources for OpenAI, diversifies its cloud dependencies, and redefines the competitive balance among hyperscalers. For Amazon, it strengthens AWS’s leadership in enterprise-grade AI infrastructure and reinforces its role in hosting global-scale AI operations that will define the next decade of digital innovation.

This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication.