AWS to offer OpenAI models following Microsoft's end of exclusivity, as OpenAI falls short of revenue goals and confronts $100 billion in infrastructure obligations.

      TL;DRAWS will begin selling OpenAI models following the end of Microsoft’s exclusive reselling rights, marking the conclusion of a restructuring that started with Amazon’s $50 billion investment in OpenAI’s $110 billion funding round. The companies have collaborated to develop a Stateful Runtime Environment for agentic AI within Bedrock. However, this announcement comes amid reports from the WSJ that OpenAI has missed revenue and user goals, with $25 billion in expected cash burn against anticipated revenue of $30 billion, alongside substantial infrastructure commitments to AWS, Azure, and Oracle that rely on growth OpenAI has yet to show.

      Amazon Web Services announced on Tuesday that it will start offering OpenAI’s models to its cloud customers, a day after Microsoft decided to terminate the exclusive reselling agreement that granted Azure sole access to OpenAI’s technology for the initial three years of the generative AI era. "This is something that our customers have long requested," stated Matt Garman, CEO of AWS, during an interview with Bloomberg Television. Select OpenAI models will be available for preview on AWS starting Tuesday, with the most advanced GPT models expected to launch within weeks. This announcement finalizes a restructuring of the AI industry's pivotal partnership that began in February when Amazon pledged up to $50 billion as part of OpenAI’s funding round, which valued the ChatGPT developer at $852 billion and marked Amazon's largest investment ever in a single company. OpenAI simultaneously committed to $100 billion in AWS computing resources and Trainium chips over eight years, utilizing two gigawatts of power capacity. The crucial question posed by the deal is not about the quality of OpenAI’s models on competing clouds, but whether OpenAI can sell enough of them across any platform to validate its spending promises.

      The restructuring process also saw Microsoft and OpenAI modify their partnership agreement on Monday, which dismantled the exclusive framework that characterized the first phase of the AI boom. Microsoft’s exclusive license to OpenAI’s intellectual property for models and products has now turned into a non-exclusive license that extends until 2032. Microsoft will maintain its 20 percent share of OpenAI’s revenue until 2030, though this is now subject to a cap. OpenAI is still required to launch its models on Azure first, unless Microsoft cannot or opts not to support the necessary capabilities, ensuring Azure retains its early-access benefit while eliminating the exclusivity that previously hindered AWS and Google Cloud customers from accessing OpenAI’s technology through their current cloud providers. The restructuring has also removed an AGI clause, a unique provision that would have ended Microsoft’s commercial rights if OpenAI’s board concluded it had achieved artificial general intelligence. This clause had transformed into a governance concern rather than a practical safeguard, and its removal indicates that both firms acknowledge their relationship is commercial rather than existential.

      Practically, this means OpenAI can now cater to customers across any cloud platform. For AWS, which spent the initial three years of the AI boom curating a collection of leading models for its Bedrock marketplace—including Anthropic’s Claude, Meta’s Llama, and Mistral—the inclusion of OpenAI mitigates Azure’s previous major competitive edge. Amazon's enhanced investment of up to $25 billion in Anthropic, announced alongside the OpenAI deal, means AWS now has exclusive or preferred access to the two most commercially relevant AI model families in the enterprise sector. Superhuman, previously known as Grammarly, initially sought AI services from Microsoft exclusively due to Azure being the only cloud hosting OpenAI’s models. That has now changed. AWS customers can now utilize OpenAI’s offerings on their existing infrastructure, complete with pre-established security configurations and compliance certifications.

      The two companies have collaboratively created a Stateful Runtime Environment, powered by OpenAI’s models, now available through Amazon Bedrock. This system provides persistent orchestration and memory for AI agents managing intricate, multi-step workflows. Rather than connecting various API requests, the runtime preserves context across tasks, retaining memory, tool states, environmental configurations, and identity boundaries. Amazon unveiled the product as Amazon Bedrock Managed Agents during an event in San Francisco, where AWS also detailed a broad initiative into business applications. "OpenAI’s business customers desire these models within a familiar and trustworthy environment," commented Denise Dresser, OpenAI’s chief revenue officer. The Stateful Runtime Environment is specifically trained to operate on AWS infrastructure and is integrated with Amazon Bedrock AgentCore, indicating that the agent capabilities are not merely an add-on to existing systems but a collaboratively developed layer designed for seamless operational integration of OpenAI’s models within AWS.

      This joint product symbolizes a significant shift in the way AI models are marketed. The initial phase of the AI boom focused on model access—identifying which cloud offered the superior models. The current phase emphasizes model integration, determining which cloud makes these models most effective within established enterprise workflows. Microsoft has started releasing its own internal AI models via Mustafa Suleyman’s superintelligence team, suggesting the company is preparing for a future where Open