OpenAI Breaks Microsoft Exclusivity to Launch Models on Amazon Web Services

OpenAI has officially announced the availability of its generative AI models on Amazon Web Services (AWS), marking a pivotal shift in the company's infrastructure strategy. The integration allows AWS customers to access OpenAI's suite of models and the Codex coding agent through the Amazon Bedrock platform, with general availability expected in the coming weeks. This expansion comes immediately after OpenAI renegotiated its long-term relationship with Microsoft. While Microsoft has been the primary provider of compute power since before the launch of ChatGPT, the new arrangement caps revenue share payments and permits OpenAI to serve enterprise clients across various cloud providers, removing previous limitations on where the company could meet its corporate customers. The partnership is supported by massive financial and technical commitments. OpenAI previously announced a $38 billion commitment with AWS and received a $50 billion investment from Amazon. To support its scaling needs, OpenAI will utilize two gigawatts of AWS's proprietary Trainium chips for the training of its AI models. Despite the strategic expansion, the announcement arrives amid a complex backdrop. Recent reports suggesting OpenAI missed internal revenue and user targets caused temporary volatility for AI hardware suppliers, including Nvidia and Broadcom. OpenAI leadership has denied these reports, asserting that the company remains fully aligned on its compute acquisition and growth strategies.