A historic partnership has been officially announced: Amazon invests $50 billion in OpenAI to jointly build a "stateful" AI future.

At a crucial turning point in artificial intelligence's transition from experimental exploration to large-scale production applications, OpenAI and Amazon officially announced a landmark multi-year strategic partnership. This deep collaboration, totaling $50 billion, not only signifies Amazon's status as a significant strategic investor in OpenAI but also signals a commitment to jointly reshaping the infrastructure landscape for enterprise AI. By integrating OpenAI's cutting-edge model capabilities with Amazon Web Services' (AWS) world-leading computing infrastructure, the two companies are dedicated to launching a revolutionary "stateful runtime environment" and the OpenAI Frontier platform. This aims to break through existing bottlenecks in AI development, providing global enterprises, startups, and developers with the powerful engine needed to build production-grade agents, thereby accelerating the deployment and innovation of AI technology in real-world business scenarios.

OpenAI and Amazon announced a multi-year strategic partnership aimed at accelerating AI innovation for global enterprises, startups, and end consumers. Amazon will also invest $50 billion in OpenAI, with an initial investment of $15 billion and a further $35 billion to be invested in the coming months subject to certain conditions.

Amazon Web Services and OpenAI will jointly create a "stateful runtime environment" driven by OpenAI models. This environment will be provided to Amazon Web Services (AWS) customers via Amazon Bedrock for building production-scale generative AI applications and agents.

AWS will be the exclusive third-party cloud distribution provider for OpenAI Frontier, enabling organizations to build, deploy, and manage AI agent teams.

OpenAI will utilize approximately 2 gigawatts of Amazon Trainium computing power through AWS infrastructure to meet the needs of the stateful runtime environment, OpenAI Frontier, and other advanced workloads.

OpenAI and Amazon will jointly develop customized models to power Amazon's various consumer-facing applications.

Amazon will invest $50 billion in OpenAI.

Working together to create new advanced AI capabilities for enterprises worldwide

OpenAI and Amazon are jointly developing a stateful runtime environment powered by OpenAI models, which will be provided via Amazon Bedrock.

Stateful development environments represent the next generation of evolution in how models are used, enabling models to seamlessly access elements such as compute, memory, and identity. Stateful runtime environments allow developers to retain context, remember previous work, work across software tools and data sources, and access compute resources. These environments are designed specifically for handling ongoing projects and workflows.

These stateful development environments will be trained to run optimally on Amazon Web Services (AWS) infrastructure and integrate with Amazon Bedrock AgentCore and infrastructure services, enabling customers' AI applications and agents to work seamlessly with the rest of the infrastructure applications running on AWS. Stateful runtime environments are expected to roll out in the coming months.

Introducing OpenAI's Advanced Enterprise-Grade Platform to AWS Customers

Amazon Web Services will serve as the exclusive third-party cloud distribution provider for OpenAI Frontier, expanding access to OpenAI's advanced enterprise-grade platform as the demand for AI deployments accelerates across industries.

Frontier enables organizations to build, deploy, and manage teams of AI agents running on real business systems with shared context, built-in governance, and enterprise-grade security, without the need to manage the underlying infrastructure. As companies move from AI experimentation to production applications, Frontier makes it simple and straightforward to integrate powerful AI quickly, securely, and globally at scale into existing workflows.

OpenAI to Utilize Amazon Trainium Computing Power to Meet Growing Amazon Customer Demand

OpenAI and Amazon Web Services (AWS) will expand their existing $38 billion multi-year partnership by $100 billion over the next eight years. This expansion includes OpenAI's commitment to utilize approximately 2 gigawatts of Amazon Trainium computing power through AWS infrastructure to meet the needs of stateful runtime environments, OpenAI Frontier, and other advanced workloads. This agreement reduces the cost and improves efficiency of mass-producing intelligence.

Within this framework, OpenAI secures long-term computing power while collaborating with AWS to deploy dedicated chips and its broader computing architecture, enabling enterprises to access intelligent computing power on demand without managing the underlying infrastructure.

This commitment also includes the Amazon Trainium 3 and the next-generation Amazon Trainium 4 chips, which will power a wide range of advanced AI workloads. The Amazon Trainium 4, expected to begin shipping in 2027, will bring another significant performance boost, including significantly improved FP4 computing performance, expanded memory bandwidth, and larger high-bandwidth memory capacity, supporting the large-scale deployment of increasingly powerful AI systems.

Customized Models Power Amazon's Consumer-Focused Applications

OpenAI and Amazon will collaborate to develop custom models for Amazon developers to enhance the experience of Amazon's consumer-facing applications. Amazon teams will be able to customize OpenAI models for use in AI products and agents that directly serve consumers. These capabilities will complement the various models currently used by Amazon developers, including the Amazon Nova family, providing another powerful tool for teams to build and deliver at scale.

Sam Altman, co-founder and CEO of OpenAI, stated, "OpenAI shares the same philosophy as Amazon: AI should be implemented in a practical way, truly creating value for humanity. Combining OpenAI's models with Amazon's infrastructure and global reach helps us deliver powerful AI at scale to businesses and users."

Andy Jassy, President and CEO of Amazon, said, "We have numerous developers and enterprises looking to run services powered by OpenAI models on Amazon Web Services, and the 'stateful runtime environment' we've created with OpenAI will open up more possibilities for customers building AI applications and agents. We have always appreciated OpenAI's R&D achievements, are delighted by its large-scale adoption of our custom AI chip, Amazon Trainium, and are excited about our long-term investment in OpenAI and deepening our collaboration."