← Back
OpenAI
OpenAI and Amazon partnership brings stateful runtime to AWS, $50B investment announced
OpenAI APIOpenAI · releasefeatureapiplatformintegrationpartnership · openai.com ↗

Partnership Overview

OpenAI and Amazon announced a comprehensive multi-year strategic partnership designed to accelerate enterprise AI adoption. Amazon will invest $50 billion in OpenAI, with an initial $15 billion investment followed by $35 billion in subsequent tranches pending certain conditions. The partnership expands their existing $38 billion computing agreement with a new $100 billion, 8-year commitment.

Stateful Runtime Environment for Amazon Bedrock

OpenAI and Amazon are jointly developing a Stateful Runtime Environment powered by OpenAI models, available through Amazon Bedrock. This represents a next-generation approach to frontier model deployment, enabling:

  • Persistent context and state across sessions
  • Seamless access to compute, memory, and identity management
  • Unified tool integration across software applications and data sources
  • Optimized performance on AWS infrastructure
  • Integration with Amazon Bedrock AgentCore services

The Stateful Runtime is expected to launch within the next few months.

AWS as Exclusive Frontier Distribution Provider

AWS will serve as the exclusive third-party cloud provider for OpenAI Frontier, OpenAI's advanced enterprise platform for building and managing teams of AI agents. Frontier enables organizations to deploy agents across business systems with shared context, built-in governance, and enterprise-grade security without managing underlying infrastructure.

Infrastructure Commitment

OpenAI will consume approximately 2 gigawatts of Trainium capacity through AWS infrastructure to support the Stateful Runtime, Frontier, and other advanced workloads. This commitment includes both current Trainium3 chips and next-generation Trainium4 processors (expected delivery in 2027), which will offer improved FP4 compute performance, expanded memory bandwidth, and increased high-bandwidth memory capacity.

Custom Models for Amazon Applications

OpenAI and Amazon will develop customized models available to Amazon developers for building customer-facing applications. These tailored models complement existing options like Amazon's Nova family, giving Amazon teams additional flexibility for AI product development and deployment at scale.