Amazon menghabiskan 50 miliar dolar AS untuk bersekutu dengan OpenAI, AWS mendapatkan hak distribusi cloud eksklusif Frontier

Phoenix Net Technology News on February 27: Today, Amazon and OpenAI officially announced the completion of a multi-year strategic partnership. Under the agreement, Amazon will invest $50 billion in OpenAI, with an initial injection of $15 billion and the remaining $35 billion to be disbursed within the next few months after certain conditions are met. This move marks a deepening of ties between these two tech giants in the areas of generative AI infrastructure and enterprise-level applications.

In terms of core technology and cloud service distribution, Amazon Web Services (AWS) will become the exclusive third-party cloud distribution provider for OpenAI’s Frontier enterprise platform, enabling enterprises to build, deploy, and manage AI agent teams. At the same time, the two parties are working together to develop a stateful runtime environment based on OpenAI models. This environment will directly integrate with Amazon Bedrock services, and is expected to be officially launched in the coming months. It is designed to help developers preserve context across data sources and handle complex, ongoing workflows.

Compute infrastructure is another key pillar of this partnership. Both sides will significantly expand the existing $38 billion multi-year agreement by adding $100 billion, and will extend the cooperation term by 8 years. OpenAI commits to consuming approximately 2 gigawatts (GW) of Trainium custom AI chip compute power through AWS infrastructure to support its advanced workloads. This plan not only covers the current Trainium3 chips, but also includes the next-generation Trainium4 chips expected to be delivered in 2027, indicating that OpenAI is adopting Amazon’s in-house silicon chip ecosystem at a large scale.

In addition, the cooperation agreement also extends to downstream application capabilities. OpenAI will assist Amazon in developing customized models for use by Amazon’s internal developers in its directly consumer-facing applications and AI agent teams. These customization capabilities will complement Amazon’s own Nova series models via a “dual-line” approach, further expanding its technical reserves and service capabilities for its C-end products. (Author/Yu Lei)