Recently, OpenAI announced a long-term cooperation agreement worth $3.8 billion with Amazon AWS, marking a significant advancement for OpenAI in AI cloud infrastructure. Previously, OpenAI mainly relied on Microsoft Azure for cloud services, but as its demand for computing power continued to grow, OpenAI chose to partner with AWS.

Human-machine collaboration

Image source note: The image is AI-generated

According to the new agreement, OpenAI will use AWS's infrastructure, especially its Amazon EC2 UltraServers, over the next seven years. This will provide OpenAI with thousands of high-performance NVIDIA GB200 and GB300 GPUs. These powerful computing resources will support OpenAI in training and reasoning for larger-scale AI models. It is expected that by the end of 2026, OpenAI will exhaust these computing capabilities and continue expanding in 2027.

AWS CEO Matt Garman stated that AWS's infrastructure will become an important support for OpenAI in advancing AI technology. Garman emphasized that the wide and immediately available computing power provided by AWS will offer unique support for OpenAI's large-scale AI workloads. This collaboration will not only help applications like ChatGPT with inference but also support the training of next-generation models.

Before finalizing the new agreement, Microsoft adjusted the terms of its partnership with OpenAI, allowing OpenAI to host non-API products, such as ChatGPT and Sora, on any cloud service platform. This change provides OpenAI with more flexibility, enabling it to better meet its computing power needs.

Amazon also noted that OpenAI's gpt-oss series models have become one of the most popular public models on its Amazon Bedrock platform, attracting the use of multiple clients including Comscore, Peloton, and Thomson Reuters. This partnership indicates further deepening and development of OpenAI and AWS in the field of artificial intelligence.

Key points:  

🌟 OpenAI has signed a long-term cooperation agreement worth $3.8 billion with Amazon AWS, expanding AI cloud infrastructure.  

💻 AWS will provide thousands of high-performance GPUs to support OpenAI in large-scale model training and reasoning.  

🚀 This collaboration enables OpenAI to flexibly use non-API products and promote the development of next-generation AI models.