Project Rainier is operational, featuring nearly half a million Trainium2 chips, making it one of the largest AI compute clusters globally. AWS launched this infrastructure less than a year after its announcement, collaborating with Anthropic, which is now using it to power its AI model, Claude. By the end of 2025, Claude is projected to run on over 1 million Trainium2 chips. The compute power available through Rainier is more than five times that of Anthropic's previous models, allowing for advanced training and deployment of AI technologies. The design of Project Rainier includes a unique “EC2 UltraCluster of Trainium2 UltraServers.” Each UltraServer integrates four physical servers, each containing 16 Trainium2 chips, connected by high-speed NeuronLinks. This setup minimizes latency and maximizes communication efficiency between chips. AWS's infrastructure is 70% larger than any previous AI platform it has created, with each Trainium2 chip capable of trillions of calculations per second. This level of processing power enables rapid training of complex AI models. AWS emphasizes reliability in operating such a vast compute cluster. By building its own hardware and maintaining oversight of the entire technology stack, AWS can optimize performance and troubleshoot issues effectively. Sustainability is also a priority; the company matched 100% of its electricity use with renewable resources in 2023 and plans to be net-zero carbon by 2040. New data center components are expected to reduce energy consumption significantly, while innovative cooling techniques will minimize water usage. For instance, data centers in Indiana will rely heavily on outside air for cooling for most of the year, showcasing AWS's commitment to environmental efficiency.