Amazon AWS announced in a joint press release with NVIDIA that it plans to deploy an additional 2 million NVIDIA GPUs across its global infrastructure during 2027~2028.
AWS had previously committed to deploying 1 million NVIDIA GPUs at GTC 2026 in March this year. The new announcement brings the total GPU deployment target to 3 million. AWS will also adopt NVIDIA Groq LPU.
AWS will also provide NVIDIA Vera CPU-based infrastructure through its own cloud services and integrate its proprietary chips with NVIDIA NVLink Fusion and NVHBM. The two parties also collaborate in artificial intelligence applications for factories, network interconnection, models, and the physical world.