AWS and Nvidia Partner to Deploy 2 Million Next-Gen GPUs for Global AI Infrastructure
Amazon Web Services (AWS) and Nvidia have announced a massive multi-year expansion of their infrastructure partnership, committing to deploy 2 million additional Nvidia GPUs across AWS availability zones through 2027 and 2028.
Subscribe to Tech Bytes Daily Briefing
Get top technology breakdowns, silicon engineering insights, and daily executive summaries delivered straight to your inbox.
No spam. Unsubscribe anytime.
Building upon an earlier commitment of 1 million GPUs, the new compute deployment will focus primarily on Nvidia's next-generation Blackwell and Ultra-architecture accelerators, integrated directly into custom AWS EC2 UltraClusters equipped with Elastic Fabric Adapter (EFA) networking.
The expanded compute capacity addresses persistent capacity shortages faced by enterprise AI developers and foundation model builders, enabling massive multi-node training clusters with petabit-per-second interconnect bandwidth.