Editor’s note: This blog was originally published on May 18, 2026, and updated on Thursday, August 27, 2026.
AWS and NVIDIA recently announced a major expansion of their collaboration, including plans to deploy 2 million additional NVIDIA GPUs and next-generation infrastructure for agentic and physical AI. The partnership will help organizations migrate and scale their AI infrastructure on AWS.
AWS has received its first NVIDIA Vera CPU servers and Vera Rubin GPUs. NVIDIA Vice President of Hyperscale and HPC Ian Buck presented the systems to AWS leaders at the company’s headquarters in Seattle.
AWS is the latest technology leader to receive NVIDIA Vera systems as the platform expands across the AI ecosystem. Previously, Buck presented Vera CPU systems to Oracle Cloud Infrastructure and three leading AI organizations: Anthropic, OpenAI and SpaceXAI.
“Agentic AI is creating new CPU moments in the AI factory, and Vera is purpose-built to keep that work moving at scale as models move from response to action,” said Buck.
Why NVIDIA Vera CPU, and why now?
AI workloads are becoming increasingly complex and demanding. From creating presentations and compiling software to analyzing data, searching files and running simulations, users need infrastructure that can keep pace with real-time, AI-powered work.
Agentic AI systems do not rely on GPUs alone. Every agent sandbox, tool call, orchestration layer and long-context retrieval operation requires CPU resources. NVIDIA Vera is a new class of CPU designed from the ground up to support these demanding agentic AI workloads.
The concurrent, real-time tasks generated by AI agents place new demands on data center CPUs—demands that traditional core-density-focused designs were not built to prioritize. NVIDIA Vera features 88 custom NVIDIA-designed Olympus cores and delivers 1.2 TB/s of memory bandwidth, enabling up to 1.8x faster per-core performance for agentic AI workloads.
By completing work faster under sustained workloads, NVIDIA Vera can improve the efficiency of the entire AI factory. This helps cloud providers and enterprises deliver faster responses, process more tokens per dollar and scale AI agents more effectively.
NVIDIA Vera is coming to @awscloud.
NVIDIA’s Ian Buck handed over AWS’s first Vera CPU server and Vera Rubin GPU to Willem Visser and Supreeth Sheshadri at AWS headquarters in Seattle.
Vera is purpose-built for agentic AI, delivering more tokens per dollar and faster results for users. pic.twitter.com/btcw2gz42x
— NVIDIA (@nvidia) August 27, 2026
Source: blogs.nvidia.com