AWS Will Add 2 Million More Nvidia GPUs in 2027 and 2028 as AI Demand Outruns Forecasts

AWS will deploy another 2 million Nvidia GPUs across its data centers in 2027 and 2028, including Blackwell Ultra, Rubin and Rubin Ultra chips, the companies announced Wednesday during Nvidia’s earnings call. It comes about five months after AWS said at GTC that it would add more than 1 million Nvidia GPUs starting this year, and Nvidia says demand has since outrun those expectations.

Neither company shared financial terms, though TechCrunch estimated that at typical GPU prices the deal is worth tens of billions of dollars. The companies cited “surging demand” from startups, enterprises, AI labs and governments.

“NVIDIA and AWS have built one of the great growth engines of the AI era, and demand is running ahead of every forecast,” said Nvidia CEO Jensen Huang. AWS CEO Matt Garman said customers “want the freedom to choose the best tools for their AI workloads.”

More than GPUs

The expanded partnership covers much more of Nvidia’s stack:

• Vera CPUs on AWS: Nvidia’s Vera CPU is coming to AWS as another option for agentic AI and reinforcement learning work such as code execution, tool use, sandboxing and data pipelines.
• Trainium ties: Amazon’s Annapurna Labs will work with Nvidia’s custom high-bandwidth memory technology, NVHBM, building on AWS’s earlier plan to support NVLink Fusion in future Trainium chips.
• Government AI factories: The companies will build AI factories for the U.S. government, including 100,000 GPUs on secure AWS infrastructure.
• Networking, open models, data processing and robotics: Nvidia technology in each of these areas will be integrated across AWS.

Alongside Amazon’s own chips

The deal lands while Amazon keeps investing in its own silicon. AWS builds Trainium AI chips and Graviton CPUs, and TechCrunch reported that AWS is in talks to sell Trainium to other companies for their data centers. AWS presents the two efforts as complementary, letting customers choose Nvidia GPUs, Trainium or both. For enterprise buyers, that means more GPU capacity on the way and more ways to mix hardware for different workloads.