AWS and NVIDIA Expand Strategic AI Alliance: 2 Million Next-Gen GPUs and Sovereign AI Factories
The race to scale artificial intelligence infrastructure has reached a historic milestone. Amazon Web Services (AWS) and NVIDIA have unveiled a massive multi-year expansion of their 16-year strategic partnership. The agreement is engineered to address skyrocketing compute demands as enterprises transition from experimental generative AI pilots to full-scale agentic production systems.
Key Highlights of the Expansion
- Massive GPU Scale: AWS will deploy an additional 2 million NVIDIA GPUs across its global cloud data centers between 2027 and 2028.
- NVIDIA Vera CPU Integration: AWS is adding next-generation NVIDIA Vera CPUs to its compute lineup, providing optimized performance for agentic AI workflows and complex reasoning pipelines.
- Sovereign & Government AI Factories: The collaboration includes building dedicated, high-security AI factories for the U.S. public sector, featuring 100,000 GPUs running on isolated AWS infrastructure.
- Hybrid Compute Flexibility: Customers gain seamless interoperability between NVIDIA accelerators and Amazon's proprietary Trainium silicon, allowing organizations to tailor workloads for cost efficiency and raw throughput.
Why This Matters for Enterprise Cloud and AI
As enterprise IT strategies pivot toward autonomous agentic workflows and multi-gigawatt compute clusters, hyperscalers face immense pressure to deliver uninterrupted scalability, low latency, and robust data governance. AWS CEO Matt Garman noted that customer AI workloads have decisively moved into large-scale production, necessitating deep co-engineering across silicon, networking, and cloud software stacks.
By coupling custom in-house accelerators with NVIDIA's hardware and CUDA software ecosystem, AWS reinforces its position against rival hyperscalers while providing enterprises with the flexibility required for the next generation of generative AI models and distributed robotic agents.