AWS and NVIDIA Supercharge AI with 2 Million More GPUs

Amazon Web Services (AWS) and NVIDIA are dramatically expanding their partnership, committing to deploy an additional 2 million NVIDIA GPUs across AWS's global infrastructure by 2028. This massive hardware influx signals a new phase in the race to build ever more powerful artificial intelligence capabilities.

The collaboration, building on 16 years of joint innovation, extends beyond just GPUs. The two tech giants are deepening their work across AI factories, CPUs, networking, and robotics, aiming to deliver co-engineered solutions that can accelerate AI development at an unprecedented scale. This move complements Amazon’s existing custom silicon efforts, offering customers a wider choice for their AI workloads, whether it's NVIDIA's specialized hardware, AWS's Trainium chips, or a combination of both.

The sheer scale of this GPU deployment is staggering. AWS plans to add over 1 million NVIDIA GPUs starting in 2026, and the demand has already outstripped expectations. The commitment for 2027, 2028 brings the total to an additional 2 million GPUs, which will power everything from complex agentic AI and scientific discovery to enterprise automation and physical AI applications.

Beyond GPUs: A Full-Stack Approach

The partnership isn't just about raw graphics processing power. AWS is bringing NVIDIA Vera CPUs to its platform, offering a new compute option specifically designed for agentic AI workloads. These CPUs are engineered to handle the CPU-intensive tasks that support AI agents, such as code execution, tool use, and orchestration, ensuring that GPUs are efficiently fed and AI agents remain responsive.

Furthermore, AWS is integrating NVIDIA's NVLink Fusion technology with its next-generation Trainium chips. By collaborating with memory suppliers on new high-bandwidth memory (NVHBM) technology, AWS aims to give Trainium access to faster, more power-efficient memory. This integration promises to enhance performance and efficiency for AI workloads by enabling seamless scaling of Trainium and GPUs within a common rack-scale architecture.

Securing AI for National Interests

A significant aspect of the expanded collaboration involves building AI factories for the U.S. government. This initiative includes deploying 100,000 NVIDIA GPUs on secure AWS infrastructure, specifically designed for federal and national security workloads classified at Impact Level 6 (IL6) and above. This move underscores the critical importance of secure, high-performance AI infrastructure for government operations.

The demand for advanced AI infrastructure continues to surge. StartupHub.ai data shows that Nvidia (NASDAQ:NVDA) holds a strong StartupHub score of 82/100 in the semiconductor space, reflecting its dominant position. Competitors like Broadcom (83/100), Celestial AI (72/100), and Cerebras Systems (70/100) are also making strides, but the scale of this AWS-NVIDIA commitment highlights the immense investment required to stay at the forefront of AI compute.

Why This Matters for the AI Industry

This expanded partnership between AWS and NVIDIA is more than just a hardware deal; it’s a strategic alignment that will shape the future of AI development and deployment. By co-engineering solutions across the full stack, from GPUs and CPUs to networking and software, they are creating an optimized environment for AI innovation.

For startups and enterprises, this means greater access to the massive compute power needed for cutting-edge AI models. It also offers more choice and flexibility, allowing them to select the best-suited hardware and software configurations for their specific needs. The availability of NVIDIA Vera CPUs on AWS, for instance, could significantly accelerate the development of sophisticated agentic AI applications.

The focus on secure government AI infrastructure also points to the growing national security implications of AI. The commitment to building dedicated AI factories for the U.S. government signifies a major push to equip defense and intelligence agencies with advanced AI capabilities, potentially influencing geopolitical dynamics.

This news builds on previous announcements, including AWS's prior plans to add over 1 million NVIDIA GPUs starting in 2026. The sheer magnitude of the current commitment, 2 million additional GPUs, indicates that the appetite for AI compute continues to grow exponentially, far exceeding even recent optimistic forecasts. The companies are not just meeting demand; they are actively shaping the infrastructure that will power the next decade of AI advancements.