AWS, NVIDIA to deploy 2 million more GPUs for AI workloads

Amazon Web Services and NVIDIA will deploy 2 million additional graphics processing units for artificial intelligence workloads and expand their collaboration into central processing units, networking and robotics.

The move further scales infrastructure tied to surging demand for AI computing, as cloud providers and chipmakers race to add processing capacity for model training and deployment.

The deployment will add to the large installed base of NVIDIA hardware already used across AWS's cloud platform. NVIDIA has become the dominant supplier of AI chips, while AWS has sought to balance demand for NVIDIA systems with its own in-house silicon, including Trainium and Inferentia.

Wider build-out

The broader tie-up points to a relationship that now extends beyond graphics processors. By naming CPUs, networking and robotics alongside GPUs, the companies signalled that they are working across a wider set of technologies at the centre of data centre expansion and AI system design.

Cloud infrastructure groups have been spending heavily to secure enough processors, networking equipment and power to keep pace with corporate and consumer use of generative AI services. NVIDIA's chips remain central to that build-out because they are widely used to train large language models and run inference tasks after deployment.

For AWS, the additional GPUs could help it serve customers that want access to NVIDIA systems through the cloud rather than build their own data centre estates. The arrangement also reflects how large cloud operators are trying to preserve flexibility by combining merchant chip supply with internally developed processors and tightly integrated networking gear.

Robotics is a newer area of overlap. NVIDIA has pushed software and hardware platforms for robotic systems and industrial automation, while Amazon has long invested in warehouse automation and machine learning systems in its logistics network. A deeper collaboration in that segment suggests the partnership could extend beyond cloud computing into physical AI applications.

Competitive pressure

The announcement comes as competition intensifies among hyperscale cloud providers to secure the equipment needed for AI services. Microsoft, Google and other large operators have been expanding their infrastructure footprints, signing supply agreements and introducing their own chip designs in an effort to reduce costs and manage dependence on external vendors.

That backdrop has increased NVIDIA's strategic importance to the biggest technology groups. Even as customers pursue alternatives, its GPUs and networking products remain deeply embedded in AI system architecture, making access to supply a priority for cloud operators serving start-ups, developers and large corporate clients.

The inclusion of CPUs in the expanded alliance is also notable because it suggests closer alignment in how AI systems are assembled. While GPUs do most of the heavy lifting for many AI tasks, CPUs remain essential for orchestrating workloads, managing memory and running broader application environments inside data centres.

Networking has become equally important as AI clusters grow larger. Moving data quickly between processors is critical to training advanced models efficiently, and NVIDIA has expanded in that segment through high-speed interconnect and data centre networking products. Closer work with AWS on networking could therefore shape how future AI clusters are configured within the cloud provider's infrastructure.

Industry context

The scale of 2 million additional GPUs underlines how AI has become a capital-intensive contest among the world's largest technology companies. Building and operating systems at that size requires substantial investment not only in chips, but also in servers, networking, software integration, cooling and electricity.

Investors and customers have been watching whether spending on AI infrastructure will translate into sustainable revenue growth for cloud providers and semiconductor groups. AWS has been under pressure to show that demand for AI services can support the cost of rapid expansion, while NVIDIA has faced questions about how long the current level of demand can continue.

Even so, the latest deployment shows that both companies are still preparing for further growth in AI use across business software, consumer applications and industrial systems. The addition of 2 million GPUs, alongside broader work in CPUs, networking and robotics, points to a partnership becoming more deeply woven into the infrastructure behind the current AI market.

NVIDIA is the dominant supplier of AI chips, while AWS remains one of the world's largest cloud operators seeking to meet rising customer demand for access to advanced computing systems.