SAN FRANCISCO, 19 March 2026 – Nvidia has struck a major multi-year agreement to supply up to one million artificial intelligence (AI) chips to Amazon’s cloud division by the end of 2027, underscoring the intensifying global race to build next-generation AI infrastructure.
The deal, which will see shipments begin in 2026 and continue through 2027, positions Nvidia at the centre of Amazon Web Services’ (AWS) expanding AI ecosystem, as demand for high-performance computing surges across industries.
While financial terms were not disclosed, the scale of the agreement highlights the enormous capital being deployed into AI infrastructure globally, with hyperscale cloud providers racing to secure computing capacity to power generative AI, automation and enterprise applications.
Beyond the headline figure of one million graphics processing units (GPUs), the agreement includes a broader suite of Nvidia technologies. These encompass networking chips such as Spectrum, as well as newly integrated AI inference hardware, including chips developed through Nvidia’s partnership with Groq.
AWS plans to deploy a combination of seven different Nvidia chips to optimise AI inference, the process by which AI systems generate outputs and perform tasks for users. This reflects a growing shift in the AI industry from model training to real-time deployment at scale, where efficiency and speed are critical.
The collaboration also marks a notable evolution in AWS’s infrastructure strategy. Nvidia’s ConnectX and Spectrum-X networking technologies will be integrated into AWS data centres, complementing Amazon’s traditionally in-house networking systems. This signals deeper alignment between the two companies as they co-develop high-performance AI infrastructure.
The deal comes amid an unprecedented boom in AI investment. Nvidia has projected that its next-generation Rubin and Blackwell chip platforms could unlock a US$1 trillion revenue opportunity over the coming years, driven by demand from cloud providers, enterprises and governments.
At the same time, competition in the AI chip market is intensifying. Major cloud players, including Amazon itself, are developing proprietary chips such as Trainium to reduce reliance on external suppliers. However, Nvidia’s continued dominance in high-performance GPUs has ensured its central role in powering large-scale AI workloads.
For Amazon, the agreement reinforces its commitment to expanding AI capabilities within AWS, enabling customers to deploy increasingly complex AI models and applications. The integration of multiple chip architectures also reflects a broader industry trend toward hybrid computing environments, where different types of processors handle specialised tasks.
For Nvidia, the deal further cements its position as the backbone of the global AI economy. With similar large-scale agreements already in place with companies like Meta, the chipmaker is rapidly scaling its presence across hyperscale cloud platforms, ensuring sustained demand for its hardware.
The implications extend beyond the technology sector. As AI becomes embedded across industries — from finance and healthcare to logistics and manufacturing, the infrastructure underpinning these systems is emerging as one of the most strategic battlegrounds in the global economy.
In this context, the Nvidia-Amazon partnership is more than a supply agreement. It is a signal of where the next phase of digital transformation is headed, one defined by massive computing power, integrated ecosystems and an accelerating race to dominate artificial intelligence at scale.
