Nvidia is set to supply 1 million graphics processing units (GPUs) to Amazon’s cloud division, Amazon Web Services, as part of a long-term strategic agreement extending through 2027.
The deal, confirmed by Nvidia executive Ian Buck, will begin rolling out this year and continue over the next several years. While both companies have not disclosed financial terms, the scale of the agreement highlights the growing demand for AI infrastructure in cloud computing.
Beyond GPUs: A Broader AI Ecosystem
The agreement goes beyond just GPUs. It includes a mix of Nvidia’s advanced technologies such as:
- Spectrum networking chips
- ConnectX networking solutions
- Newly introduced Groq AI chips
- Additional AI-focused hardware from Nvidia’s portfolio
These components will be deployed across AWS data centers to support large-scale AI workloads.
Focus on AI Inference
A key focus of the partnership is AI inference—the process where trained AI models generate outputs and perform real-world tasks.
According to Ian Buck:
“Inference is hard. It’s not a one-chip solution—we use multiple chips to achieve optimal performance.”
AWS plans to utilize a combination of seven different Nvidia chips to improve efficiency and performance in inference workloads.
Strategic Importance for AI Growth
This collaboration aligns with Nvidia CEO Jensen Huang’s broader vision, where the company sees a $1 trillion market opportunity driven by its next-generation Rubin and Blackwell chip architectures.
Additionally, AWS integrating Nvidia’s networking technologies marks a notable shift, as AWS has traditionally relied on its own custom-built infrastructure.
What This Means
- Strengthens Nvidia’s dominance in AI hardware
- Expands AWS’s capabilities in AI and machine learning
- Accelerates innovation in large-scale AI deployment
- Signals massive growth in AI cloud infrastructure demand
