Amazon Buys 2M Nvidia AI Chips for 2027-28 Data Centers
By Business Desk
Amazon orders an additional 2 million Nvidia AI GPUs for its 2027-2028 data centers, boosting its AI infrastructure and underscoring a massive investment in the AI boom.
Amazon.com Inc. is set to acquire an additional two million high-end Nvidia graphics processing units (GPUs) for its data centers, with deployment scheduled across 2027 and 2028. This significant procurement bolsters Amazon’s infrastructure for training and running advanced artificial intelligence models.
- Additional procurement: 2 million Nvidia GPUs
- Deployment timeline: 2027 and 2028
- Chip types include: Nvidia Blackwell Ultra, Rubin, and Rubin Ultra GPUs
- Supplements: 1 million Nvidia chips announced in March for deployment this year
While financial terms were not disclosed, these advanced GPUs command a substantial price point, underscoring the scale of Amazon’s investment. Nvidia’s GPUs remain fundamental to the ongoing artificial intelligence boom, providing critical computational power for large language model development.
Amazon’s Dual Chip Strategy
Despite its reliance on Nvidia products, Amazon is also actively investing in and developing its own proprietary chips, a strategy mirrored by other major cloud providers like Microsoft and Google.
- Amazon’s Annapurna Labs is creating an AI accelerator designed to compete with Nvidia’s GPUs.
- The forthcoming Trainium AI chip from Amazon is slated to incorporate Nvidia’s networking technology.
This dual approach enables cloud giants to customize hardware for their specific data center environments, ultimately enhancing efficiency and tailored service offerings for their clientele.