Meta commits to a multiyear deal to buy Nvidia chips, including Vera Rubin; source: Meta's in-house chip strategy had suffered technical challenges and delays (Financial Times)

ThinkingNews Desk · how this was written
Meta has entered a multiyear agreement to purchase millions of Nvidia AI chips, including the Vera Rubin GPU as well as Grace CPUs and Blackwell GPUs. The deal aims to bolster Meta’s AI hardware after technical challenges and delays in its own chip strategy, with the components slated for deployment in Meta’s data-center infrastructure.
Written from all 3 reports below, not from any single one.
How it was reported
- TechMeme·Meta commits to a multiyear deal to buy Nvidia chips, including Vera Rubin; source: Meta's in-house chip strategy had suffered technical challenges and delays (Financial Times)
Meta has committed to a multiyear deal to buy Nvidia chips, including Vera Rubin. The purchase is due to technical challenges and delays in Meta's in-house chip strategy. Meta will buy millions of chips to support its AI hardware development.
- The Verge·Meta’s new deal with Nvidia buys up millions of AI chips
Meta has struck a deal with Nvidia for millions of AI chips. The deal includes Nvidia's Grace and Vera CPUs and Blackwell and Rubin GPUs. Meta will deploy these chips in its data centers for significant performance improvements.
- Wired·Nvidia’s Deal With Meta Signals a New Era in Computing Power
Nvidia's deal with Meta marks a shift in computing power. AI companies now require a range of components, including GPUs and CPUs. This signals a new era in computing.
Related stories
- Nvidia and AI chip startup d-Matrix are combining their hardware in a new system to power AI models, the latest example of Nvidia partnering with rivals (Phoebe Liu/The Information)3 outlets
- Meta may trade AI chips for shares in its latest AMD deal4 outlets
- Meta signs multibillion-dollar deal for Amazon Graviton5 chips as AI compute demand outstrips $135B capex budget3 outlets
- Nvidia announces the Nemotron Coalition, which includes Thinking Machines Lab, Cursor, and Mistral and will develop an open model trained on Nvidia DGX Cloud (Sebastian Moss/DatacenterDynamics)5 outlets