Meta has secured a multiyear agreement to enhance its data centers by integrating millions of Nvidia’s Grace and Vera CPUs along with Blackwell and Rubin GPUs. Although Nvidia’s hardware has been a staple for Meta’s AI initiatives for some time, this partnership marks the “first large-scale deployment” of Nvidia’s Grace CPUs exclusively, promising “substantial performance-per-watt improvements” within Meta’s data facilities. Additionally, the deal outlines the intention to incorporate Nvidia’s forthcoming Vera CPUs into Meta’s infrastructure by 2027.

Meanwhile, Meta is developing its own proprietary chips to support AI models, but as reported by the Financial Times, the company has encountered “technical challenges and rollout delays” with this initiative. Nvidia is also navigating issues related to depreciation, chip-backed loans for financing AI expansions, and competitive pressures. Following a November report regarding Meta’s potential interest in Google’s Tensor chips for AI, Nvidia’s stock experienced a four percent dip. Moreover, late last year, AMD announced chip partnerships with both OpenAI and Oracle, intensifying the competitive landscape.

Although the financial terms of the Nvidia and Meta deal remain undisclosed, the collective AI expenditure by tech giants such as Meta, Microsoft, Google, and Amazon is projected to surpass the total cost of the Apollo space program this year.