Meta has entered into a significant multiyear agreement to enhance its data center capabilities by acquiring millions of Nvidia's advanced Grace and Vera CPUs, along with Blackwell and Rubin GPUs. This partnership marks a pivotal moment for Meta as it signifies the first large-scale deployment solely utilizing Nvidia's Grace technology. According to Nvidia, this initiative is expected to yield remarkable improvements in performance per watt across Meta's data centers.
In addition to this collaboration, Meta is also developing its own proprietary chips aimed at optimizing AI model operations. However, reports indicate that the company has encountered some technical hurdles and delays in implementing its chip strategy. Meanwhile, Nvidia faces challenges related to asset depreciation and financing for AI expansion, compounded by an increasingly competitive landscape.
Although the financial details of the agreement have not been disclosed, analysts estimate that the overall AI investments from major tech players such as Meta, Microsoft, Google, and Amazon this year could surpass the total expenditure of the Apollo space program.