What happened
Meta Platforms Inc. plans to develop custom silicon for training future artificial intelligence models, according to CFO Susan Li at a Morgan Stanley technology conference. The company currently deploys custom chips at scale for ranking and recommendation workloads. This initiative proceeds alongside recent agreements with market leaders Nvidia Corp and Advanced Micro Devices Inc. for AI chips and equipment, indicating a dual strategy for its extensive data centre operations.
Why it matters
Meta's expansion into custom AI training silicon signals a strategic shift for large-scale AI operators, impacting procurement teams and hardware architects. This move, despite recent external chip deals, demonstrates a drive to tailor hardware for specific, high-volume AI workloads, potentially reducing reliance on general-purpose accelerators. CTOs and founders should assess the long-term cost and performance benefits of in-house chip development versus external vendor dependence for their unique AI infrastructure needs.
Subscribe for Weekly Updates
Stay ahead with our weekly AI and tech briefings, delivered every Tuesday.
