Meta is reportedly testing an in-house chip for training AI systems, a part of a strategy to reduce its reliance on hardware makers like Nvidia.
According to Reuters, Meta’s chip, which is designed to handle AI-specific workloads, was manufactured in partnership with Taiwan-based firm TSMC. The company is piloting a “small deployment” of the chip and plans to scale up production if the test is successful.
Meta has deployed custom AI chips before, but only to run models — not train them. As Reuters notes, several of the company’s chip design efforts have been canceled or otherwise scaled back after failing to meet internal expectations.
Meta expects to spend $65 billion on capital expenditure this year, much of which will go toward Nvidia GPUs. If the company manages to reduce even a fraction of that cost by shifting to in-house chips, it’d be a big win for the social media giant.