Meta begins testing its first internal AI training chip to reduce its reliance on NVIDIA and others

According to foreign media reports, two sources have revealed that Meta is testing its first in-house chip for training artificial intelligence systems, a major milestone in the company's shift towards designing more of its own custom chips and reducing its reliance on external suppliers such as Nvidia. The world's largest social media company has started small-scale deployments of the chip, and if the testing goes well, the company is ready to expand production for large-scale use, the sources said. Developing the in-house chip is part of Meta's long-term plan to reduce its massive infrastructure costs. Meta's new training chip is a dedicated accelerator designed to handle only tasks aimed at artificial intelligence, one of the sources said. This could make it more energy efficient than the integrated graphics processing units (GPUs) typically used for artificial intelligence workloads. Meta is working with Taiwanese chipmaker TSMC to produce the chip, the person said.