Meta’s Bold Move: Testing In-House AI Training Chips

Meta is taking a bold step toward reducing its dependence on hardware giants like Nvidia by testing its own AI training chip. According to Reuters, the social media powerhouse has developed the chip in collaboration with Taiwan-based TSMC and is currently running a small-scale pilot deployment. If the tests prove successful, Meta aims to ramp up production reportedly.

While the company has previously introduced custom AI chips, they were designed solely for running models—not training them. Past chip design projects at Meta have faced setbacks, with some being scrapped or scaled back due to falling short of expectations.

Read more: Meta Enters the Race to Develop AI-Powered Humanoid Robots

This year, Meta plans to pour a staggering $65 billion into capital expenditures, with a significant chunk allocated to Nvidia GPUs. However, if its in-house chip strategy pays off, even a modest reduction in these costs could mark a major breakthrough for the tech giant.

For more updates, be with Markedium.