Meta is preparing to put its own artificial-intelligence chip, code-named Iris, into production in September, according to reporting on an internal company memo cited by CNA. The move is part of Meta's push to sharply expand its computing power — the memo indicates the company is aiming to roughly double its computing capacity.
The effort places Meta alongside partners in a broader industry shift. According to Digitimes, Meta, MediaTek, and TSMC are jointly advancing AI chips in a bid to challenge Google, which has long built its own custom silicon. Building a proprietary chip lets a company tailor hardware to its specific AI workloads and reduce reliance on outside suppliers.
Much of the coverage frames Iris as a potential challenge to Nvidia, whose graphics processors dominate the market for training and running AI systems. Finance.biggo.com describes the chip as "posing a new challenge to Nvidia's dominance."
But analysts urge caution about how far that challenge goes. According to Yahoo Finance, analyst Daniel Newman said Meta is not replacing Nvidia or AMD with its in-house chips — instead, "it is augmenting" its existing hardware. In other words, custom silicon is expected to supplement, rather than displace, the chips Meta already buys.
Investors have responded positively to Meta's chip ambitions. According to TIKR.com, Meta's stock gained 12% last week as the company moved toward Iris production.
Why it matters: If big tech companies like Meta can design more of their own AI chips, they gain leverage over cost and supply in one of the most expensive and supply-constrained parts of the AI boom — a shift that could gradually reshape the balance of power among chipmakers like Nvidia and AMD.