Meta Platforms surged 7.3% on Thursday after Reuters reported the company will begin manufacturing its own AI chips in September.
Meta Platforms surged 7.3% on Thursday after Reuters reported the company will begin manufacturing its own AI chips in September.

Meta Platforms surged 7.3% on Thursday after Reuters reported the company will begin manufacturing its own AI chips in September.
Meta's plan to produce in-house AI chips starting September threatens Nvidia's dominance in the data center GPU market, where it commands an estimated 80% share of the $62 billion segment.
"Meta's move to build custom silicon is the clearest signal yet that hyperscalers see in-house chips as the only path to controlling AI infrastructure costs," Ross Mayfield, investment strategy analyst at Baird, said.
The stock hit an intraday high of 7.3% on Thursday, building on gains from earlier in the week after Erste Group upgraded Meta to Buy from Hold. The broader tech rally also supported the move: the Nasdaq Composite rose 1.3% to 26,206.89 on Wednesday, and the PHLX Semiconductor Index gained 3.06%. The S&P 500 trades at about 20 times expected earnings, with analysts projecting 24% year-over-year earnings growth for the index.
If Meta's chip achieves comparable performance to Nvidia's H100 — which delivers 990 TFLOPS of FP16 performance — at a lower cost, the savings could run into billions annually. Meta's stock trades at roughly 22 times forward earnings, a premium to the S&P 500 that reflects its AI growth potential but leaves little room for execution missteps.
The strategy mirrors moves by Amazon and Google, both of which have developed custom AI accelerators to reduce reliance on Nvidia's GPUs. Amazon's Trainium3, for example, offers twice the memory bandwidth of Nvidia's H100 at 40% lower cost, according to the company. Meta has not disclosed specifications or benchmark results for its chip, leaving direct comparisons unverified.
For investors, the key question is whether Meta can match Nvidia's performance benchmarks while delivering cost savings. If successful, the in-house chip program could reduce Meta's GPU procurement costs by an estimated $2 billion to $4 billion annually, based on the company's disclosed AI infrastructure spending. Nvidia shares, which have gained 56% over the past 12 months, face renewed pressure as hyperscalers increasingly pursue vertical integration.
This article is for informational purposes only and does not constitute investment advice.