Meta's MTIA chip ambitions threaten Nvidia's AI dominance at 21x earnings

Meta Platforms' custom MTIA chips, refreshed every six months, threaten Nvidia's GPU dominance as hyperscalers shift toward in-house silicon for inference workloads, potentially reshaping the $62 billion AI chip market. The move mirrors similar efforts at Alphabet's Google, whose Tensor Processing Units now power a growing share of its AI workloads, and Amazon, whose Trainium and Inferentia chips target specific AI compute tasks.

"Meta's six-month innovation loop with MTIA shows how serious the company is about securing its custom silicon future," Joey Frenette, a 24/7 Wall St. contributor, wrote. "The whole Meta AI story is fundamentally misunderstood."

Meta's MTIA chips target generative AI and recommendation inference — the fastest-growing segment of AI compute demand. The six-month iteration cycle outpaces Nvidia's roughly two-year GPU architecture cadence, though Nvidia's H100 delivers 990 TFLOPS of FP16 performance versus MTIA's more specialized inference focus. Meta trades at 20.6 times trailing earnings, a fraction of Nvidia's 31 times forward multiple, suggesting the market has not priced in potential savings from in-house silicon.

If Meta's custom chips reduce GPU procurement costs by even 20%, the savings could exceed $4 billion annually based on the company's $37 billion in 2025 capital expenditure. That margin expansion is not reflected in Meta's current valuation, creating upside for investors who see the chip strategy as more than a defensive move.

Semiconductor Sector Volatility Adds Risk to Nvidia's Premium

The broader semiconductor sector is showing cracks. The PHLX Semiconductor Sector index lost more than 10% on June 5, erasing $1.4 trillion in market value after a stronger-than-expected jobs report raised the odds of Federal Reserve rate hikes. Nvidia fell 6% in that session. Broadcom, despite reporting AI revenue up 143% year over year to $10.8 billion, dropped nearly 8%.

The selloff highlights a growing risk for Nvidia investors: the stock's 31 times forward earnings multiple assumes uninterrupted AI chip demand growth. Any sign of CapEx moderation from hyperscalers — Meta, Google, Amazon, Microsoft — could trigger a re-rating. Bear ETFs targeting the semiconductor sector have seen increased inflows in recent weeks, reflecting hedging demand.

Meta's Foundry Strategy Carries Its Own Risks

Meta designs its MTIA chips but relies on Taiwan Semiconductor Manufacturing Co. for fabrication. TSMC controls 73% of the global foundry market, according to Counterpoint Research, creating single-source risk for any custom silicon program. Any disruption at TSMC's fabs — whether from geopolitical tension or natural disaster — would halt MTIA production and force Meta back to the merchant GPU market at potentially higher prices.

The foundry dependency also means Meta competes for TSMC's advanced 3nm and upcoming 2nm capacity against the same companies it seeks to displace. Nvidia, Broadcom, and AMD all rely on TSMC for their most advanced chips, creating a capacity bottleneck that could limit MTIA's production scale.

Meta shares, trading at 21 times trailing earnings, offer a lower-risk entry point to the AI chip theme compared with Nvidia at 31 times forward earnings. If Meta's MTIA program delivers on its promise, the stock could see multiple expansion as margins improve. If it falters, Meta retains the option to continue buying Nvidia GPUs — a hedge that pure-play chip companies lack.

This article is for informational purposes only and does not constitute investment advice.