What happened
Meta plans to begin manufacturing its own data-center AI chip, code-named Iris, in September. The chip was designed with help from Broadcom and will be built by Taiwan Semiconductor Manufacturing. Testing took about six weeks with no major issues reported, and Meta plans to launch a new chip about every six months through 2027.
Why it matters
Meta expects to spend as much as $145 billion（約23兆円） on AI infrastructure this year. Custom silicon is designed to extract more computing power from each dollar spent. The company plans to bring about 7 gigawatts of computing capacity online this year and double that to 14 gigawatts in 2027, with Iris augmenting (not replacing) GPUs from Nvidia and Advanced Micro Devices. This suggests Meta may need Nvidia somewhat less over time, while handing more business to Broadcom and TSMC.
What to watch
Meta raised its 2026 capital expenditure forecast to a range of $125 billion（約20兆円） to $145 billion（約23兆円） (up from $115 billion（約18兆円） to $135 billion（約22兆円）), and is guiding for second-quarter revenue of $58 billion（約9.3兆円） to $61 billion（約9.8兆円）. In the first quarter, capital expenditures alone were $19.8 billion（約3.2兆円）, while first-quarter revenue rose 33% year over year to $56.3 billion（約9兆円） and earnings per share grew 62% year over year.
Summaries like this, in your inbox every morning.
Meta's announcement of in-house chip manufacturing marks a strategic shift in how the company approaches its massive AI infrastructure investments. Rather than cutting spending, the custom chip program is designed to make existing spending more efficient—Meta is still doubling its computing capacity from 7 gigawatts this year to 14 gigawatts in 2027, but aims to reduce the per-unit cost. The chip development program appears ahead of investor expectations; testing took only about six weeks with no major issues, and the planned launch cadence of roughly one new chip every six months is faster than the industry norm of about one per year.
Meta's core business remains strong enough to fund this expansion. First-quarter revenue grew 33% year over year to $56.3 billion（約9兆円）, accelerating from 24% growth in the fourth quarter of 2025, while operating margins reached 41% and earnings per share grew 62% year over year. This financial momentum is what enables the company to pursue a four-generation family of custom chips without cutting its overall spending.
For chip industry participants, the news carries a nuanced message. Broadcom gains design work and TSMC gains manufacturing volume, but Nvidia faces potential long-term pricing pressure as each Iris chip Meta deploys represents a marginal reduction in Nvidia GPU purchases. However, Nvidia remains central to Meta's strategy, and the success of this program depends on whether the AI investments ultimately produce more user engagement and better ad economics—if they do not, the cost savings from custom silicon alone may prove insufficient.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
