Meta Starts MTIA AI Chip Production in September as 7 Gigawatts of Compute Loom This Year
Updated
Updated · TechCrunch · Jul 9
Meta Starts MTIA AI Chip Production in September as 7 Gigawatts of Compute Loom This Year
3 articles · Updated · TechCrunch · Jul 9
Summary
September is Meta’s target to begin manufacturing the latest MTIA AI chips, after at least one chip cleared testing in about six weeks, according to a Reuters-cited internal memo.
The in-house chips are meant to cut reliance on costly Nvidia and AMD GPUs during a component shortage, while handling training, recommendation and broader inference workloads across Meta’s apps.
Broadcom is designing the chips, TSMC will manufacture them, and Meta is sourcing RAM from Samsung, storage from Sandisk and fiber-optic equipment from Sumitomo Electric.
Meta has produced AI chips since 2023, but its spending remains huge: it forecast $125 billion to $145 billion in 2026 capital expenditures and plans to deploy 7 gigawatts of compute this year, doubling that next year.
The push mirrors a wider industry shift as OpenAI, Amazon and Google pursue custom AI silicon to curb surging infrastructure costs and dependence on Nvidia.
Amid a global chip shortage, can Meta's new silicon truly challenge Nvidia or will it hit a manufacturing wall?
As AI's energy thirst grows, will the data center boom force a choice between tech progress and affordable public utilities?
Meta’s Iris AI Chip and $145 Billion Investment: Transforming AI Infrastructure and Industry Competition
Overview
Meta is launching its Iris AI chip (MTIA 400) in September 2026, marking a major step in its move toward in-house silicon. After a rapid and successful testing phase, Meta is aggressively pushing into custom hardware development to optimize its vast AI infrastructure. This shift aims to reduce reliance on external chip manufacturers and give Meta greater efficiency and control over its technology stack. The introduction of the Iris chip is a pivotal moment in Meta’s strategy, supporting its goal to lead in AI innovation and manage the growing demands of its global platforms.