Meta Greenlit to Begin Manufacturing Custom AI Chip 'Iris' in September
Meta has cleared the way to begin manufacturing its custom data-center AI chip "Iris" in September as part of a broader strategy to reduce reliance on Nvidia and scale its computing infrastructure to 14 gigawatts by 2027.
Custom Silicon Production Ramping
Meta Platforms plans to begin manufacturing its custom data-center AI chip, codenamed "Iris," in September as part of a broader four-generation MTIA roadmap, with the company aiming to scale its computing infrastructure to 14 gigawatts by 2027, up from seven gigawatts targeted for 2026. The chip, developed with Broadcom and manufactured by TSMC, completed testing in just six weeks with no major issues, and represents Meta's most aggressive push yet to reduce dependence on Nvidia and AMD GPUs while supporting its expanding AI ambitions across recommendation systems, content moderation, and generative tools.
Infrastructure Scaling
Meta's aggressive doubling of power capacity from 7 gigawatts in 2026 to 14 gigawatts in 2027 signals massive expansion in AI compute resources. This infrastructure buildup will support the company's internal AI training and inference needs for Facebook, Instagram, and other platforms. The custom chip development parallels similar vertical integration efforts by other hyperscale tech companies seeking to optimize their AI hardware stacks.
Strategic Implications
Meta's investment in custom silicon reflects the broader industry shift toward vertical integration as AI infrastructure costs escalate. By designing chips specifically tailored to its workloads, Meta can improve performance-per-watt and reduce costs compared to relying entirely on commodity GPUs. The rapid testing cycle suggests confidence in the design and manufacturing partnership with TSMC and Broadcom.
Competitive Positioning
The move positions Meta among a small group of companies—including Google, Amazon, and others—developing custom AI accelerators. This strategy allows them to differentiate their AI capabilities and control their supply chains amid geopolitical tensions and chip export restrictions.