AI & Technology
Meta Plans September Production for Iris AI Chip as Compute Push Accelerates
The custom processor is part of Meta’s effort to expand AI infrastructure, control costs and reduce dependence on external chip suppliers.
By 19Network Editorial Team · Jul 10, 2026 · 5 min read
Meta plans to put its in-house Iris AI chip into production in September as it expands computing capacity.
Why Meta wants its own AI chips Meta operates Facebook, Instagram, WhatsApp and a growing range of AI products. Serving recommendations, advertisements, generative assistants and other machine-learning features requires enormous amounts of computing capacity. External suppliers such as Nvidia and AMD remain important to the industry, particularly for advanced accelerators. But a custom chip designed for a company’s specific workloads can improve efficiency and give the company more control over hardware planning, software integration and costs. Iris is being developed within Meta’s Training and Inference Accelerators programme. The initiative is intended to create processors suited to the company’s internal systems rather than compete directly as a general-purpose chip sold to outside customers. A broader move toward vertical integration Meta is not alone. Major cloud and consumer-technology companies have spent years developing proprietary processors for AI and data-centre tasks. The trend reflects a strategic shift: AI infrastructure is no longer treated simply as equipment that can always be purchased when needed. It is becoming a core capability tied to product road maps and…