FIRST LOOK: Meta's Iris Chip and the $145B Plan to Ditch Nvidia

2026-07-10 · Source: AIM Network · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, short

Summary

Meta Platforms is initiating manufacturing of its in-house AI chip, code-named Iris, in September, marking a significant advancement in its Meta Training and Inference Accelerators (MTIA) program. This fourth-generation chip project, designed with Broadcom and manufactured by TSMC, aims to power AI systems for Facebook, Instagram, WhatsApp, including recommendation engines, advertising tools, and large-scale model training. Testing of Iris reportedly took just 6 weeks with no major issues, signaling rapid progress after years of stalled efforts. Meta plans an aggressive release schedule, shipping new chips every 6 months through 2027, and projects spending up to \$145 billion on AI infrastructure this year, part of a broader \$700 billion industry spend. The company intends to deploy 7 GW of computing infrastructure this year, scaling to 14 GW by 2027, and has secured long-term supply deals with Samsung Electronics for memory chips, SanDisk for flash storage, and Sumitomo Electric for supporting hardware. Iris is designed to augment, not replace, existing Nvidia GPUs for heavy training workloads, reflecting a broader industry trend where major AI companies are developing proprietary hardware to reduce dependence on external suppliers and mitigate supply chain bottlenecks.

Key takeaway

For AI Architects or VPs of Engineering managing large-scale AI infrastructure, Meta's Iris chip initiative highlights the strategic imperative of vertical integration. Your organization should evaluate the long-term cost and supply chain risks of relying solely on external GPU providers. Consider investing in custom hardware development or securing diversified supply agreements to ensure future AI compute capacity and reduce dependency on a single vendor.

Key insights

Meta's Iris chip program aims to reduce Nvidia dependence and secure AI infrastructure through aggressive in-house hardware development.

Principles

• In-house chip development reduces external dependency.
• Aggressive release cycles accelerate infrastructure build-out.
• AI dominance requires owning underlying hardware.

Method

Meta's MTIA program involves designing chips with Broadcom, manufacturing with TSMC, and deploying them to power internal AI systems like recommendation engines and advertising tools.

In practice

• Develop custom silicon for core AI workloads.
• Secure long-term supply deals for critical components.
• Plan for rapid, iterative hardware releases.

Topics

• Meta Iris Chip
• AI Accelerators
• Custom Silicon
• Supply Chain Resilience
• NVIDIA Dependence
• Data Center Infrastructure

Best for: Investor, CTO, Director of AI/ML, AI Architect, VP of Engineering/Data

Related on AIssential

• Nvidia and Meta Agree to Wide-Ranging new AI Chip Deal — Meta and Nvidia deepen their AI hardware partnership, focusing on next-gen GPUs, CPUs, and networking for large-scale AI infrastructure.
• Meta’s new AI chips will begin production in September — Major tech companies are developing custom AI chips to reduce costs and optimize for specific workloads.
• Why Everyone Is Moving Away from NVIDIA — Hyperscalers are vertically integrating AI infrastructure, developing custom silicon and energy solutions to overcome GPU bottlenecks.
• Deepseek is designing its own AI chip — Deepseek, like other AI leaders, is developing custom inference chips to reduce external hardware dependency and control costs.
• 615: Nvidia Q4, Ben Evans on AI Economics, Meta + Google TPUs, China Solar at 1TW/Year, Anthropic vs DoD, AWS Outages, METR, Sora, and Marty Supreme — Diversifying AI compute suppliers and architectures is crucial for large-scale AI development, despite integration challenges.
• Nvidia invests $2bn each in Lumentum and Coherent for AI optics R&D — Nvidia is strategically investing in optical and physical AI technologies to secure supply chains and accelerate AI infrastructure development.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AIM Network.