Broadcom Meta AI Partnership Extends Through 2029
Broadcom Meta AI partnership will co-develop 2nm accelerators and staged silicon, reshaping Meta's hardware roadmap and prompting trader repositioning.
KEY TAKEAWAYS
- Partnership will co-develop the industry's first 2nm AI accelerator for Meta's MTIA program.
- Initial commitment exceeds 1GW of custom silicon as phase one of a multi-gigawatt rollout.
- Broadcom XPU and Ethernet integration scale Meta's datacenter networking for real-time generative AI.
HIGH POTENTIAL TRADES SENT DIRECTLY TO YOUR INBOX
Add your email to receive our free daily newsletter. No spam, unsubscribe anytime.
Broadcom Inc. (AVGO) and Meta Platforms (META) announced on April 14, 2026, a partnership to co-develop next-generation AI accelerators for Meta’s MTIA program. The deal includes a staged supply of custom silicon and integrated networking to expand Meta’s AI infrastructure.
Technology Development and Capacity Commitments
The collaboration focuses on creating the industry’s first 2nm AI accelerator for Meta’s MTIA (Meta Training and Inference Accelerator) chips. This effort will use Broadcom’s XPU custom-accelerator platform, which integrates logic, memory, and high-speed input/output across multiple silicon generations to support both AI training and inference workloads.
The initial supply commitment exceeds 1GW of custom silicon as the first phase of a sustained, multi-gigawatt rollout through 2029. This program aims to provide the computing power necessary for large-scale AI model training and deployment. The partnership also includes Broadcom’s Ethernet technologies to enable scale-up, scale-out, and scale-across networking within Meta’s AI clusters, linking chip design with datacenter networking.
Strategic Roadmap and Governance
Broadcom and Meta outlined a shared hardware roadmap designed to enable real-time generative AI features across Meta’s consumer apps, including WhatsApp, Instagram, and Threads. By combining next-generation accelerators with networking and systems engineering, the initiative seeks to align device-level and datacenter computing with the operational demands of generative AI models.
The staged supply commitment and networking integration position Broadcom as a long-term supplier for Meta’s rapidly expanding AI infrastructure. The partnership aligns multi-year engineering efforts between the companies. As part of the arrangement, Broadcom CEO Hock Tan is reported to be leaving Meta’s board. The deal formalizes a multi-year collaboration on silicon and systems to support Meta’s goal of running real-time generative AI features at scale.
