Broadcom and Meta announced a multi-year, multi-generation strategic partnership to support Meta’s rapidly scaling artificial intelligence compute infrastructure. Building on their existing partnership, Broadcom will deliver technology supporting Meta Training and Inference Accelerator (MTIA) chips, with plans to extend through 2029. This technology will serve as the foundational backbone for Meta’s deployment of state-of-the-art AI data centers.
Meta is partnering with Broadcom to rollout industry’s first 2nm AI compute accelerator; foundation for sustained multi-year infrastructure rollout. Over the next three years, the companies are collaborating on next generations of AI accelerator chips to meet Meta’s escalating compute demands for next-generation AI models. Partnership built on Broadcom’s industry-leading XPU platform to deeply co-develop and optimize Meta’s AI infrastructure across multiple silicon generations. Deployment utilizes Broadcom’s advanced Ethernet technologies to enable seamless, high-bandwidth scale-up, scale-out, scale-across networking, eliminating bottlenecks throughout Meta’s rapidly expanding AI compute clusters.
The initial commitment exceeds 1GW and is the first phase of a sustained, multi-gigawatt rollout. This reinforces a shared roadmap to co-design and scale the hardware required to bring real-time generative AI features and “personal superintelligence” to billions of people globally across apps like WhatsApp, Instagram, and Threads.
Central to this expanding infrastructure footprint is the rapid deployment of Meta’s MTIA silicon, made possible by deep engineering co-design through Broadcom's foundational XPU (Custom Accelerator) platform. This platform approach allows Broadcom and Meta to tightly couple logic, memory, and high-speed I/O for current deployments, while establishing a highly adaptable, multi-generation blueprint to co-develop future iterations of the MTIA portfolio over the coming years. MTIA is a key pillar of Meta's broader silicon strategy, which deploys different accelerators for different workloads — the MTIA portfolio matches purpose-built hardware to optimize both performance and total cost of ownership at scale.
“We are pleased to expand our strategic collaboration with Meta as they pioneer the next frontier of artificial intelligence,” said Hock Tan, President and CEO, Broadcom. “This initial MTIA deployment is just the beginning of a sustained, multi-generation roadmap to serve the trajectory of massive growth over the next few years highlighting Broadcom’s unmatched leadership in AI networking and the power of our foundational XPU custom accelerator platform.”
“Meta is partnering with Broadcom across chip design, packaging, and networking to build out the massive computing foundation we need to deliver personal superintelligence to billions of people,” said Meta founder and CEO Mark Zuckerberg. “As we roll out more than 1GW of our custom silicon to start and then multiple gigawatts over time, this partnership will give us greater performance and efficiency for everything we’re building.”
To support the massive compute density of MTIA clusters and future expansions, Broadcom is delivering its industry-leading Ethernet networking solutions to address scale-up, scale-out, and scale-across requirements. Broadcom’s high-radix Ethernet switches, Optical Connectivity products, PCIe switches, and high-speed SerDes capabilities provide a standards-based, low-latency fabric. This future-proof Ethernet backbone is critical for scaling-up compute bandwidth within individual MTIA racks, and scaling-out and scaling-across network capacity over tens of thousands of nodes, eliminating congestion during the most intensive AI workloads.
This comprehensive, multi-generational suite of MTIA and networking solutions will be instrumental in scaling the latest and future generations of MTIA clusters.
The partnership extends well beyond hardware supply, focusing heavily on continuous system-level optimization and forward-looking R&D. Because MTIA prioritizes inference and low-precision processing for maximum efficiency, the supporting infrastructure must offer near-zero latency. Broadcom’s Ethernet-based rack-scale interconnects will ensure that current and future MTIA grids remain constantly utilized, gracefully handling evolving memory hierarchies while drastically reducing the Total Cost of Ownership (TCO) across the lifespan of Meta's multi-generation infrastructure.
As Meta continues to invest in the infrastructure required to bring AI to billions of people, the Broadcom partnership and MTIA program will remain a critical piece of that strategy delivering the custom, workload-optimized silicon that underpins Meta's AI future.
Given the scale of this expanded partnership, Hock Tan will transition off of Meta’s Board of Directors and into an advisor role for Meta, where he'll provide guidance on Meta’s custom silicon roadmap and help shape the future of their infrastructure investments.
