Just days after launching its Muse Spark AI model, Meta is now partnering with Broadcom to develop custom silicon to ensure it has the compute foundation to deliver on its long-term AI ambitions. The partnership will see the pair develop the next generation of Meta Training and Inference Accelerator (MTIA) chips.
The social media company said that this custom silicon will be what’s used to help power AI across its apps and services. It recently announced that it is developing and deploying four new generations of MTIA chips within the next two years. This partnership with broadcom will help to accelerate this.
Under the agreement, Broadcom will work with Meta to design the chips, create advanced packaging, and work on networking to help build out a “massive” computing foundation needed to deliver AI to billions of users. The work will be built on Broadcom’s XPU platform and its advanced Ethernet technologies, which will enable high-bandwidth networking across Meta’s expanding AI compute clusters.
Commenting on the deal, Meta founder and CEO Mark Zuckerberg said:
“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. 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.”
As part of this development, Broadcom President and CEO, Hock Tan, will transition off Meta’s Board of Directors and move to an advisor role for the company. There, he will provide guidance on Meta’s custom silicon roadmap.