Meta is building a custom chip to power AI models, the company said in a blog post. This comes after the company’s CEO Mark Zuckerberg recently said they see “an opportunity to introduce AI agents to billions of people in ways that will be useful and meaningful.”

The Meta Training and Inference Accelerator (MTIA) chip is an in-house, custom accelerator chip family targeting inference workloads. Meta’s VP and head of infrastructure, Santosh Janardhan, said the chip provides “greater compute power and efficiency” than CPUs.

In addition, Meta has introduced its first in-house ASIC solution, MSVP, designed to power constantly growing video workloads. The company has also touted the Research SuperCluster (RSC) AI Supercomputer, which features 16,000 GPUs and is built to power new augmented reality tools, content management systems, and real-time translation technology.

“Over the next decade, we will see increased specialisation and customisation in chip design, purpose-built and workload-specific AI infrastructure, new systems and tooling for deployment at scale, and improved efficiency in product and design support,” the company said in its statement.