• by Colin Kirkland , October 15, 2025

Meta has expanded its partnership with semiconductor design company Arm, a competitor of GPU chip-manufacturer Nvidia. The deal shows Meta making a major investment in the global efficiency of its future AI systems.

According to a statement from Arm, Meta is planning to utilize the British semiconductor company’s software and data center infrastructure resources to scale its AI systems, relocating its AI ranking and recommendation systems to Arm's cloud-optimized Neoverse platform.

Meta’s Head of Infrastructure, Santosh Janardhan, says that partnering with Arm will help Meta scale its AI systems to its 3 billion app and technology users, while Arm CEO Rene Haas says that “AI's next era will be defined by delivering efficiency at scale.”

By moving the AI systems that power discovery and personalization across its family of social media apps, Meta will begin leveraging Arm’s Neoverse-based data center platforms, ultimately aiming for higher performance and lower power consumption.

Arm’s statement claims that its Neoverse platform is designed to help Meta achieve “performance-per-watt parity” via “the joint tuning of open source components,” which invite developers to build AI systems, as well as the exploitation of Arm’s vector extensions and performance libraries.

Unlike other similar deals, both Meta and Arm are maintaining their individual ownership stakes and physical infrastructure.

The decision to partner with Arm comes as Meta plans a vast expansion of its environmentally controversial multi-gigawatt AI data centers, with its first center – called “Prometheus” – slated for 2026. The company is also building “Hyperion,” a data center campus the company has said will be able to scale up to 5GW of computation power over several years.

After a brief hiring freeze, Meta has also begun investing millions in AI talent, recently hiring Andrew Tulloch, the co-founder of AI research and product company Thinking Machines Lab.