Meta Builds an AI‑Hardware Team
Meta has created a new AI hardware division and hired veteran engineer Xu Rui to lead development of AI devices inside its superintelligence unit, signalling the company is moving from research toward productised AI form factors. The move reinforces a broader industry push to pair specialized hardware with new interfaces rather than only scaling cloud models. (thebridgechronicle.com)
Meta has built plenty of hardware before. Quest headsets, Ray-Ban smart glasses, and the still-experimental Orion AR glasses all came out of Reality Labs. What changed this week is where the next devices will be born. Meta has created a separate AI hardware team inside Meta Superintelligence Labs, the company’s top AI unit, and hired veteran engineer Rui Xu to lead it. The move was first reported by Business Insider and quickly echoed across other outlets on April 4 and April 5. It places device building much closer to the people trying to build Meta’s next generation of AI systems. (thebridgechronicle.com) That matters because Meta’s AI group is not some side project. Mark Zuckerberg formally created Meta Superintelligence Labs on June 30, 2025, putting former Scale AI chief Alexandr Wang at the center of the company’s AI reorganization. The lab was supposed to unify Meta’s work on models, products, and long-term AI research. A hardware team inside that structure suggests Meta no longer sees devices as a downstream packaging exercise. It wants hardware to shape the product from the start. (cnbc.com) Xu’s background explains why Meta picked him. Recent reports describe him as a hardware operator who has worked across AI devices, robotics, and consumer electronics. Before this hire, he led hardware at Dreamer, an AI agent startup whose team Meta acqui-hired in late March 2026. Earlier roles tied him to ByteDance smart-device work, robotics startup K-Scale, and senior posts at Xiaomi and Lenovo. That is not the résumé of someone brought in to tweak an existing headset. It is the résumé of someone asked to turn vague AI ambitions into a thing people can wear, hold, or keep nearby all day. (scmp.com) The timing is the real story. Meta already has proof that AI hardware can escape the demo stage. Reuters reported on March 31 that the company expanded its Ray-Ban lineup again, after those glasses became one of the few clear commercial wins in the scramble for AI gadgets. At the same time, Meta is still pushing toward more ambitious form factors. In September 2024, it unveiled Orion, a prototype AR system that showed how far it wants to push glasses as a computing platform. A new team inside Superintelligence Labs sits between those two poles: today’s sellable smart glasses and tomorrow’s more radical AI-native devices. (msn.com) That shift also says something blunt about the industry. For the past two years, tech companies acted as if better models alone would define the AI race. Now they are rediscovering an older truth from the smartphone era: the interface decides what people actually use. Meta is hardly alone in chasing that idea, but it has a reason to move faster than most. It already owns a social graph, a family of apps, a growing glasses business, and a lab built to chase “superintelligence.” If it can fuse those pieces into a device category that feels natural, it gets a new front door to its AI. If it cannot, it is left with models that still reach people through someone else’s screen. (businessinsider.com) So this is less about one executive hire than about where Meta thinks the next battle will happen. The company is not just trying to make smarter software. It is trying to decide what shape that software should take when it leaves the cloud and enters daily life. This new team sits inside the unit created to build Meta’s most advanced AI, and it is being led by an engineer whose recent work spans agents, robots, and consumer devices. That is a very specific combination. (scmp.com)