Meta Assembles Hardware Team Led by Xiaomi and ByteDance Veteran
Meta's Super Intelligence Lab is building a dedicated hardware team led by Xu Rui, a product executive with years of experience at Xiaomi and ByteDance. The move signals Meta is developing new AI devices beyond its existing smart glasses and VR headsets.
Some engineers from Meta's Reality Labs have transferred to the Super Intelligence Lab to support the effort. The exact specifications of the new hardware remain undisclosed.
Who is Xu Rui?
Xu Rui spent two decades working on consumer hardware across multiple continents. He started at LG Electronics, where he helped develop one of the world's first Linux-based mobile phones-five years before Android launched. After stints at Intel and Amazon Lab 126 (where he worked on Fire TV), he returned to China and held senior roles at Xiaomi, Lenovo, and ByteDance.
Most recently, Xu Rui was senior director of XR products at Tencent. He then joined K-Scale Labs, a Silicon Valley robotics startup, as chief operating officer, but the company shut down after running out of funding. He moved to Dreamer, an AI agent startup, as head of hardware. Meta acquired Dreamer last month, bringing Xu Rui into the company.
What Meta's Hardware Won't Be
Xu Rui has been explicit about what he thinks AI hardware should avoid. He dismisses phones with AI buttons-like ByteDance's DouBao Phone-as "button wizards" that crudely automate finger taps. He compares them to "nuclear-powered pencil sharpeners": using cutting-edge AI models to operate outdated mobile apps on a traditional smartphone screen.
His criticism hinges on a mismatch between eras. The PC era had "mouse and menu." Mobile introduced "touch and app." Each generation, he argues, needs hardware and software designed together from the ground up. AI era devices need their own native logic, not retrofitted intelligence.
The Philosophy Behind the Hardware
Xu Rui believes true AI hardware solves physical-world problems that existing devices cannot. He points to smart rice cookers as an example of wasted potential. A quality heating element can maintain temperatures between 30°C and 180°C-enough to make yogurt, cook sous-vide steak, or ferment dough. Yet product managers lock it into preset functions: rice, soup, porridge.
AI changes this equation. Instead of predetermined firmware, hardware becomes an extension of intelligent agents. The rice cooker finally makes yogurt because an AI agent understands what the user actually needs, not what engineers decided in advance.
Xu Rui calls this concept "dumb devices"-hardware with excellent intrinsic capabilities rather than embedded intelligence. The smarts come from AI agents, not the device itself.
Meta's Broader AI Agent Strategy
Alexandr Wang, who leads Meta's Super Intelligence Lab, said the lab is building personalized AI agents that operate across devices. These agents will follow users throughout their day, capable of "seeing what you see and hearing what you hear" while staying constantly connected.
Wang told a podcast audience in February: "In the coming months, you will witness us advancing all of this at an astonishing pace."
For product development professionals, Meta's move signals a shift in how companies think about AI hardware. Rather than bolting AI onto existing device categories, the focus is on designing hardware and intelligence as integrated systems from inception. Understanding this philosophy-and the constraints of retrofitting AI onto legacy platforms-will shape how teams approach their own AI product roadmaps.
Learn more about AI for Product Development and Generative AI and LLM to deepen your knowledge of these emerging strategies.
Your membership also unlocks:
