Mumbai: Meta is going all-in on humanoid robots (Meta Humanoid Robots – Metabots), and this isn’t just another gadget bet. According to CTO Andrew Bosworth, the company is funneling billions into what he calls an “AR-size bet”, but this time, the hardware isn’t the challenge. The real bottleneck? Software.
The plan: combine AI, dexterity, and ambition to make robots smart enough to move like humans but without breaking a sweat over hardware hurdles.
Meta Humanoid Robots: Software Over Hardware
During a sit-down at Meta HQ, Bosworth didn’t mince words. “I’m not saying the hardware isn’t also hard, but it’s not the bottleneck. The bottleneck is the software,” he said. He demonstrated the challenge by picking up a glass of water, something humans do without thinking, but robots routinely fail at, often crushing or spilling objects.
Dexterous manipulation is the Holy Grail. Robots can walk, run, and even flip, because the ground is stable. Handling delicate objects? That’s a different ballgame. Meta’s software team, collaborating with the Superintelligence AI lab, is trying to create a “world model” that can simulate this precision.
Licensing, Not Manufacturing
Meta’s approach mirrors Google’s software licensing model. The company isn’t trying to churn out every Metabot. Instead, Bosworth plans to license the software platform to other robot manufacturers. “I don’t care about us being the hardware manufacturers,” he said. Companies can use the Meta blueprint as long as their robots meet certain specs.
It opens the door to scale without the production headaches, and keeps Meta focused on what it does best: AI and software.
Talent-First Strategy
Bosworth highlighted the high-profile team behind the robots. Marc Whitten, ex-Cruise CEO, is leading the robotics division. MIT’s Sangbae Kim joined earlier this year, hailed as “the greatest tactical roboticist in the game.” Jinsong Yu, who built the Orion AR glasses prototype, was coaxed out of retirement. Ning Li, a 15-year Meta veteran, now leads engineering.
Bosworth made it clear: the story isn’t just the robots, it’s the talent. In robotics, brains matter as much as bolts.
Tesla vs. Meta: Two Approaches
Bosworth doesn’t buy the Elon Musk approach to humanoids. Tesla relies on vision and mountains of data from human tasks. Meta, in contrast, is building data sets from scratch, simulating real-world dexterity before letting robots loose. “I can’t figure out how they’re going to get robotic data,” Bosworth quipped.
Meta isn’t chasing humanoids that mimic every human movement. “I don’t think you need 23 degrees of freedom in your hand. Two thumbs would be nice. I only need two thumbs.” Pragmatic, not theatrical.
Challenges in Humanoid Robotics
Meta’s vision is audacious, but not naive. The company is aware of silicon-level bottlenecks. Nvidia, Qualcomm, and other players are in the mix. The robots themselves will never be simple. Walking, balancing, and manipulating objects all require precise software that’s still under development.
Meanwhile, famous roboticist Rodney Brooks isn’t impressed. He warns investors that humanoid robot hype is overblown, calling methods like Tesla’s “pure fantasy thinking.” Human hands have about 17,000 touch receptors, a benchmark that robots are nowhere near hitting. Safety is another problem: larger robots store enormous kinetic energy. One misstep, and a humanoid can become dangerous.
Brooks predicts practical humanoid robots will likely ditch the human form altogether. Wheels, multiple arms, specialized sensors, that’s the future.
Why Meta Might Succeed
Meta isn’t chasing theatrics. The company is leveraging AI, talent, and simulation. By building world models and software-first platforms, it could bypass many pitfalls other humanoid projects face. Bosworth’s approach is grounded: it’s about usable dexterity, not perfect imitation.
In India, robotics startups are experimenting with automation in logistics and healthcare. Meta’s platform could eventually become a software backbone for companies globally, including those in Asia. The potential: billions of dollars, multiplied by global adoption.
The Takeaway
Meta humanoid robots are more than a headline. They’re a statement: software drives the future of robotics. Hardware is important, but brains come first. Talent, AI, and data-centric design are what could make these Metabots more than just walking prototypes.
And while critics scoff, Meta’s approach is pragmatic and scalable. If the company nails this, it won’t just license software; it might set the standard for the next generation of intelligent machines.
Also Read: Meta Connect 2025: Hypernova Glasses and a Bold AI Gamble
