Who Is Wang He? The Peking University Professor Behind Galbot

Most of the humanoid story is told in walking demos — robots taking careful steps across a stage while a crowd films. Wang He (王鹤) is quietly betting that none of that is the point. His robots don't walk; they roll. And while the rest of the field argues about bipedal gaits, his are already standing in real Beijing pharmacies, picking your medicine off the shelf at three in the morning.

Wang is a Peking University professor who co-founded Galbot (银河通用) and serves as its chief technology officer — not its CEO — and the company he helped start has a deliberately unglamorous thesis: build robots that actually do real work.

Wang He: the 30-second version

Wang He is the co-founder and CTO of Galbot, a Beijing embodied-AI company building humanoid robots that run on its own multimodal models. He is also a working academic — a Peking University professor — and that scientist's instinct shapes the company: Galbot is less interested in the spectacle of a robot walking than in whether a robot can finish a real job, reliably, without a human holding a controller.

From Stanford to a Peking University lab

Wang earned his bachelor's at Tsinghua University in 2014 and his PhD at Stanford in 2021, advised by the computer-graphics and geometry authority Leonidas Guibas. He returned to China as a tenure-track assistant professor at Peking University's Center on Frontiers of Computing Studies, where he founded and directs the EPIC Lab, working on embodied AI and the multimodal large models meant to give robots general-purpose skills.

In May 2023 he co-founded Galbot in Beijing's Haidian District with Yao Tengzhou (姚腾洲), taking the role of founder and CTO rather than CEO — keeping, in other words, one hand on the technology and one foot in the university.

Galbot's pitch is a rebuke to the walking-demo era: a robot that pours a perfect cup of coffee on stage is worthless if it can't restock a shelf at 3 a.m. Wang He's company chose the shelf.

The robot that stocks the shelves

The flagship Galbot G1 doesn't look like a Tesla Optimus or a rope-driven Astribot. It's a wheeled, dual-arm mobile humanoid: instead of legs, it rides a wheeled-leg base that moves in any direction, drops into a kneeling mode to reach the floor and rises into a standing mode to reach objects above two metres. It stands about 173 cm tall, with a vertical working range of about 0 to 240 cm.

What matters more than the chassis is how it's driven. Galbot says the G1's retail picking runs on a self-developed end-to-end vision-language-action model — fully autonomous, with no teleoperation, leaning heavily on synthetic simulation data to generalize. And it is genuinely deployed, not just demoed: around ten Galbot G1-staffed pharmacies were reported in regular operation in Beijing during 2025, including an unmanned 24-hour smart-pharmacy solution built with Meituan Pharmacy, where the robot picks ordered medicines off the shelf, packages them and handles replenishment. (The G1's published weight, payload and joint counts vary between sources, so we'll leave those numbers out — a good habit when judging any humanoid.)

Galbot S1 and the CATL line

Galbot has since moved into heavy industry. In January 2026 it announced the Galbot S1, an industrial-grade robot with a continuous dual-arm payload of up to 50 kg, reported as operating on the production line at battery giant CATL (宁德时代) — which is also one of the company's investors.

The money

Capital has poured in. Galbot raised a roughly RMB 700 million angel round in June 2024 (with Meituan, Qiming, IDG and others), an RMB 1.1 billion round led by CATL in June 2025, a round of over US$300 million at about a US$3 billion valuation in December 2025 (reportedly taking cumulative funding to roughly US$800 million), and a further RMB 2.5 billion round in March 2026 backed by China's National AI Industry Investment Fund and Sinopec. Galbot describes itself as China's highest-valued unlisted humanoid company — a label Caixin reports as the company's own claim rather than independently confirming — so treat the ranking as Galbot's own. It is still private, though Caixin reported in March 2026 that it is preparing a Hong Kong IPO, and the valuations sit squarely in the froth we keep flagging.

Where Galbot is ahead, and where it's exposed

• Real deployments beat demos. Autonomous pharmacy picking in unmanned 24-hour stores is a narrow, checkable claim — more convincing than a stage performance.
• The wheeled bet is pragmatic. Skipping legs for a wheeled base trades the "humanoid" narrative for something that works in a shop today.
• The superlatives are PR. Per-store throughput figures, store and city counts, and the "highest-valued unlisted" label are company-supplied claims — worth attributing, not repeating.
• The form factor is also the risk. A wheeled robot may cede the long-term humanoid story to the bipedal players, and a ~US$3 billion private valuation is built largely on promise.

The Wang He fact sheet