The lab is the new garage: how Tsinghua doctoral students became robotics founders

The garage story has a Chinese revision. The company starts in a university laboratory, the founder is a doctoral student with a supervisor’s blessing and a publication record, and the first cheque arrives before the first product. Between early 2025 and the summer of 2026, 36Kr and others recorded a cluster of such cases at Tsinghua University. Read together, they describe a pipeline more than a fashion.

Four founders, one template

Min Yuheng, born in 2000, joined the Tsinghua AI & Robot Laboratory as a master’s student in April 2024. He released a humanoid robot in October and registered Zero Power Robotics in January 2025, according to a June 2026 profile by the Chinese outlet Zimubang. Qin Shentao, born in 2001, was in the third year of a doctorate at the School of Vehicle and Mobility when he registered OriginFlow in August 2025. Gasgoo reports more than 500 million yuan raised by May 2026. Hu Yucheng, also born in 2001, resigned from ByteDance’s Seed unit in April 2026 and announced a seed round of several hundred million yuan in August for LatentVerse, a spin-out of his institute’s embodied-intelligence lab, per 36Kr and Gasgoo.

A fourth, Liu Songming, completes the set. Zimubang describes a doctoral student in Tsinghua’s ISR Lab who founded Jiangxian Technology and released a dexterous-hand manipulation model within four months, then took 500 million yuan in angel funding about six months after founding. VnExpress, citing Securities Times, writes about a Tsinghua doctoral student of the same name who founded a physical-AI company called LiberAI, with more than 1 billion yuan raised across six rounds in ten months. The two accounts agree on founder and field; they may be the same company under different English names, but neither outlet says so.

The template is consistent. A young founder with a strong research record. A lab that supplies a team and a technical direction. Investors who write early cheques of tens or hundreds of millions of yuan. And, in the case of Hu, a supervisor, Chen Jianyu, who founded a robot maker, Robot Era, that then invested in the student’s company.

Why robotics, and why now

The robotics field suits students for a plain reason. A humanoid or a manipulation model is still a research problem, which means the people who write the papers are the people who can build the product. Zimubang quotes the view of Wang Xingxing, founder of Unitree Robotics, that emerging fields offer students low maturity and less competition. The cost of entry is set by compute and data rather than factories, and both can be bought with seed money.

Money has been willing. Zimubang’s report lists lead investors including Shunwei Capital and Meituan on one company and MiraclePlus, the Lu Qi incubator, on several smaller ones. At an investor roadshow for Tsinghua alumni in Shenzhen in August 2025, Ding Baoyu, chief investment officer at Tongchuang Weiye, said Tsinghua alumni had founded 53 hard-tech unicorns, more than 30 per cent of them in Shenzhen, and that four of China’s six leading large-model companies came out of the Tsinghua group.

An older machine underneath

None of this began in 2025. Tsinghua launched x-lab in April 2013 at its School of Economics and Management, co-built with 14 departments, as a platform for innovation and entrepreneurship teaching. By 2017 the university said it had received more than 1,000 project teams. VnExpress, citing Business Insider and the research firm Tracxn, reports that alumni have founded more than 1,000 companies, including over a dozen unicorns.

The pipeline also extends well beyond one campus. Zimubang’s list of founders includes a doctoral student at Shanghai Jiao Tong University who founded a flapping-wing robot company, a Northeastern University student making underwater robots, and Qin’s earlier years at the Harbin Institute of Technology. Several of them passed through programmes tied to two people: Li Zexiang, the professor whose students and institutes recur in these biographies and who, Zimubang says, incubated DJI, and Lu Qi’s MiraclePlus. Min studied in Li’s experimental class at Chongqing University; Qin entered the Shenzhen institute Li founded.

What to be careful about

The numbers here are loosely reconciled. Zerith’s funding is given in different totals by different outlets, and in a single month 36Kr’s own English pages spelled the company two ways. Investor lists vary. Round names are not standardised, and “hundreds of millions of yuan” is a range, not a figure. Almost none of the coverage reports revenue.

There is also a selection effect. The stories that travel are those of students who raised money, not of those who did not. Zimubang itself notes that Min’s first venture, a cleaning robot, failed. And the principal test, set out by Gasgoo of LatentVerse, applies across the group: whether the work can move beyond papers and demonstrations.

What the cases do establish is narrower, and more useful. In 2026 a Chinese doctoral student with a lab, a supervisor’s network and a credible research record can plausibly start a funded company in the middle of a degree. The institutional choice, to treat the laboratory as the front door of the start-up, is the thing to watch. The founders’ youth is the headline; the infrastructure behind them is the story.