Who Is Zhao Tongyang? The Showman Behind EngineAI

In December 2025, a Chinese humanoid robot kicked a man across a studio floor, and the internet couldn't look away — more than 100 million views in 24 hours. The man on the receiving end of the 75-kilo flying kick wasn't a stuntman. He was the robot company's own founder and CEO, Zhao Tongyang (赵同阳), who had strapped on padding and offered himself up as a crash-test dummy to prove the robot was real and not, as critics had claimed, CGI.

That single act tells you almost everything about the man behind EngineAI's T-800. In an industry full of soft-spoken PhDs, Zhao is the showman — a self-taught, vocational-college "grassroots" founder who treats a viral video as a weapon and once called a robot's front flip "worth 200 million yuan." So let's get to know him: where he came from, why he built XPeng's first humanoid before walking away to compete with it, and whether the substance behind EngineAI matches the spectacle.

First, the short answer: who is Zhao Tongyang?

Zhao Tongyang is the founder and CEO of EngineAI, one of the noisiest companies in China's humanoid-robot boom. What makes him stand out from the other founders we've profiled — Unitree's Wang Xingxing, AgiBot's Peng Zhihui, XPeng's He Xiaopeng — is that he is the least credentialed of all of them. Chinese media repeatedly call him a 大专生, a vocational-college graduate, in a field crowded with Tsinghua and Peking University doctorates. He's the underdog who got here the hard way.

A note on the record, because honesty matters in a profile: a lot of the usual biographical basics about Zhao are genuinely murky. No reliable source confirms his birth year or birthplace — English reporting only places him "in his mid-30s." Even his university is unverified; Chinese deep-dives note that the school he's often said to have attended doesn't exist under the name given. What is well documented is everything he's built — and that's a more interesting story anyway.

The grassroots founder

Zhao's path into robotics was long and littered with failures. He first made money in 2012 founding the Shenzhen IoT components firm Ai-Thinker (安信可), where he reportedly drove the price of WiFi modules down from around 40 yuan to 10 — an early taste of the cost-cutting instinct that would later define his robots. Then he chased the harder dream. By his own telling in Chinese profiles, the loss of a family member is what set him on the path of building robots, and he poured his IoT profits into two humanoid startups in 2016 and 2018. Both failed for lack of funding.

In late 2019 he tried again — pivoting from two-legged robots to four-legged ones with Dogotix (多够机器人), selling robot dogs for around 18,000 yuan, less than a tenth of what a Boston Dynamics machine cost. It was classic Zhao: undercut everyone, ship something real, make noise. And this time, someone important was watching.

Zhao Tongyang's whole career is the same move on repeat: walk into a field everyone says is too hard, arrive with no pedigree and no money, build something cheaper than the incumbents think possible, and make sure the world sees it. EngineAI is that instinct, finally backed by real capital.

Building XPeng's first humanoid — then leaving

In December 2020, the electric-car maker XPeng absorbed Dogotix's team, and its CEO He Xiaopeng set up a robotics joint venture, Pengxing Intelligent (鹏行智能), with Zhao as general manager. This is the part of his story most English coverage misses: Zhao didn't come from nowhere — he built XPeng's robotics program from the ground up. He led a team that grew to 300-plus people and developed XPeng's robot "horse" and its first humanoid prototype, the PX5, unveiled at XPeng's tech day in late 2023.

And then, like the engineer-founder he is, he left to do it himself. In October 2023 — taking talent and know-how with him — Zhao registered EngineAI in Shenzhen. The man who built the EV giant's humanoid now had his own company to point at it. It's a useful reminder that today's humanoid "twin stars," Unitree and AgiBot, sit atop a much deeper bench of founders, many of whom trained inside the carmakers we covered in why every carmaker is building a humanoid.

EngineAI and the viral playbook

EngineAI moved fast and loud. Its lineup quickly spanned the budget SA01 bipedal research robot (around 38,500 yuan, ~$5,400), the full-size SE01 flagship (170 cm, 55 kg, a remarkably natural neural-network walk that debuted on CCTV), the compact PM01 (138 cm, with a 320-degree rotating waist), and the heavy-duty T800. But hardware specs aren't what made EngineAI famous. Stunts did.

In February 2025, EngineAI posted a video of the little PM01 performing what it billed as the world's first front flip by a humanoid robot — a genuinely harder feat than the backflips Boston Dynamics' Atlas and Unitree had already done, because a front flip demands you land blind. It racked up more than 10 million views in three days and a nod from state media. The same robot went on to dance the "Kung Fu Hustle" axe-gang routine alongside US streamer IShowSpeed and to complete the world's first humanoid half-marathon. Zhao reportedly summed up the flip's real value bluntly: it was, he said, a front flip "worth 200 million yuan" — the amount it helped him raise.

Then came the T800, a 1.73 m, 75 kg flagship whose name is a deliberate wink at the Terminator. When EngineAI's dark, cinematic launch video of the T800 throwing flying kicks drew accusations of being CGI, Zhao's answer was pure showman: he had the robot kick him to the ground, on camera, from multiple angles. He's open about the strategy, telling one interviewer that it is "necessary to create a certain amount of panic. Otherwise, they don't realize how valuable it is." He pairs that with a surprisingly soft philosophy — that robots should have "a soul," and that "AI will never feel a gust of wind or smell the fragrance of flowers."

The price war and the money

The stunts are in service of a hard-nosed commercial strategy: be the cheapest. EngineAI helped ignite China's humanoid price war by launching the PM01 at 88,000 yuan, deliberately undercutting Unitree's G1 at 99,000. That aggression — plus the viral fame — turned a near-bankrupt founder into a fundraising magnet. EngineAI raised nearly a billion yuan across 2025 rounds led by XPeng-affiliated Rockets Capital and JD.com, then closed a $200 million Series B in April 2026 that pushed its valuation above 10 billion yuan (~$1.5 billion), co-led by a fund tied to Henan Investment Group and Apple supplier Luxshare Precision. In June 2026 it filed confidentially for a Hong Kong IPO, advised by CICC and CITIC Securities.

To back the orders, EngineAI opened a 12,000-square-metre Shenzhen factory around June 2026 and began delivering its first batch of T800s, claiming a line that can finish one humanoid every 15 minutes. It's the same demo-to-factory test we apply to everyone in how you actually judge a humanoid robot.

Is any of it real? The honest verdict

Here's where the showmanship cuts both ways, and where a review site has to be straight with you. For all the noise, EngineAI is not a volume leader. It reportedly shipped only around 500 humanoids in 2025 — respectable for a young company, but a fraction of Unitree's ~5,500 or AgiBot's ~5,168. The viral T800 combat clips were accused of being faked for a reason: like much of this industry, EngineAI's robots still operate at a low level of genuine autonomy, and even its own people gave conflicting accounts of whether the famous kick was a programmed move or a glitch. The kind of stage-managed demo we flagged at ICRA 2026 is exactly what skeptics see here.

To his credit, Zhao is unusually candid about the hype — including his own industry's. He's warned that any humanoid company without 5 billion yuan in the bank "needs to be very careful," and predicts a shakeout that leaves only five to ten Chinese survivors, with the top two taking ~80% of the market — the same brutal winnowing we argued is coming in our piece on the robot bubble. He's also combative: he publicly clashed with a prominent VC who was "batch exiting" humanoid startups, accusing him of "using the present to deny the future," and has openly disagreed with Unitree's Wang Xingxing about whether humanoid hardware is really "ready" yet.

What he gets right — and the risks

Strip away the kicks and the Terminator branding, and the case for Zhao Tongyang — and the worry about him — both come into focus:

• He understands attention. In a field where most demos look identical, Zhao's stunts have bought EngineAI a global profile — and the funding that comes with it — far out of proportion to its size.
• Grassroots hunger and cost obsession. A founder who survived two failed startups and priced WiFi modules to the bone brings a ruthlessness about cost that the price war rewards.
• Real pedigree underneath. He built XPeng's first humanoid; this is not a marketer who wandered into robotics, but an engineer-operator who can ship hardware.
• The risk: spectacle outrunning substance. EngineAI trails the leaders on the metric that ultimately matters — robots doing real, sustained work — and a brand built on viral stunts is vulnerable the moment the stunts stop landing or a "fake video" accusation actually sticks. By Zhao's own math, most companies like his won't survive the decade.

The quick facts on Zhao Tongyang

Frequently Asked Questions