Unitree popped 629% on its Shanghai debut. By September, half that gain was gone -- and so was its profit growth.

Unitree, the Hangzhou-based humanoid- and quadruped-robot maker, priced its Shanghai STAR Market initial public offering at ¥150.80 a share on August 6, 2026 -- valuing the company just above ¥61 billion (about $9 billion) before a single share traded. The retail portion of the deal drew the largest oversubscription in STAR Market history: more than 8,000 times the shares on offer, leaving ordinary retail applicants roughly a 0.018% chance of an allocation. When trading opened August 19, shares immediately spiked to ¥1,100 -- 629% above the IPO price -- before drifting down to close the day at ¥845, a 460% gain. Both figures were reported accurately by different outlets measuring different moments of the same session, and that gap between a number's peak and its close turns out to be the pattern for the whole story that follows.

The listing itself moved at a pace regulators don't usually allow. Unitree's IPO application was accepted by the Shanghai Stock Exchange on March 20, 2026, and approved by the listing committee on June 1 -- 73 days, a record for the STAR Market's review process. The debut was also timed, whether by design or coincidence, to land during the World Robot Conference in Beijing, the industry's highest-visibility annual showcase in China. A fast-tracked review and a conference-week debut aren't evidence of anything improper on their own, but they're both facts about how badly Beijing's capital markets wanted this listing to succeed, on top of whatever ordinary investors wanted.

The debut landed three weeks after Washington had already made Unitree's largest overseas market more complicated to reach. On July 29, the FCC's Public Safety and Homeland Security Bureau added foreign-made humanoid and quadruped robots to its Covered List, blocking new US import authorizations for models not already cleared -- a national-security order that named the kind of hardware Unitree builds, alongside a separate Pentagon designation alleging the company has links to China's military. The rule isn't retroactive: Unitree's currently shipping models had already secured US authorization and can keep selling. What the order actually blocks is Unitree's *next* generation of hardware from reaching the US specifically -- 18% of Unitree's 2025 revenue by Fortune's count, within a broader 45% overseas share that spans every market outside China.

By September 2 -- just 11 trading days after the debut -- the stock had fallen below ¥550, erasing more than ¥220 billion in market value from its opening-day peak. It closed September 8 at ¥523.68, down roughly 52% from the ¥1,100 debut peak but still up about 247% from the ¥150.80 IPO price -- a reminder that 'crash' and 're-rating' can both be true descriptions of the same chart, depending which starting point is chosen.

The financial picture behind the stock is genuinely two different stories depending which period is measured. Full-year 2025 was a breakout: revenue grew 335% to ¥1.708 billion, and Unitree's preferred profit figure -- net profit excluding non-recurring items -- grew 674% to ¥600 million. That excluded-items framing matters: Unitree's plain statutory net profit for the same year, the number after non-recurring gains and losses are left in rather than stripped out, was ¥278 million, less than half the figure the company and most coverage actually lead with. Both are real, audited numbers; they answer different questions, and a reader comparing this year's press coverage against last year's needs to know which one is being quoted.

2026 broke that streak. Adjusted net profit fell 53% year-over-year in Q1, to just ¥40 million. The first-half figure -- Q1 and Q2 combined -- came in less severe: revenue up 48.5% to ¥1.152 billion, but adjusted profit still down 19.3% to ¥244 million (net income attributable to shareholders, a related but separate line, was ¥274 million for the half). Read the two quarters together and Q2 clearly clawed back some of Q1's damage -- but two straight quarters of shrinking adjusted profit against accelerating revenue is a margin story, not a rounding error, and it's the number the IPO prospectus's growth narrative didn't have to reckon with yet.

"The surge in shipments for robot makers could be illusionary." -- HSBC analyst note, mid-July 2026, on humanoid-robot order volumes ahead of Unitree's listing

Unitree is, for what it's worth, one of the only companies in this field that is profitable at all. Ubtech, its closest listed domestic rival, reported 2025 revenue of ¥2 billion against a ¥790 million loss -- higher revenue than Unitree, and a nine-figure loss where Unitree posted a nine-figure profit. Nomura's own post-debut note credited Unitree's "rapid product iteration and continuous innovation" as the reason it, and not the sector broadly, earned the premium. (A Morningstar analyst separately flagged the US import rule as the more durable risk to Unitree specifically, since its overseas sales strategy leaned on exactly the market the FCC order now complicates.)

The mechanics of who actually captured the debut-day pop are also worth separating from the headline gain. With retail investors facing roughly a 0.018% allocation chance in an offering oversubscribed more than 8,000 times, the shares that caught the ¥1,100 opening spike were overwhelmingly institutional -- retail lottery winners who did get an allocation were still buying into a stock that had already run up before most of them could act on the pop at all. The 460%/629% and $50B/$66B figures describe the stock, not the typical investor's actual return, and conflating the two is a shorthand this coverage cycle has repeated without noting it.

None of this makes Unitree's business fake, and the FCC order alone doesn't explain the stock's retreat -- the profit deceleration started before the ban's practical effects could show up in a filing, and the stock gave back its peak faster than any single piece of news can fully account for. What the last six weeks actually settled is narrower and more useful than a verdict on the whole humanoid-robot sector: Unitree entered its public life as the industry's clearest profitable company, priced at multiples that assumed the profit line would keep accelerating the way it did in 2025, and its first two quarters as a public company instead showed that line decelerating. The stock's math had to catch up to that, and by September 8, it largely had.