Unitree's $50 Billion Shanghai IPO and China's Humanoid Robotics Boom

Unitree's August 19 debut on Shanghai's STAR Market turned into a retail feeding frenzy. Priced at 150.8 yuan a share, the stock briefly spiked as much as 620-630% intraday before settling into a 460% gain, closing its first session at 845 yuan ^[1]. That pushed Unitree's market value to roughly $50 billion by the close of trading ^[2].

That price tag sits on top of 2025 revenue of just 1.70-1.71 billion yuan (roughly $240 million) and net profit of 278 million yuan ^[1]- real numbers by humanoid-robotics standards, but nowhere close to justifying a valuation running into the hundreds of times trailing sales on their own.

The demand behind that price was almost entirely retail. The IPO raised about $904 million by selling roughly 40.4 million shares - about 10% of the company - and individual investors submitted an estimated 8.1 trillion yuan (about $1.2 trillion) in orders ^[3], a STAR Market record confirmed at 5,526 times subscribed ^[4]. The froth cooled almost as fast as it built: Unitree shares fell nearly 19% the very next trading day, giving back a meaningful chunk of the debut-day pop ^[5].

Nine months before Unitree's Shanghai debut, China's National Development and Reform Commission - the country's top economic planning agency - issued a rare public warning that the humanoid robotics sector was showing classic frontier-industry bubble symptoms. Spokeswoman Li Chao noted that frontier industries have long grappled with balancing growth speed against bubble risk, and that more than 150 humanoid robot makers were already operating in China, with the number still rising and threatening a flood of highly similar models ^[6].

Goldman Sachs raised a related but distinct concern from the investment-banking side: that China's humanoid robot sector risks overcapacity as production scales up quickly without matching real orders ^[5]. The unusual part of this story isn't that skeptics exist - it's that the warning came from inside China's own economic-planning apparatus, months ahead of the wave of retail enthusiasm that would eventually send Unitree up 460% in a single session.

What separates Unitree from the rest of the field is that its numbers are directionally real. 2025 revenue rose to about 1.7 billion yuan from 392.77 million yuan in 2024 - roughly 335% growth - alongside a 278 million yuan net profit ^[1], and founder Wang Xingxing, who built the company from its 2016 founding and retains about a third of it post-IPO, assembled that revenue base years before the IPO frenzy arrived ^[7]. The IPO itself drew backing from Chinese AI company DeepSeek, which invested about 140.8 million yuan, on top of earlier stakes from Tencent and Meituan ^[8]- a sign of how tightly China's AI-model makers and its robotics hardware boom are now intertwined.

Mech-Mind Robotics, the industrial 3D-camera and AI-perception maker chasing its own roughly $300 million Hong Kong listing, tells a different story. Its 2025 adjusted net loss was 109 million yuan - improved from a 214 million yuan loss the year before, but still red ink - and it's expected to stay unprofitable through 2026 as it keeps spending on R&D and sales ^[9]. Independent analyst Rinat Mirzaitov of Humanoid Analytics frames Unitree's position bluntly: it is "one of the few companies in humanoid robotics with genuine revenue, profitability and meaningful production volume" ^[10]- the open question is whether it can convert that hardware and manufacturing lead into leadership on the application and workflow side, where the durable margins are likely to sit.

The same week Unitree went public, the 2026 World Robot Conference opened in Beijing's E-Town district, showcasing roughly 3,000 products and setting up the World Humanoid Robot Games, where more than 2,000 robots were set to compete in athletic events starting August 22 ^[11]. The showcase didn't go entirely smoothly - several humanoid robots publicly malfunctioned during opening-day demonstrations, a reminder that the industry's marketing cadence is still outrunning its reliability ^[12].

China shipped more than 40,000 humanoid robots in the first half of 2026, representing 97% of global shipments ^[13]- the deployment-at-scale story the government wants told. But community reaction split along familiar lines. On X, the highest-engagement commentary on the conference noted that despite the 'World' branding, the exhibition floor was overwhelmingly Chinese with almost no foreign firms present, a gap tied to US export controls that makes the ecosystem look increasingly self-contained rather than globally integrated. On Reddit, the skepticism ran more toward the robots themselves - commenters questioned how much of the conference-floor choreography is remote-controlled or pre-scripted rather than autonomous, and noted that a large share of Unitree's units still ship to universities and research labs rather than paying commercial customers, a distinction that matters for how much of that headline shipment figure has actually translated into industrial deployment rather than research demos.