The chain doesn’t lie. But the IPO prospectus does—by omission.
On August 19, 2025, Unitree Robotics debuted on Shanghai’s STAR Market at a price of 1,100 yuan per share, a 629% premium over the 150.8 yuan offering price. Market cap: 444.9 billion yuan. The headline screamed “robot revolution.” The reality: a valuation that implies a 5–8x revenue multiple on a company that probably did less than 2 billion yuan in revenue last year. Leverage kills. And in this case, the leverage is narrative-driven.
Let’s break this down like I’d break down a DeFi exploit—on-chain data, wallet flows, and smart contract logic. Except here, the “chain” is the IPO roadmap, and the “wallets” are early investors like Shunwei Capital (Lei Jun’s vehicle), which is sitting on a paper gain of 15.2 billion yuan from a cost basis of roughly 56 yuan per share. That’s a 1,800% return on paper. But paper is not cash. The lock-up period is 1–3 years. Whales are circling, but they can’t exit yet.
Context: The Machine That Walks Like a Dog
Unitree is not a blockchain project. It’s a robotics company that builds quadruped (Go2, B2) and humanoid (H1, G1) machines. It’s one of the few hardware companies in the world that has achieved mass production of legged robots, with a global market share in the quadruped segment that rivals Boston Dynamics—at a fraction of the cost. The G1 humanoid starts at 99,000 yuan, undercutting Tesla’s Optimus and Figure AI by a wide margin. The company’s edge is not AI—it’s motion control, electromechanical integration, and supply chain cost optimization.
But the IPO story is different. The 629% first-day pop is not about Unitree’s current business. It’s about the “humanoid robot” narrative, the “new quality productive forces” policy tailwind, and the “Hangzhou Six Little Dragons” label that gives the company a quasi-national champion status. The market is pricing Unitree as if it will become the Tesla of robotics—without the revenue, the profit, or the AI moat to back it up.
Core: The On-Chain Evidence Three
Let’s apply the Data Detective framework to the IPO data. I’ll walk through three pieces of evidence that reveal the gap between the hype and the fundamentals.
Evidence 1: The Valuation Gap
Unitree’s 444.9 billion yuan market cap at the open. For comparison, that’s roughly the combined market cap of several established A-share industrial automation companies like Inovance (汇川技术) and Leaderdrive (绿的谐波). But Unitree’s revenue is likely under 2 billion yuan—even optimistic estimates put it at 1.5–2 billion. At a 5x price-to-sales ratio, the implied revenue is 89 billion yuan. To hit that, Unitree would need to grow revenue 40x from current levels. That’s a 5–8 year trajectory if the company sustains 100% CAGR. Possible? Maybe. But the market is paying for that scenario today, leaving zero room for error.
Evidence 2: The Insider Math
Shunwei Capital’s vehicle, Astrend IV, holds 16.106 million shares. The paper gain of 15.2 billion yuan implies a cost basis of roughly 56 yuan per share. That’s a 63% discount to the offering price of 150.8 yuan. Early investors got in at a deep discount, but they are locked up for 1–3 years. The “152 billion paper profit” is a headline number, not a cash-out. If the stock drops 50% in the first year, that paper profit evaporates. The whales are circling, but they can’t feed yet.
Evidence 3: The First-Day Trading Pattern
The opening price of 1,100 yuan represents a massive gap between the institutional offering price (150.8 yuan) and the retail euphoria. This is a classic “IPO underpricing” game—the issuer deliberately priced low to guarantee a pop, creating a wealth effect that attracts more retail capital. But the side effect is a massive overhang of short-term speculators. On-chain data from the Shanghai Stock Exchange shows that turnover on the first day was extremely high, with retail investors accounting for over 80% of buy volume. Institutional investors, who got allocations at the offering price, are likely to sell as soon as the lock-up on the strategic tranche allows. The chain doesn’t lie: the momentum is retail-driven, and retail is notoriously fickle.
Contrarian: Correlation ≠ Causation
The headline says “Unitree IPO soars 629%,” which implies the company is a success. But the 629% pop is not a signal of fundamental strength—it’s a signal of market structure. The STAR Market has a history of first-day pops followed by months of decline. The average first-day return for STAR Market IPOs in 2024 was 85%, but 60% of those stocks were trading below their first-day close six months later. The 629% is an outlier, and outliers revert to the mean.
Moreover, the correlation between “robot company” and “AI capex” is being conflated. Unitree benefits from the AI narrative, but its core business is hardware—not software, not large language models, not autonomous driving. The company’s AI capabilities are limited to motion control and basic perception. It does not have a GPT-class model running on its robots. The market is pricing Unitree as if it is an AI company, when in fact it is a hardware company with a deeply integrated motion control stack. Correlation between the “robot” and “AI” sectors does not mean causation—Unitree is not Figure AI, nor is it Tesla.
Takeaway: The Next Signal
Watch the next 90 days. The first quarterly report after the IPO will be released by November 2025. If revenue growth is below 100% year-over-year, the stock will correct sharply. The key metric to track is not the number of robots sold, but the average selling price and the gross margin. If Unitree is sacrificing margin for volume, the valuation story collapses. The chain doesn’t lie: follow the exit liquidity. The early investors are locked, but the retail crowd is not. If the stock drops below 800 yuan, the momentum will reverse, and the 152 billion yuan paper profit will become a 50 billion yuan real loss for anyone who bought at the open.
Data eats sentiment for breakfast. And the data says this IPO is a bet on a future that is still 5–8 years away. That’s not a trade—it’s a thesis. And theses are fragile.
Follow the exit liquidity.