Code doesn't lie. The balance sheet does.
Nomura just dropped its first coverage on Yuzhu Technology (Unitree Robotics) with a 'Buy' rating. The headline numbers are brutal in their precision: 63.2% gross margin on humanoid robots, 5,500+ units shipped in 2025 (global first), and a 2026–2028 revenue CAGR of 122%. This is not a drill.
Context: Why Now?
Yuzhu is a Chinese humanoid robot maker that has moved from lab toy to mass-deliverable product in 26 months, across four generations (H1, G1, R1, H2). The market is small—total humanoid shipments globally are under 30,000 units as of early 2025—but Yuzhu claims the top spot. Nomura's report is the first major institutional coverage, and it reads like a crypto token whitepaper: high margins, vertical integration, and a 'data flywheel' narrative.
Core: The Forensic Breakdown
Let's verify the numbers. Yuzhu's key advantage is hardware self-sufficiency. Only 10–20% of components are outsourced—likely including AI chips. This gives them a structural cost edge. Compare that to Figure AI or Tesla Optimus, which rely on external supply chains. The result: 63% gross margin on humanoids, far above typical consumer electronics (20–40%).
But the real story is the revenue acceleration. Nomura forecasts: - 2026 revenue: 26.87 billion yuan (+58% YoY) - 2027: 53.96 billion yuan (+101%) - 2028: 131.84 billion yuan (+144%)
A leap from 58% to 101% in one year is not natural. It implies a step-function event—a large industrial contract, a new product line, or a capacity expansion. The report does not name the catalyst. That's a red flag.
Based on my audit experience during the 2017 ICO boom, I've seen this pattern before. A company with a strong hardware story but a revenue projection that assumes a binary trigger. The question is: is the trigger real, or is it a narrative?
Contrarian: The Unreported Angles
The data doesn't care about your narrative. Here are three blind spots Nomura glossed over:
- Chinese competitors are missing from the analysis. The 'global first' claim may not hold if you include Zhiyuan Robotics, UBTECH, or Kepler. All three are racing to mass-produce humanoids. Nomura's report is written for international investors who may not know the domestic landscape. That's a selective omission.
- The revenue CAGR depends on industrial adoption, but current shipments are to research, education, and government. These are 'demo' buyers, not productivity buyers. The physical interaction data from a university lab is fundamentally different from a factory floor. If the data flywheel can't transfer to industrial skills, the 122% CAGR collapses.
- The 25x P/S valuation on 2027 revenue is a bet on the future, not the present. At 2027 revenue of 53.96 billion yuan, the implied market cap is ~1.35 trillion yuan (~$190 billion). That's higher than many mid-cap tech stocks. The premium is justified only if Yuzhu becomes the Tesla of humanoids—a huge assumption given the technology gap in manipulation and autonomy.
Takeaway: What to Watch
The next 6 months are critical. Yuzhu must show industrial repeat orders—not just government contracts. If the 2027 revenue acceleration materializes, it will be the signal. If not, the 25x P/S will look like a meme.
For crypto investors: this is a 'buy the transition, not the destination' play. The hardware edge is real, but the revenue narrative is untested. Watch the data, not the hype.