UBTech & Humanoid Investment Framework
UBTECH Overview
UBTECH is one of China’s earliest and most prominent humanoid-robot companies. Headquartered in Shenzhen — the country’s leading tech hub — it has built deep ties with industrial partners such as BYD, sharing a major venture investor and counting BYD among its largest pilot customers. UBTECH was also the first humanoid-robot firm to list on the Hong Kong Stock Exchange, and today trades at roughly US $6 billion in market value.
While its technology leadership is clear, UBTECH is still at an early stage of commercialization and is expected to remain loss-making for the next two to three years. Because of that, it does not yet fit tightly into our core public-market investment framework; we would view it instead as a venture-style holding.
That said, I am intentionally broadening my investing approach to gain exposure to a wider set of opportunities. I believe the long-term future of humanoids is compelling, and if UBTECH maintains its leadership over the next decade this could be a genuine “ten-bagger” opportunity.
1. Long-term Perspective: A Trillion-RMB Market with Wave-Like Growth
Humanoid robots have become an industry consensus — a “must-build, bound-to-succeed” trillion-RMB track. Like most emerging industries, growth will be wave-like: an initial boom driven by subsidies and policy support, followed by 30–50 % pullbacks or cooling sentiment. That’s normal and doesn’t change the long-term trend. Investors need to approach it with patience, give the technology time and space, and not fixate on short-term revenue. As long as a company is scaling fast, temporary losses are acceptable.
New-energy vehicles and solar PV were once accused of “subsidy fraud,” with many firms collapsing mid-cycle, yet they became global industries. Humanoid robots have similar industrial and government support behind them. If a truly competitive full-system manufacturer emerges, its scale and market cap could easily reach hundreds of billions of RMB within ten years — like BYD today. Cyclical swings can even give long-term investors a chance to buy low, sell high, and improve positioning.
Another difference now: governments, enterprises, and investors are committing to these “high-certainty” tracks with far more conviction and speed, taking five- to ten-year horizons — rare in the past. The hardest phase for humanoid robots is already over. For years the field felt like science fiction: Japan and the U.S. tried and stalled. Now Tesla leads, government directives repeatedly endorse the sector, and commercialization should accelerate. A PV-style crash (mass subsidy withdrawal, years of stagnation) looks less likely. It’s a race against time: whoever ships and deploys first wins a huge edge.
2. Rapid Economics
The commercial logic of humanoids is simple: “24/7 work, no social insurance or HR management” — essentially no hidden labor costs. At a unit price of RMB 100–150k, one robot equals roughly two years of a worker’s wages yet can run more than three years. Payback in two years, then pure incremental efficiency.
Combined with “dark factories” (lights-off, fully automated), ROI is especially high in labor-tight or high-cost industries like electronics assembly, logistics, and hazardous work — the core driver of large-scale adoption.
3. Why “Humanoid” Specifically
Buildings, factories, and cities are designed around the human form: door widths, stair heights, workbench dimensions, tools, transport, public facilities. Non-humanoid robots require environment redesign or special interfaces — costly and inflexible.
Humanoid robots fit existing infrastructure and coexist more naturally and safely with people — a key reason governments and big firms are betting on this format. Wheels or specialized grippers can still be used in niche settings to cut costs (e.g., no dexterous hands where not needed).
The most valuable humanoids will be “intelligent humanoids” — not only doing physical work but also orchestrating other robots and tasks, acting as a central hub (factory logistics coordinator, household “butler”). Compared with wheeled or modular bots, humanoids maximize value in natural environments and can take on higher-level decision and coordination roles.
4. Why Full-System Makers Beat Component Suppliers Long-Term
As with EVs and smartphones, value will shift from components to system integrators, software ecosystems, and brand premium. Four main reasons:
Component commoditization. Early on, dexterous hands, actuators, sensors, and batteries have breakthroughs and high margins. But as the supply chain matures, China’s cost advantage flattens gaps; price wars squeeze margins; high ROE is hard to sustain.
System integration & software. Humanoids need sensing, motion control, energy management, AI, and scenario adaptation to work together. True competitiveness lies in full-system control of hardware + software, rapid customization, and OTA/cloud updates — hard for component firms to replicate.
Brand, channels & customer lock-in. In B2B (factories, care, hazardous work) and future B2C (homes, assistants), the system maker owns the customer relationship, data, and service network. Component makers are easily replaced.
History. Smartphones: chips and screens had high margins first, but Apple (device + OS + ecosystem) captured long-term value. EVs: batteries were key early, but Tesla/BYD (vehicle + data) now dominate. Same logic for humanoids.
Thus while component stocks may enjoy high multiples in the short term, full-system makers are better placed to capture the eventual system, brand, and ecosystem premium. Picking one or two winners among dozens of component suppliers is tough — especially with cross-industry entrants like BYD making their own parts at lower cost. ETFs are also tricky: many listed names are overhyped or low-quality.
5. How Capital Decides Whom to Invest
Investors focus not on near-term profits but on who can truly “run out ahead.” Key factors:
Founding team & execution. Can they drive continuous breakthroughs, iteration, and mass production?
Credible tech roadmap. Software, system, and scenario design must be scalable.
First-mover & sustained lead. New entrants abound; maintaining the lead is hard, but winners become kings of the hill.
Capital backs companies that can become scarce “head” assets — like BYD or Cambricon in the early EV/AI chip days. Once a system maker leads in capacity, tech, and ecosystem, it can remain a long-term king despite followers.
Regional clusters to watch: Unitree in Hangzhou (“Hangzhou Four Little Dragons”), UBTECH in the Greater Bay Area, Zhiyuan in Shanghai. These hubs have complete supply chains, local government support, and dense capital — more likely to produce winners.
6. “Industrial First + Software Ecosystem” Route
I believe the correct rollout path is to commercialize first in industrial/high-value scenarios, then expand to services and households:
Paying power. Factories, logistics, and hazardous jobs have labor shortages and high costs; they’ll pay for efficiency, generating stable orders.
Physical data gap. Open-environment human-robot interaction standards are immature. Industrial settings provide safer, faster real-world data accumulation.
Consumer too early. All-purpose home humanoids are still more of a showcase; no clear price model or replicable scenes; weak short-term profitability.
What matters isn’t “who debuts at a trade show” but “who lands in real factories and keeps building hardware + software capability.” Long term, the real moat is software/platform — editability, cloud upgrades, open ecosystem, even partial open-source — like a “robot OS” with developers and apps. Components will platformize quickly and lose their moat.
7. UBTECH’s Edge: ROSA + Hive Mind
UBTECH started this layout earliest and most systematically. Its self-developed ROSA (Robot Operating System Agent) deeply decouples hardware from applications, giving all its robots a unified platform with shared architecture, APIs, and tools.
Core functions:
Motion control & navigation (gait/path planning, balance, obstacle avoidance)
Perception & interaction (voice, face/object recognition, SLAM)
Task scheduling & decision (multi-task, scenario adaptation)
Hardware abstraction (servo joints, motors, sensors)
Cloud & multi-device connectivity (remote upgrades, data analytics)
Modular architecture = scalability and efficiency, faster new products and lower R&D cost: “hardware + software + ecosystem.”
ROSA also supports an open ecosystem: partners can build apps on generic hardware. Together with OTA upgrades and “Hive Mind” (swarm intelligence), multiple robots share data/experience:
Experience sharing. One robot’s task experience uploads to others.
Task coordination. Complex jobs split among multiple robots.
Data accumulation. Sensor data uploaded, models trained in the cloud, pushed back — robots get smarter with use.
This “hardware + software + cloud” triad makes UBTECH not just a hardware vendor but a platform company. Once customers and developers lock in, switching costs soar and the moat deepens.
Unitree comparison. Unitree excels in cost control, agile leg mechanics, and quick small-batch delivery — currently the lower-price, faster-ship humanoid/quadruped vendor. But its software relies heavily on open-source ROS; its ecosystem is nascent, more hardware-driven, exploring industrial orders. UBTECH dug deeper into system + OS + ecosystem, aiming for an “Android/iOS for robots” — the closest in China to that goal.
8. What to Watch on UBTECH
Walker S3 launch — speed of improvement vs previous gen; timeline to reach worker-level efficiency (commercial inflection).
Tech breakthroughs expanding scenarios — when does it move to consumer market?
Scaling to break-even in 3–4 years — secondary but still a positive.
9. Market Critiques & My View
No Spring Festival Gala / low public exposure. Some think a humanoid missing high-profile events lacks “brand.” I see this as strategy: UBTECH leaves the stage to others, fosters “a hundred flowers,” avoids hype, and focuses on industrial landings. Real value lies in maturity and utility, not noise. TOC products naturally get more buzz but less maturity.
Shareholder selling. Market reads sales as bearish; I see it as short-term overhang cleared. Now UBTECH is shifting from “non-consensus” to “super-consensus”: the market is recognizing its full-system R&D moat and industrial potential. Early speculators have exited; now comes rational competition over tech delivery and capacity.
Limited media attention. Industrial focus means less TOC buzz, but that signals more mature hardware, software, ROSA platform, and mass-production ability. High exposure ≠ high value. Tech + deployment = long-term competitiveness.
10. Investment Takeaways
Prioritize full-system makers with integration, software, and scenario fit (UBTECH, Unitree, Zhiyuan).
Watch commoditization of components; true edge is in algorithms + system.
Track tech + rollout path (3–5 years to full commercialization; long-term R&D and business plan key).
Benchmark global valuations & sentiment but stay rational.
Policy & cluster advantage — favor firms with government support and strong local supply chains.
