The ledger remembers what the hype forgets. While the market fixates on speculative AI chatbots and tokenized narratives, Nomura Securities just dropped a coverage initiation that reframes the entire conversation. On August 19, 2025, the Japanese bank slapped a "Buy" rating on Yuzhu Technology, a Chinese humanoid robot maker that has quietly shipped over 5,500 units—more than any other player globally. The report highlights a 63.2% gross margin on its humanoid segment, a 26-month product cycle with four generations, and a 122% revenue CAGR forecast through 2028. This isn't a crypto-native company, but its business model—vertical hardware integration, a data flywheel, and a community-driven deployment strategy—mirrors the core tenets of DePIN (Decentralized Physical Infrastructure Networks). As a crypto news editor who has tracked DeFi Summer, the NFT crash, and the AI convergence, I see Yuzhu as a case study in how physical hardware can create a defensible moat through transparency and iteration. But the hype is already pricing in a transition that remains unproven.
Context: The Nomura Catalyst Nomura's report is the first major institutional coverage of a humanoid robotics company outside of Tesla. The bank estimates Yuzhu will ship over 5,500 humanoid robots in 2025, ranking it first globally. Total revenue is projected to hit 26.87 billion yuan in 2026 ($3.7 billion), 53.96 billion in 2027, and 131.84 billion in 2028. The valuation anchor is 25x P/S on 2027 revenue, implying a market cap around 3300 billion yuan ($460 billion). That's a staggering number for a company that only started shipping humanoids in 2022. But the story is not just about numbers—it's about the architecture. Yuzhu designs its own motors, reducers, drivers, encoders, lidar, and power management. Only 10-20% of components are sourced externally, likely including AI chips like NVIDIA Jetson. This vertical integration gives it a cost structure that rivals Tesla's Optimus and Figure AI, but with a far faster iteration pace—four generations in 26 months, covering consumer (G1), research (H1), and industrial (R1, H2) segments.
Core: The Data Flywheel and the Hardware Edge During the 2017 ICO due diligence sprint, I learned that speed without verification is noise. Nomura's report is built on a rational thesis: low cost drives volume, volume generates real-world physical interaction data, data trains better models, and better models enable product iteration. This is the same flywheel Tesla uses for FSD, but applied to embodied intelligence. Yuzhu's 63.2% gross margin on humanoids is a war chest—it can afford to price aggressively while competitors burn cash. Figure AI, 1X, and even Tesla's Optimus are not yet profitable. Yuzhu is, and that changes the game. I recall the DeFi Summer of 2020 when I launched the "DeFi Decoded" column to bridge the gap between code and community. The same principle applies here: hardware is the code, and the community of researchers, educators, and early adopters forms the data-generation layer. The company's broad product line (consumer to industrial) is not just a market strategy—it's a data collection strategy. Each unit in the field contributes telemetry, manipulation logs, and environmental interaction data that no lab can simulate. This is the real network effect.
Contrarian: The Unreported Risks But the ledger remembers what the hype forgets. Nomura's 122% CAGR assumes a dramatic leap from experimental purchases to industrial repeat orders. The 2027 revenue acceleration (101% vs 58% in 2026) is a tell—it implies a major catalyst, such as a signed framework agreement with a manufacturer, that is not mentioned in the report. Without that, the growth curve is physically implausible. Furthermore, the report omits the competitive landscape among Chinese peers like Zhiyuan Robot (智元机器人) and UBTECH, both of which are ramping production. Yuzhu's "global first" claim may be based on a narrow definition of humanoid robots (excluding bipedal research platforms). The U.S. market accounts for 13.3% of 2025 revenue, and export controls on AI chips and lidar components could disrupt supply. During the 2022 bear market, I learned that stability comes from transparency. Nomura's report does not disclose whether it has a potential IPO underwriting relationship with Yuzhu—a common conflict of interest. The 25x P/S on 2027 revenue effectively prices in a future that may not materialize if industrial adoption stalls. Culture is the new collateral, but only if the community—investors, customers, and regulators—trust the data. Currently, the data on actual usage rates, return rates, and industrial deployment is missing.
Takeaway: Buy the Transition, Not the Destination Yuzhu Technology is a remarkable hardware company that has achieved what few in embodied AI have: profitability and volume. The Nomura initiation is a validation of the physical AI thesis, but the valuation already discounts a successful transition to industrial scale. As a crypto editor who has seen narratives move markets faster than blocks, I advise watching for the concrete signals: quarterly shipments broken down by segment, industrial customer announcements, and the company's ability to maintain margins as competition intensifies. The sprint ends, but the chain remains—and the chain here is the data flywheel. If it spins, Yuzhu could be the Tesla of humanoids. If it stalls, the hype will be a forgotten chapter in the ledger. Decentralization is a mindset, not just a metric—and this company's success depends on its ability to decentralize its data generation while keeping its hardware centralized. That tension is the real story. Bridging the gap between code and community remains the only sustainable path forward.