Embodied Intelligence Hype vs. Reality: Zhishen Technology's 15,000-Unit Production and 'Lay Eggs Along the Way' Strategy at WRC 2026

Forum topic · 小凯 · 2026-08-26

Summary

At the 2026 World Robot Conference (WRC), Zhishen Technology co-founder Liu Yulong challenged the industry's 'demo bubble' with a strategy he calls 'build the legs first, then attach the working hands': focus on quadruped robots whose all-terrain mobility and spatial navigation are already mature, monetize them in real deployments, and defer humanoid manipulation intelligence. The company reports 15,000+ units cumulatively produced, monthly capacity of 5,000+ units, thousands of quadrupeds on security patrol duty, 100+ power inspection projects, and coverage of 100+ fire brigades. Its engineering metrics include the P85MAX-S joint module (1 kg driving 90 kg loads), the Gangbang L2 quadruped (7:1 payload-to-weight ratio, IP66 rating, submersion to 1 m), and the NE01 humanoid (180 N·m peak joint torque). Liu contrasts this 'lay eggs along the way' path-revenue rolling back into R&D-with competitors who 'wait for the perfect humanoid' while burning cash. He also flags the industry's lack of shared standards for performance metrics, evaluation, and certification as the biggest bottleneck. The company operates a 'chain leader + module output' model, supplying one million CHAMP joint modules annually and partnering with 300+ integrators.

> Source: Zhishen Technology co-founder Liu Yulong interview with China.com (Aug 26, 2026) + WRC 2026 coverage

One-Sentence Warning

While the whole industry shows off demos, Zhishen Technology brought a set of numbers every peer understands: 15,000+ units cumulatively produced, 5,000+ units monthly capacity, thousands of units deployed in security, 100+ power inspection projects delivered, 100+ fire brigades covered.

At WRC 2026, co-founder Liu Yulong played a completely different hand: "Build the legs that run first, then attach the hands that work." The company first focused on quadruped robots because all-terrain mobility and spatial navigation are already solved; once these modules prove out in real scenarios (security, emergency response, power), it returns to fill in manipulation intelligence for humanoids.

On the 2026 H2 embodied intelligence map, this forms a mirror image of the WHRG World Humanoid Robot Games: WHRG is embodied intelligence on stage (record-breaking, autonomous assembly), while Zhishen is embodied intelligence in the real world (24/7 community patrol, fire brigade solutions, routine substation inspection).

The 'Demo Bubble': 100,000 Units on Slides vs. 100 Units in Reality

At WRC 2026, the industry claims 400+ robot products and an annual output targeting 100,000 units. Yet Unitree chairman Wang Xingxing said bluntly: "Task success rates approach 100% in fixed scenarios, but plummet once objects or environments change."

This is the 'demo bubble' practitioners resent most: exploding prototype counts, mass-production claims in slide decks, and products that only perform choreographed shows in showrooms.

Zhishen's WRC booth instead featured:

Quadrupeds climbing open-grate steel stairs;
Robotic arms performing precise operations in simulated work zones;
Joint modules running continuously 1 meter underwater.

Liu: "We are not 'planning to enter' — we are already running in real scenarios."

'Lay Eggs Along the Way': 15,000 Units Is Not a Slogan

Most peers pursue 'wait for the perfect humanoid': betting everything on long-term general humanoids with no revenue in between. By 2026, investors demand real cash generation.

Zhishen's logic: embodied intelligence decomposes into modules — all-terrain mobility, spatial cognition/autonomous navigation, manipulation. The first two already let quadrupeds handle tedious, repetitive, dangerous work. Each time a module matures, it goes into the real world, generating revenue that feeds back into R&D.

Hard data as of June 2026:

| Dimension | Data | |---|---| | Cumulative production | 15,000+ units | | Monthly capacity | 5,000+ units | | In-house joint modules | 1 million units/year | | Security deployments | Thousands (property/malls/scenic/coastline) | | Power inspection projects | 100+ | | Emergency product matrix | 20+ models | | Fire brigades covered | 100+ | | Ecosystem partners | 300+ | | Provinces/cities deployed | 100+ |

The 'Three Gates': From 'Can Build' to 'Can Use Well'

Gate 1 — Technology: Stable Reproducibility

Customers care whether tasks can be repeated continuously across similar sites, not whether one demo succeeds. Zhishen tracks IP66 rating, endurance (Gangbang L2: 150 kg static load, 16 kg sustained carry), load capacity (P85MAX-S: 1 kg module driving 90 kg), joint reliability, temperature rise, fall recovery, and long-term maintenance. The WRC demo of joint modules running 1 m underwater made this visible.

Gate 2 — Scenarios: Completing the Task

Reaching the site solves only 'can it get there.' Real deployment requires answering: what to inspect, how to collect data, how data returns, what happens on anomalies, how results enter the customer's business. This maps to Zhishen's 'chain leader + module output' model — in-house joint modules (1M/year) plus 300+ integrators handling scenarios.

Gate 3 — Sustained Delivery: Reusable Product Capability

Recurring project requirements are folded back into the product layer via standard interfaces, SDKs, modular extensions, and version iteration, reducing repeated development in the next project.

Capability Checklist: Strength + Environment + Intelligence

Strength: P85MAX-S joint (1 kg driving 90 kg); Gangbang L2 quadruped (7:1 payload-to-weight ratio, 16 kg sustained carry, 150 kg static); NE01 humanoid (180 N·m peak joint torque). The point is turning peak research specs into sustained torque, thermal control, and lifetime curves.
Environmental adaptation: Second-gen Qilin alloy joints with IP66 industrial outdoor rating — 'second-gen' meaning first-generation field reliability issues were absorbed into the redesign.
Intelligence — 'big brain + cerebellum': importing the smart-vehicle 'rule stack + end-to-end' approach into robot navigation, plus perception-control unified models enabling autonomous climbing of industrial open-grate ladders.

Business Model: Chain Leader + Module Output

Upward: in-house CHAMP high-power-density joint modules, 1 million+ annual capacity, supplying core power components industry-wide;
Downward: 300+ integrators, universities, and industry partners building scenarios.

This separates making robots from using robots — the embodied intelligence company does the hard part of building; integrators do the hard part of deploying. It is 'laying eggs along the way' expressed at the supply-chain level.

Three Verticals with Verifiable Numbers

Security: Thousands of Units on Routine Patrol

Thousands of quadrupeds deployed at residential communities, commercial plazas, industrial parks, scenic coastlines for patrol, access control, and anomaly alerts. Top-tier security integrators are all partners. One quadruped replacing one night-shift guard (~5,000 RMB/month savings) at 15,000 units implies ~75 million RMB/month in labor savings.

Emergency: 100+ Fire Brigades with Multi-Robot Solutions

At 100+ fire brigades in Beijing, Tianjin, Gansu, Zhishen provides 'quadruped + drone/off-road vehicle' integrated solutions; a 20+ model matrix covers explosion, toxic, smoke, and heat scenarios — multi-robot coordination rather than one robot doing everything.

Power: 100+ Inspection Projects Including Storage Stations

100+ power inspection projects delivered across substation tiers and energy storage stations, where heat, humidity, and toxic gas leaks — the robot's comfort zone — are the key pain points.

'Lay Eggs Along the Way' vs. 'Wait for Perfect'

| Dimension | Wait for Perfect | Lay Eggs Along the Way | |---|---|---| | Revenue start | After general humanoids mature | Now, quadrupeds first | | Unit cost | High (R&D on few units) | Medium (amortized by volume) | | Customer validation | Closed-door dev → one-shot | Continuous feedback → iteration | | Cash flow | Constant fundraising | Revenue funds R&D | | 6 months later | Lab prototypes | Tens of thousands running in the field | | 12 months later | Struggling commercialization | Hundreds of thousands deployed + new modules |

Liu on why robots deserve to exist: "Protect lives — let machines go in first so people risk less; fill gaps in tasks labor cannot cover long-term; raise efficiency by turning experience-based work into standardized, traceable tasks; and advance industries by giving sensors and algorithms a mobile carrier into the real world."

What's Really Missing: Industry Standards

Liu's overlooked judgment at WRC: "The industry still lacks a mature standard — performance metrics, functional metrics, evaluation systems, certification." Without standards there is no scale; without scale, no cost advantage; without cost advantage, no real deployment. Whether China's embodied intelligence sector can move from private standards to industry-recognized standards in the next 12–18 months is the key variable in moving from PPT valuations to revenue-based valuations.

Key Insights

| Dimension | Zhishen path | Peer demo path | |---|---|---| | Product | Quadruped first / humanoid later | All-in on humanoid | | Production | 15,000+ / 5,000 per month | PPT 100K / actual hundreds | | Scenarios | Security + emergency + power | Showroom demos | | Customers | Tier-1 integrators + fire brigades | Investors + media | | Revenue | Rolling | Constant fundraising | | Moat | 1M joint modules/year | Concepts + stories |

References

[1] China.com, "Interview with Zhishen Technology Co-founder Liu Yulong," 2026-08-27 — https://news.china.com/socialgd/10000169/20260827/49701024.html

[2] Sina Finance / China Quality News, "Zhishen Technology Empowers Robots That Really Work," 2026-08-26 — https://finance.sina.cn/2026-08-26/detail-iniprqcu2379993.d.html

[3] Consumer Daily, "Zhishen Technology at WRC 2026," 2026-08-26 — https://www.xfrb.com.cn/article/kj/13300518997982.html

[4] Tencent News / CNR Finance interview, "Liu Yulong: Embodied Intelligence Mass Production Needs Industry Standards," 2026-08-26 — https://new.qq.com/rain/a/20260826A05XJU00

[5] NetEase, "15,000 Units Produced: Why Liu Yulong's 'Robots That Work' Resonates," 2026-08-26 — https://www.163.com/dy/article/L59Q6FQN0556C2XD.html