ChinaBizInsight

RaaS, Solutions, and Ecosystems: The Three Business Models Powering China’s Embodied AI Boom

RaaS, Solutions, and Ecosystems: The Three Business Models Powering China’s Embodied AI Boom

In 2025, the first Chinese embodied AI companies crossed the “ten-billion club” revenue threshold. By mid-2026, the industry had attracted 935 billion yuan ($128 billion) in investment across 322 funding rounds — a fivefold increase year-over-year.

But behind these staggering numbers lies a more fundamental question: How do these companies actually make money?

The answer is more complex — and more interesting — than simply “selling robots.” China’s embodied AI industry has developed four distinct business models, each with its own economics, customer profiles, and strategic implications.

For global businesses evaluating Chinese partners — whether as buyers, investors, or strategic collaborators — understanding these models is not optional. It’s the key to assessing financial sustainability, customer traction, and long-term viability.

1. The Four Models at a Glance

China’s embodied AI industry has converged around four primary business models, each representing a different stage of market maturity and customer relationship:

ModelDescriptionRepresentative CompaniesKey Economics
Hardware Sales Direct sale of robot units to customers Unitree, UBTECH, Galaxy General Scale-driven gross margins (Unitree: ~60%)
RaaS / Leasing Subscription-based robot access, pay-per-use ZhiYuan (擎天租), Figure AI, Agility Robotics Recurring revenue, lower customer entry barrier
Solutions / Systems Integration End-to-end industry-specific deployments UBTECH (industrial), ZhiYuan, Weizhi Manufacturing High ticket price, high customer stickiness
Ecosystem / Platform Open platforms, APIs, developer ecosystems Tesla, ZhiYuan (open-source), 51WORLD High upfront investment, very low marginal cost

Importantly, these models are not mutually exclusive. Leading companies are increasingly adopting hybrid approaches — combining hardware sales with RaaS subscriptions, solutions delivery, and ecosystem building. As one industry executive put it: “The future is a mixed business model — there will be whole-machine sales, RaaS leasing, and MaaS (Model-as-a-Service) cloud-based models.”

2. Hardware Sales: The Scale Game

🏭 Hardware Sales Volume-Driven

The most straightforward model — sell robots, generate revenue

The Logic: Hardware sales are the foundation of the embodied AI industry. The economics are simple: manufacture at scale, drive down unit costs, and capture margin through volume. This is the model that has powered China’s dominance in global shipments.

Unitree (宇树科技): The standout example. Unitree has achieved gross margins of approximately 60% on its humanoid robot business, with revenue reaching 1.7 billion yuan ($233 million) in 2025 — a fourfold increase year-over-year. The company’s IPO filing shows gross margins rising for three consecutive years, reaching 60% in 2025, with the humanoid segment alone contributing 868 million yuan ($119 million) in revenue at a 60.13% gross margin.

The Scale Effect: In the first half of 2026, China shipped approximately 11,900 humanoid robots globally (out of 19,100 total, or 97% market share). ZhiYuan alone accounted for 44% (approximately 5,000 units), while Unitree held 31%. This scale is driving dramatic cost reductions — the average cost of a planetary roller screw has fallen to approximately ¥1,850, and is expected to decline further as volume increases.

The Trade-off: Hardware sales are capital-intensive and require significant manufacturing capacity. They also create one-time, transactional customer relationships rather than recurring revenue streams. As Unitree’s IPO filing reveals, despite revenue growth, net profit after non-recurring items actually decreased — highlighting the pressure of scaling.

3. RaaS / Leasing: The Subscription Revolution

🔄 RaaS (Robot-as-a-Service) Recurring Revenue

Pay-per-use access to robot capabilities — no large upfront capital expenditure

The Logic: RaaS lowers the barrier to adoption by allowing customers to access robot capabilities without large upfront capital expenditure. It transforms the relationship from transactional (one-time sale) to relational (ongoing service) and creates predictable, recurring revenue streams.

ZhiYuan’s 擎天租 (Qingtianzu): The most prominent example. Launched in December 2025 as a joint venture between ZhiYuan and Feikuo Technology, 擎天租 is the world’s first open humanoid robot RaaS platform. By mid-2026, the platform had:

  • Raised hundreds of millions of yuan in Series A and A+ funding at a 7 billion yuan ($960 million) valuation
  • Integrated seven major robot manufacturers including ZhiYuan, Unitree, and Lingchu
  • Built a network of over 4,000 deployable robots
  • Expanded to cover more than 100 cities across China
  • Signed a strategic partnership with JD.com for logistics and distribution

RaaS Economics: Daily rental rates have fallen from ¥15,000–20,000 at peak to ¥3,000–5,000 as capacity has scaled, with the market shifting from “hard to find” to “platformized, standardized operations.” The platform’s model is simple: standardize delivery, make robot leasing as convenient as sharing a power bank.

Global Parallels: This isn’t unique to China. Agility Robotics uses RaaS subscriptions with multi-year binding contracts, accumulating over 65,000 operating hours and moving more than 100,000 logistics boxes. Figure AI has achieved stable operation in BMW factories.

IDC’s View: IDC’s April 2026 report found that user acceptance of RaaS models doubled year-over-year, and projects global humanoid robot shipments to exceed 510,000 units by 2030. The report noted that “future industry competition will shift from technology demonstrations to actual application capabilities and commercial value delivery.”

4. Solutions / Systems Integration: The High-Ticket, High-Stickiness Model

🏗️ Solutions & Systems Integration High Value, High Stickiness

End-to-end deployments for specific industries — not just robots, but complete workflows

The Logic: Solutions go beyond selling hardware. They involve understanding a customer’s specific operational challenges, designing a complete workflow, deploying robots, and providing ongoing support. This model creates high customer stickiness and high barriers to switching, but requires deep industry expertise and long sales cycles.

UBTECH (优必选): The leading example in industrial solutions. In 2026, UBTECH positioned itself for “大规模商业化” (large-scale commercialization), focusing on three key scenarios: aerospace manufacturing, automotive, and smart manufacturing.

  • New orders exceeded 1 billion yuan in the first half of 2026, including three orders exceeding 100 million yuan each
  • Airbus placed an initial order for 100 Walker S2 units
  • Walker S series shipment target raised from 2,000–3,000 to 5,000 units for 2026, with annual production capacity targeting 10,000 units
  • 2026 is defined as the “verification year” — the transition from proof-of-concept (POC) to large-scale industrial deployment

Other Solution Providers: Companies like Weizhi Manufacturing (微亿智造) have built their entire business around industrial solutions, combining AI vision, robotics, and process optimization for specific manufacturing verticals.

The Economics: Solutions generate higher average revenue per customer than pure hardware sales, but require longer sales cycles and higher upfront investment in customer acquisition and deployment. The payoff is higher customer lifetime value and lower churn.

5. Ecosystem / Platform: The Long Game

🌐 Ecosystem & Platform Network Effects

Open platforms, APIs, developer communities — building the infrastructure others build on

The Logic: Ecosystem models are the most ambitious and highest-risk — but also potentially the most valuable. The goal is to become the operating system or infrastructure layer that other companies build on, capturing value through network effects, developer lock-in, and data aggregation.

Tesla: The most visible example globally. Tesla’s Optimus program is not just about selling robots — it’s about building a unified foundation model that spans both autonomous driving and embodied AI, leveraging shared data, shared compute, and shared manufacturing.

ZhiYuan’s Open-Source Strategy: In June 2026, ZhiYuan launched the “YuanSheng Ecosystem Plan,” committing 2 billion yuan over five years to foster innovation and partnerships. The company has open-sourced real-scene datasets like “agibot world 2026,” covering commercial and home environments with structured, finely annotated data.

51WORLD (五一视界): A Hong Kong-listed company building embodied data infrastructure. In August 2026, it launched two core products: AperData (an embodied data infrastructure) and AperOne (an embodied application platform). The company’s strategy is to become the data and simulation backbone for the entire industry.

The Economics: Ecosystem models require massive upfront investment in platform development, community building, and data infrastructure. But once established, they enjoy very low marginal costs and can capture value across multiple layers of the stack — hardware, software, data, and services.

6. Comparative Analysis: Which Model Wins?

The answer depends on who you are and what you’re trying to achieve. Here’s a framework for evaluating each model:

Hardware Sales
Best for: Scale & Volume

Unitree: 60% gross margin, 1.7B revenue

RaaS / Leasing
Best for: Adoption & Recurring Revenue

擎天租: 7B valuation, 4,000+ robots

Solutions
Best for: High-Value Enterprise Deployments

UBTECH: 1B+ new orders, Airbus 100 units

Ecosystem / Platform
Best for: Long-Term Network Effects

51WORLD: Data infrastructure play

Key Insight: The most successful companies are not choosing one model — they are building combinations. ZhiYuan sells hardware, operates a RaaS platform, delivers industrial solutions, and builds an open ecosystem. UBTECH sells hardware and delivers high-value solutions. Unitree is expanding from pure hardware sales into RaaS through partnerships like 擎天租.

📈 The Hybrid Future: Industry consensus is that the winning business model will be a hybrid — combining hardware sales, RaaS subscriptions, solutions delivery, and ecosystem participation. As one executive put it: “The future is a mixed business model — there will be whole-machine sales, RaaS leasing, and MaaS (Model-as-a-Service) cloud-based models.”

7. What This Means for Global Businesses

If you’re evaluating a Chinese embodied AI company — whether as a potential partner, investment target, or customer — here are the critical questions to ask:

✅ Due Diligence Checklist:

  1. What is the company’s primary revenue model? Is it hardware sales, RaaS, solutions, or a hybrid? Verify the revenue mix — not just top-line numbers.
  2. How sustainable is the model? Hardware sales require continuous volume growth. RaaS requires high utilization and low churn. Solutions require deep industry expertise and long customer relationships. Ecosystems require critical mass.
  3. What is the customer concentration risk? Does the company depend on a small number of large customers? UBTECH’s Airbus order is impressive, but also creates concentration risk.
  4. What are the margins and unit economics? Unitree’s 60% gross margin is exceptional — but net profit after non-recurring items decreased, highlighting the cost of scaling.
  5. Is the company building a moat? Hardware scale, RaaS network effects, solutions expertise, and ecosystem lock-in are all defensible moats — but they require different capabilities and investment profiles.

8. How ChinaBizInsight Supports Your Due Diligence

At ChinaBizInsight, we understand that verifying the claims of potential Chinese partners is essential — and challenging. Language barriers, limited access to authoritative data sources, and the rapid pace of change make it difficult to separate real capability from marketing hype.

We provide verified, authoritative intelligence on Chinese embodied AI companies — exactly the kind of data you need to make informed partnership, investment, and sourcing decisions:

  • Official Enterprise Credit Reports — directly from China’s National Enterprise Credit Information Publicity System, with official watermarks and seals. Verify a company’s legal status, business scope, and regulatory compliance.
  • Custom Due Diligence Reports — combining multiple authoritative sources to provide a 360-degree view of a company’s financial, legal, and operational status.
  • Executive Background Checks — detailed reports on directors, supervisors, and senior management, including their investment history, external positions, and risk records.
  • IP Searches — trademark, patent, and copyright verification to ensure your partner actually owns the technology they claim.
  • Document Legalization & Apostille — ensuring that Chinese corporate documents are recognized internationally for contracts, compliance, and legal proceedings.

All our data is traceable to official sources, presented in English, and delivered with the speed and professionalism that global businesses demand.

🔎 Know your Chinese partners before you commit.

Get the verified intelligence you need to make confident partnership and investment decisions in China’s embodied AI sector.

Contact ChinaBizInsight →

📍 Based in China · Serving the world · Know your Chinese partners

Sources:

EO Intelligence, 2026 Global Embodied AI Commercialization Report (2026)

Unitree IPO filings and public disclosures (2025–2026)

ZhiYuan / 擎天租 public announcements and media reports (2026)

UBTECH investor presentations and public disclosures (2026)

IDC, Global Humanoid Robot Market Forecast (April 2026)

Various industry reports: DoNews, Jiemian, 163.com, China Business News

Counterpoint Research, Humanoid Robot Shipments H1 2026 (2026)

Your strategic bridge to transparent business in China.

Native Expertise
Direct Access
Official Sources
VIEW SAMPLES CONSULT EXPERT

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top