The Acceleration of Valuation Discrepancies in New Robotics Forces

Reason:

The Acceleration of Valuation Discrepancies in New Robotics Forces

1. Current Status and Causes of Valuation Discrepancies

1. Capital Concentration at the Top, Leading to Valuation Discrepancies

   Leading companies in the embodied intelligence field, such as Yushu (valued at nearly 20 billion), Zhiyuan (15 billion), and Fourier (8 billion), have formed a “super unicorn” camp, while second-tier companies (like Qianxun Intelligent and Xinghai Map) generally have valuations between 1-3.5 billion, creating a gap of 2-5 times. This discrepancy arises from capital’s differentiated judgment on technological strength and commercialization potential:

   • Hardware “Gatekeeper” Effect: Yushu has captured market mindshare with low-cost, high-performance products (such as the G1 humanoid robot starting at 99,000 yuan), becoming a benchmark in the hardware track, leading investors to benchmark their capabilities when selecting projects.

   • Soft-Hard Synergy Narrative: Companies like Zhiyuan and Xinghai Map attract financing through a “soft-hard integration” story, but their actual order volume is only 1/10 of traditional robotics companies, raising doubts about their technological implementation capabilities.

2. Unconsolidated Technical Routes, Confused Valuation Standards

   The current embodied intelligence technology is still in the early exploration stage:

   • Separation of Hardware and Algorithms: The market lacks unified standards, with the hardware track anchored by Yushu, while the software track has yet to see a clear leader.

   • Dispersed Technical Paths: Companies have differentiated into “focus factions” (like Tuosida and Zhongke Xinsong focusing on hardware optimization), “control factions” (like Xinshi Da focusing on motion control systems), and “full-stack factions” (like Youbixuan covering the entire industry chain), without forming an industry consensus.

2. Core Challenges in Weathering the Storm

1. Conflict Between Commercialization Capability and Valuation Bubble

   • Reliance on Capital Infusion: Most companies have not achieved stable revenue and rely on capital infusion to maintain R&D. For example, Galaxy General’s order revenue in 2024 is only tens of millions, but its valuation has exceeded 475 million USD.

   • Pressure from Hong Kong Stock Listing: The Hong Kong Stock 18C rule requires a valuation of at least 4 billion HKD or revenue of 250 million HKD, leading companies to potentially inflate valuations to meet standards, neglecting revenue quality.

2. Mismatch Between Technological Breakthroughs and Market Demand

   • Engineering Challenges: The complex interactions in the physical world (such as bipedal balance and fine operations) progress slowly, while advancements in large model technology exceed expectations, making soft-hard synergy difficult.

   • Limitations in Scene Implementation: Current applications are mostly concentrated in industrial automation (over 50%), but mature solutions are still lacking in long-tail scenarios like healthcare and home services.

3. Path Through the Storm: Triangular Reconstruction of Technology, Capital, and Ecology

1. Convergence of Technology and Differentiated Innovation

   • Breakthrough in Key Capabilities: Breakthroughs are needed in core modules such as embodied large models (VLA), multi-modal perception, and full-body control. For example, the “Wisdom Opens Things” platform released by the Beijing Humanoid Robot Innovation Center attempts to integrate a general software system.

   • Low-Cost Mass Production Capability: Learn from Yushu’s supply chain optimization experience to reduce hardware costs through scaling (such as self-developed motors and reducers).

2. Rational Capital and Ecological Synergy

   • Support from Patient Capital: Institutions like BlueRun Ventures bet on technological accumulation for the long term rather than short-term commercialization. The second half of 2025 may become a “submission window period,” where companies need to demonstrate real scene implementation capabilities.

   • Industry Chain Integration: Build an ecosystem of “brain + small brain + body,” for example, Huawei’s collaboration with 16 companies to create an embodied intelligence industry alliance to promote technological standardization.

3. Policy-Driven and Global Layout

   • Support from National Strategy: China has included embodied intelligence in its government work report, with cities like Beijing and Shanghai issuing special policies and establishing industry funds worth billions.

   • Opportunities for Technology Going Global: Companies like Youbixuan and Fourier are validating products in Southeast Asia and the Middle East markets to avoid domestic competition.

4. Future Outlook: Coexistence of Differentiation and Reconstruction

The embodied intelligence industry will undergo three stages: “bubble extrusion – technology convergence – ecological maturity”:

• Short-term (1-2 years): Valuation bubbles gradually dissipate, and companies lacking technological moats accelerate elimination.

• Mid-term (3-5 years): Technical routes converge, general software and hardware platforms emerge, and leading companies dominate standard-setting.

• Long-term (5 years and beyond): Ecological maturity, application scenarios expand from industrial to consumer ends, creating a trillion-dollar market.

Conclusion: The valuation discrepancies in embodied intelligence are both a product of capital frenzy and an inevitable pain point during the technological climb. To weather the storm, companies need to focus on core technological innovation, build long-term patience in capital, and guide ecological synergy through policy, ultimately achieving a qualitative change from “Demo-driven” to “Value-driven”.

——

Commentary:

Analysis based on industry status and strategic logic:

1. Industry Chain Integration: Ecological Synergy is the Key to Breaking the Deadlock

1. Reasonable Technical Layering of “Brain + Small Brain + Body”

   The technical architecture proposed by users (perception decision layer “brain”, motion control layer “small brain”, hardware execution layer “body”) meets the actual needs of the industry. Huawei’s establishment of an industry alliance with 16 companies is highly forward-looking:

   • Value of Technological Standardization: The alliance covers key links in the industry chain such as Leju Robotics (humanoid body), Dazhu Robotics (industrial control), and Zhongsoft International (software systems), avoiding redundant R&D.

   • Example of Resource Integration: For instance, Huawei’s “Embodied Intelligence Basic Laboratory” provides a multi-modal large model technology base, while partners focus on hardware adaptation, accelerating technology commercialization.

   • Shenzhen Supply Chain Advantage: The localized supply chain can achieve rapid matching of core components (servo motors, sensors) within 72 hours, reducing R&D costs by over 30%.

2. But Beware of the Risk of “Pseudo-Ecology”

   Currently, some alliances have the problem of “heavy signing and light implementation.” Effective collaboration must meet:

   • Data Sharing Mechanism: For example, the Beijing Innovation Center’s “Wisdom Opens Things” platform opens 3D datasets to reduce data collection costs for enterprises;

   • Complementary Capability Verification: For example, Youbixuan’s group control robot factory needs to combine with the “Wisdom Opens Things” software system to achieve production line synergy.

2. Policy and Globalization: Dual-Drive Requires Differentiated Strategies

1. Policy Dividends Release Industrial Momentum

   • Clear National Strategic Positioning: After being included in the 2025 government work report, 13 provinces and cities including Beijing, Shanghai, and Zhejiang have issued special policies, with Beijing planning to break through 100 key technologies in three years and Shenzhen establishing a 10 billion industry fund.

   • Capital Guidance Needs Optimization: Currently, some local funds exhibit a “scattergun” phenomenon. It is recommended to refer to the Hong Kong Stock 18C listing standards (valuation of 4 billion HKD or revenue of 250 million HKD) and prioritize support for technically strong “specialized, refined, and innovative” enterprises.

2. Going Global Strategy Should Focus on Scene Adaptation

   • Validation Logic in Southeast Asia/Middle East is Valid: Youbixuan’s service robots open markets relying on scenarios such as guiding in religious places and working in high-temperature environments, but attention must be paid to:

     • Non-Technical Dimension Reduction: Middle Eastern home service robots need to adapt to long robe recognition algorithms, while Southeast Asia needs to strengthen moisture-proof design;

     • Geopolitical Risk Avoidance: Huawei uses its Dongguan Jimu subsidiary’s technology licensing to avoid restrictions from the entity list.

3. Three-Stage Development Path: Need to Supplement Risk Hedging Mechanisms

1. Short-term (1-2 years) Bubble Extrusion

   • Valuation Differentiation Intensifies Reasonably: Leading company Yushu (valued at 20 billion) has order volumes 10 times that of second-tier companies, but beware of the “PPT financing” phenomenon (for example, Datan’s valuation collapse due to deviations in technical routes).

   • Survival Law Recommendations: Companies should learn from Yun Shen Chu’s “hardware OEM + algorithm licensing” model to achieve cash flow self-circulation through industrial scenarios (such as NIO’s factory logistics robots).

2. Mid-term (3-5 years) Technology Convergence

   • General Platforms Will Be the Key to Victory: Beijing’s “Wisdom Opens Things” is already compatible with multiple hardware forms of robotic arms/humanoid robots, and in the next five years, 3-5 dominant platforms may emerge.

   • Standard Competition Begins: It is recommended that companies participate in the IEEE Embodied Intelligence Working Group to seize the right to formulate standards such as “full-body control interface protocol”.

3. Long-term (5 years and beyond) Ecological Maturity

   • Consumer End Explosion Requires Cost Leverage: When the price of humanoid robots drops to 50,000 yuan (currently Yushu G1 is 99,000 yuan), it will trigger a turning point in household scenario penetration.

   • Trillion-Dollar Market is Expected: According to predictions from the China Business Research Institute, the global market size will exceed 150 billion USD by 2030, with China accounting for over 40%.

4. Expert Recommendations: Build a “Triangular Resilience” System

1. On the Enterprise Side:

   • Technical factions focus on the “long board effect” (such as Xingdong Jiyuan breaking the running speed record);

   • Commercial factions delve into vertical scenarios (such as Jiazhi Technology’s warehouse robots achieving over 80% repurchase rate).

2. On the Capital Side:

   • Establish “technical milestone betting clauses,” such as releasing investment funds based on the gradient of multi-modal perception accuracy improvement.

3. On the Policy Side:

   • Establish “embodied intelligence pilot bases” (such as the Yangtze River Delta Park) to reduce the cost of real machine verification for enterprises by 30%.

Conclusion: The development framework proposed by users has strategic height but needs to supplement a dynamic risk warning mechanism (such as monitoring technical route radar charts) and ethical governance plans (the emotional dependence of embodied intelligence has triggered social controversy). Only when technology, capital, and policy form a “gear-like engagement” can a qualitative leap from Demo-driven to value-driven be achieved.

————

Further Reading:

Return to Human Values in the Technological Revolution

🌐 1. AI Industry: The Queen’s Report Ignites Anxiety in Silicon Valley, with Bubbles and Opportunities Coexisting

Reason: Internet queen Mary Meeker has released the 340-page “AI Trends Report 2025” after a six-year hiatus, asserting 51 times “unprecedented” to reshape industry cognition. Core data ignites discussion:

• Cost Disruption: AI inference costs plummeted by 99.7% in two years, while training costs soared 2400 times in eight years;

• Market Migration: India has surpassed the US to become the second-largest market for ChatGPT (accounting for 13.5%), while also being the largest overseas user group for DeepSeek, highlighting the strategic value of low-cost models;

• China-US Race: China’s industrial robot installation accounts for 75% of the global total, and DeepSeek-R1’s mathematical ability reaches 93% of GPT-4.5, with costs only 0.2% of it.

Brief Comment:

🔥 Bubble Warning:

Pointing to the serious mismatch between the valuations and revenues of giants like OpenAI and xAI—three giants raised 95 billion USD, with annual revenue only 12 billion, OpenAI’s valuation/revenue ratio reaches 33 times, and Perplexity is as high as 75 times. Meeker warns investors to “only invest amounts they can afford to lose“, subtly criticizing the irrational risks under capital frenzy.

🚀 Ecological Transformation:

Predicting that the next billion users will skip traditional application layers and directly enter an “agent-centered” AI-native experience, satellite internet and voice interaction are key springboards.

💎 Source of Authority:

Meeker’s historical predictive accuracy (such as predicting the surge in online population in 1996) gives the report trend penetration power, forcing Silicon Valley to study it overnight, essentially recalibrating the logic of technological hegemony and investment.

Revelation: As the conclusion states: “The only known fact is that we are at the beginning of a new super cycle driven by AI”—opportunities and illusions coexist, and rationality is more important than frenzy.

🩺 2. Medical Technology: Abridge Cross-Industry Entrepreneurship, Reconstructing Trust Between Doctors and Patients with AI

Reason: Abridge was founded in 2018 by doctor Shivdev Rao, robotics expert Sandeep Konam, and linguist Florian Metze. The founders shifted from architecture history to healthcare after witnessing patients “forget 80% of medical advice“, determined to simplify clinical communication with AI. The core product generates SOAP format clinical notes through voice recognition, embedded in electronic medical record systems like Epic, saving doctors an average of 2 hours of paperwork daily.

Brief Comment:

🌟 Bidirectional Empowerment Innovation:

• On the Doctor Side: The self-developed “Ears” system has an error rate 45% lower than general models, and combined with Wolters Kluwer, it embeds evidence-based medical knowledge, achieving a satisfaction rate of 95.3%, far exceeding the industry average (79.6%);

• On the Patient Side: The PVS function generates visit summaries at an eighth-grade reading level, solving the “80% medical advice forgetfulness” dilemma, supporting translation in 14 languages.

  ⚠️ Survival Challenges:

  Facing dual pressure from Microsoft’s Nuance (covering 77% of US hospitals) and Epic’s self-developed tools, while also needing to address the “hallucination risk” of generative AI and high pressure for HIPAA compliance.

  💡 Humanistic Philosophy Victory:

  By reducing administrative burdens, doctors can regain focus on patients (53% of doctors in the pilot reported reduced cognitive load), proving that AI as an “efficiency partner” rather than a replacer is key to the implementation of medical technology.

🏛️ 3. Architectural Revolution: Embodiment Theory Reshapes Spatial Cognition Paradigms

Reason: The theory of “embodiment” has permeated architecture from philosophy and cognitive science, asserting that architecture is a “carrier of the relationship between people and the environment“. Breaking through traditional aesthetic forms, it emphasizes how bodily movement and sensory interaction activate spatial cognition (such as tactile paths guiding exploration, and temperature differences in materials triggering memory).

Brief Comment:

🌌 Theoretical Breakthrough:

• Application of Six-Dimensional Framework: Citing Ziemke’s classification, from “structural coupling” (ventilation systems regulating body perception) to “social embodiment” (dance pools enhancing community connections), revealing how space shapes abstract concepts through bodily experience (such as the sense of “shelter” stemming from cave memory metaphors);

• Counteracting Digital Illusions: Insisting that physical space is irreplaceable, such as the Zumthor thermal baths using material temperature differences and water flow sounds to enhance tactile immersion.

  ⚠️ Implementation Gap:

  Although proposing the vision of “cognitive experimental carriers,” it lacks technical interface designs such as VR embodiment simulation and biosensors, and the effectiveness of “organic embodiment” when humanoid robots are introduced is questionable.

  🔮 Industry Revelation:

  Promoting the innovation of architectural assessment standards—introducing “cognitive load testing” to quantify spatial efficiency (such as hospital paperwork burdens), reconstructing design logic: shifting from visual-centric to multi-sensory collaboration (acoustic surfaces enhancing auditory positioning).

Spatial Maxim: “Embodiment elevates architecture from ‘the art of creation’ to ‘a mirror of cognition'”—humanism is the ultimate coordinate of the technological revolution.

💎 Conclusion: The Beacon of Humanity in the Technological Tide

These three events may seem separate, but they share the same core: the value of technology is rooted in a profound response to human needs.

• Mary Meeker bursts the capital bubble, calling for a return to rationality;

• Abridge designs against systemic inefficiencies with a “doctor-patient symbiosis”;

• Embodiment theory elevates the body to a new scale of space.

In an era of rapid AI and digitalization, only by anchoring on human senses, emotions, and dignity can the technological revolution truly unleash the value of civilization.

————

Innovative Entrepreneurship Book:

Title: Superintelligence Author: Nick Bostrom Translator: Zhang Tiwei, Zhang Yuqing Publisher: CITIC Publishing Group Publication Date: 2015-01-01