Analysis of the Impact of the 'Physical AI' Era on the Investment Value of China's AI Industry

Based on the above in-depth research and data analysis, I now provide you with a complete investment analysis report:

---

NVIDIA CEO Jensen Huang clearly stated at the 2026 International Consumer Electronics Show (CES 2026):

[1]. This declaration marks a revolutionary leap for artificial intelligence from “understanding language” to “understanding the physical world”[2].

Jensen Huang’s core definition of “Physical AI” is:

, with its essential characteristics including:

Technical Features	Connotation Explanation
World Understanding	AI needs to understand physical common sense such as object permanence (objects do not disappear out of thin air) and causal relationships (thrust causes objects to move)
Interaction Capability	AI shifts from screen interaction to physical interaction with the real world
Closed-Loop Decision-Making	Establish a complete “perception-reasoning-action” closed loop, rather than staying at the information processing level only

Jensen Huang pointed out that the breakthrough in physical AI will bring the

its own “ChatGPT Moment”, and AI will move from screens to factories and streets to become an intelligent entity with physical action capabilities[3]. Robotaxis are explicitly identified by Jensen Huang as the first batch of beneficiary application scenarios.

: NVIDIA did not release new GPU products for the first time in 5 years at this CES, instead focusing on software, models, and system-level capabilities for physical AI and inference AI, releasing open-source tools such as the Cosmos series models and the Isaac GR00T humanoid robot model[4].

---

As the world’s largest manufacturing country, China has a

in physical AI application scenarios:

Advantage Areas	Specific Performance	Data Support
Industrial Robots	World’s largest industrial robot market	China accounts for approximately 45% of the global industrial robot market share[5]
Intelligent Manufacturing	Complete manufacturing industrial chain	Penetration rate of L2-level assisted driving exceeds 50% [6]
Digital Twin	More than half of prefecture-level administrative regions have carried out related construction	National digital twin city coverage rate exceeds 50% [7]

:

• China’s local supply chain has
  strong cost reduction capabilities
  in core motors, drivers, and sensors
• Product iteration speed: approximately
  one generation every six months to one year
  (European companies typically take 2-3 years)
• Mass production cycle: approximately
  1 year
  in China,
  3-5 years
  in Europe[8]
• The “domestic GPU + physical AI” collaboration model has been formed (in-depth binding between Moore Threads and 51WORLD)

National-level policies continue to increase, clearing institutional obstacles for the development of physical AI:

• The “14th Five-Year Plan for Digital Economic Development” explicitly supports the integration of digital twins and physical AI
• The State Council’s “Implementation Opinions on Accelerating Scene Cultivation and Opening Up to Promote Large-Scale Application of New Scenarios”
• The Ministry of Industry and Information Technology’s “Implementation Opinions on the ‘Artificial Intelligence + Manufacturing’ Special Action”
• 30% subsidy policy
  for industrial software procurement[9]

China demonstrates

advantages in data accumulation and commercial closed-loop in complex urban scenarios[10]:

• Autonomous driving field
  : The penetration rate of L2-level passenger vehicles is expected to rise to
  90%
  by 2030
• Unmanned driving financing
  : Total financing in Q3 2025 exceeded
  14.6 billion yuan
  , with robotaxis accounting for 5.7 billion yuan
• Embodied intelligence investment
  : 8 major internet giants (Baidu, Alibaba, Tencent, Meituan, etc.) have invested a total of
  62 times
  , with an amount of
  1.45-3.4 billion yuan
  [11]

---

┌─────────────────────────────────────────────────────────────────┐
│                      Panorama of the Physical AI Industrial Chain                            │
├─────────────────────────────────────────────────────────────────┤
│  Upstream (Computing Power Layer)         Midstream (Platform Layer)         Downstream (Application Layer)      │
│  ┌─────────┐           ┌─────────┐           ┌─────────┐       │
│  │ Domestic GPU  │──────────▶│ Simulation Engine │──────────▶│ Robot  │       │
│  │ Cambricon   │           │ Digital Twin│           │ Autonomous Driving│       │
│  │ Moore Threads │           │ Spatial Intelligence│           │ Intelligent Manufacturing│       │
│  └─────────┘           └─────────┘           └─────────┘       │
│     27%                   28%                   45%            │
│  [Highest Technical Barrier]      [Core Value Link]       [Demand-Driven]        │
└─────────────────────────────────────────────────────────────────┘

Industrial Chain Link	Value Proportion	Investment Logic	Core Targets
Underlying Computing Power	27%	Infrastructure for large-scale implementation of physical AI, large space for domestic substitution	Moore Threads, Cambricon, Hygon Information
Platform Tools	28%	Highest technical barrier, key hub connecting computing power and applications	Fantuo Digital Creation, 51WORLD, iFLYTEK
Terminal Applications	45%	Most direct commercialization, demand explosion drives upstream development	UBTECH, BYD, XPeng Motors

:

Year	Market Size (USD 100 million)	Compound Annual Growth Rate
2024	37.8	-
2026	67.2	33.49%
2030	212.5	-
2034	679.1	33.49% [12]

: Expected to increase from 2024’s 20% to 2030’s 26%, showing a rapid upward trend.

---

:

• Scarcity
  : The only A-share digital twin enterprise covering the three tracks of intelligent driving, embodied intelligence, and industrial simulation
• Technical Barrier
  : Independently developed FTE engine with spatial positioning accuracy of
  0.05 mm
  , leading the industry
• Growth
  : Business transformation from traditional 3D visualization to high-value-added AI digital twins, with a continuous increase in revenue proportion

:

• A photovoltaic factory in Zhejiang: Through the AI 3D digital twin platform, energy consumption per 10,000 yuan of output value was reduced by
  12.8%
  , and annual CO₂ emissions were reduced by more than
  3,500 tons
• A precision machine tool enterprise: After applying the predictive maintenance system, the MTBF of key equipment increased to
  5,000 hours
  , and maintenance costs decreased by
  30%
  [13]

:

• Total orders in 2025 reached nearly
  1.4 billion yuan
  , with customers including leading automakers such as BYD and Geely
• Production capacity of industrial humanoid robots exceeds
  1,000 units
  , with over
  500 units
  delivered
• Target production capacity of
  10,000 units
  in 2026[14]

:

• 2025 revenue of HK$1.305 billion (YoY +23.6%)
• Gross profit margin of 23%-38%
• Full-stack self-developed capabilities (from hardware control to ROSA 2.0 system)

:

• Built a complete closed loop of “synthetic data - spatial intelligence model - simulation training platform”
• CAGR of revenue from 2022 to 2024 was
  30.02%
• H1 2025 revenue increased by
  62.04% YoY
• Reached a strategic cooperation with Moore Threads to form a “domestic GPU + physical AI” collaboration[15]

Investment Portfolio	Allocation Ratio	Target Type	Expected Return
Core Allocation	40%	Blue-chip leaders (intelligent driving leaders, chip leaders)	Steady growth
Growth Allocation	35%	High-potential targets (digital twin, robot manufacturers)	High elasticity
Speculative Allocation	15%	Early-stage targets (embodied intelligence, emerging players)	High risk and high return
Hedging Allocation	10%	Steady targets (beneficiaries of traditional manufacturing upgrading)	Defensive

---

2024-2025: Infrastructure Construction Period
    ├── Chip R&D, Algorithm Training, Infrastructure Construction
    └── Investment Focus: Underlying Computing Power, Core Algorithms

2025-2026: Initial Deployment Period
    ├── Industrial Robot Pilots, Digital Twin Project Implementation
    └── Investment Focus: Platform Tools, Solutions

2026-2027: Large-Scale Application Period
    ├── L4-Level Intelligent Driving, Robot Mass Production
    └── Investment Focus: Terminal Equipment, Application Scenarios

2027-2030: Full Maturity Period
    ├── Full Scene Coverage, Significant Cost Reduction
    └── Investment Focus: Ecological Integration, Global Layout

Risk Type	Specific Performance	Response Strategy
Technology Iteration Risk	Rapid evolution of algorithms may lead to changes in technical routes	Focus on enterprises with core technical barriers
Commercialization Progress Risk	Some tracks are still in the early stage, and profitability needs to be verified	Focus on enterprises with large-scale orders already
Increased Competition Risk	Increasing industry entrants may compress profit margins	Focus on enterprises with first-mover advantages and customer resources
Cost Reduction Falling Short of Expectations	High cost of physical AI equipment limits application promotion	Focus on enterprises with strong supply chain capabilities

: Jensen Huang’s judgment on the “ChatGPT Moment for Physical AI” , mainly reflected in:

1. Scenario Advantages
   : The world’s largest manufacturing base provides rich application scenarios
2. Supply Chain Advantages
   : Efficient industrial chain collaboration and rapid iteration capabilities
3. Data Advantages
   : Massive training data accumulated in complex urban scenarios
4. Policy Advantages
   : Strong national policy support

:

• Short-term (2025-2026)
  : Focus on commercially mature tracks such as intelligent driving simulation and industrial robots
• Mid-term (2026-2028)
  : Focus on high-growth tracks such as embodied intelligence and digital twins
• Long-term (2028+)
  : Lay out leading enterprises across the industrial chain to benefit from ecological maturity dividends

:

1. Upstream Computing Power Layer
   : Domestic GPU breakthroughs, focus on Moore Threads
2. Midstream Platform Layer
   : Digital twins and simulation engines, Fantuo Digital Creation, 51WORLD
3. Downstream Application Layer
   : Industrial robots and intelligent driving, UBTECH, BYD, XPeng Motors

---

[1] Guancha.com - “Jensen Huang: The ChatGPT Moment for Physical AI Has Arrived” (https://www.guancha.cn/qiche/2026_01_06_802929.shtml)

[2] 36Kr - “Jensen Huang Sets the Tone, Physical AI Sounds the Bugle” (https://m.36kr.com/p/3629230900643080)

[3] Economic Daily - “Jensen Huang Calls for Physical AI Implementation, This Disruptive Innovation ETF Leads the Charge” (https://money.udn.com/money/story/5607/9249982)

[4] Business Next - “Jensen Huang: The ChatGPT Moment for Physical AI Has Arrived!” (https://www.bnext.com.tw/article/89703/nvidia-ces-2025)

[5] The Paper - “China-US Competition in the Physical AI Track: Consensus and Surpassing Between XPeng and Tesla” (https://m.thepaper.cn/newsDetail_forward_32291112)

[6] Eastmoney - “Intelligent Driving and Embodied Intelligence Take the Lead, Fantuo Digital Creation’s Leading Path in Digital Twins” (https://caifuhao.eastmoney.com/news/20260102220532348479770)

[7] The Paper - “After NVIDIA, the Infrastructure Battle for Physical AI Has Begun” (https://m.thepaper.cn/newsDetail_forward_32206654)

[8] 21st Century Business Herald - “CES 2026 Witnesses AI Ecological Changes, Chinese Manufacturers Enter the Global Core Camp” (https://www.21jingji.com/article/20260107/herald/f1bce8e64fa75c8da92051abb6174441.html)

[9] Eastmoney - “Physical AI, A New Track” (https://caifuhao.eastmoney.com/news/20260103154505641748870)

[10] OFweek AI Network - “Jensen Huang’s Leather Jacket is Cool, But Factory Ledgers Don’t Recognize Physical AI” (https://m.ofweek.com/ai/2026-01/ART-201700-8420-30678402.html)

[11] 36Kr - “2025 for Humanoid Robots: Value Verification, Capital Boom, and Crossroads” (https://m.36kr.com/p/3618801339483396)

[12] Deloitte Official Website - “Deloitte Asia-Pacific Physical AI Lab Unveiled in Shanghai” (https://www.deloitte.com/cn/zh/about/press-room/asia-pacific-physical-ai-lab.html)

[13] Founder Securities - “Comment on the Surge of Intelligent Vehicle ETF” (https://www.nbd.com.cn/articles/2026-01-06/4209200.html)

[14] Caifuhao - “4000-Word In-Depth Research Report on UBTECH” (https://caifuhao.eastmoney.com/news/20260104025908670319240)

[15] BOC International - “Artificial Intelligence Enters the Physical AI Era, Supply Chain Stockpiling is Expected to Accelerate” (https://finance.sina.com.cn/stock/t/2026-01-07/doc-inhfnktr4729826.shtml)

---

:

1. Market Size Analysis Chart
   - Shows the forecast comparison of global and Chinese physical AI market sizes, industrial chain value distribution, and advantage comparison of China in various physical AI fields

2. Investment Radar Chart
   - Evaluates each investment field from four dimensions: short-term potential, long-term potential, policy support, and industrial chain maturity

3. Investment Timeline Chart
   - Evaluation of physical AI industrial development stages and investment maturity

4. Comprehensive Analysis Chart
   - Industrial chain investment value distribution, China-foreign development comparison, market growth trajectory, and risk-return matrix

5. Investment Portfolio Allocation Chart
   - Suggestions on investment portfolio allocation strategies for the physical AI track