The wave of humanoid robots is sweeping in, with major tech giants entering the fray to compete for the high ground in future intelligent manufacturing and services. HSBC expects that under the baseline scenario, the market size of humanoid robots will grow from USD 900 million in 2025 to USD 73 billion by 2034. However, since humanoid robots have not yet entered mass production, the supply chain may still undergo changes.
On February 28, HSBC analysts Helen Fang, Kenneth Chin, and Jennifer Pan released a research report focusing on three leading humanoid robot companies: Tesla, Unitree Technology, and Figure AI. These three companies not only have unique characteristics in robot design and functionality, but their underlying supply chains also exhibit diversity and intense competition.
<h2>Who supplies the &#34;Future Warriors&#34;?</h2>
The research report shows that in the bill of materials (BOM) for humanoid robots, hardware components such as drives, motors, reducers, ball screws, and bearings account for as much as 55% of the value share, far exceeding sensors (17%) and AI chips/software (13%).
Currently, humanoid robots have not yet entered large-scale production, and the supply chain landscape remains uncertain. However, HSBC has identified some key Chinese suppliers that provide samples or small batch production for leading companies like Figure AI, Unitree, and Tesla.
Among them, Figure AI has made significant breakthroughs in robot training efficiency. The Helix vision-language-action (VLA) model developed by the company greatly enhances training efficiency and is expected to accelerate the commercialization process of humanoid robots. However, battery life remains a critical bottleneck limiting the large-scale application of humanoid robots.
In the supply chain of Figure AI, Xusheng Group provides magnesium alloy shells, Everwin Precision provides joints, bearings, and sensors, and Leaderdrive provides harmonic reducers.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/ec52d1aa-df06-4f15-9428-28b75b1c20c1.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/ec52d1aa-df06-4f15-9428-28b75b1c20c1.png"/>
In the supply chain of Unitree Technology, MOONS and Zhaowei Machinery &amp; Electronics provide coreless motors, Leaderdrive provides harmonic reducers, and CSB Bearing provides bearings and ball screws.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/24ede01d-5f2b-4d5e-8f8e-3bafcd1b185b.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/24ede01d-5f2b-4d5e-8f8e-3bafcd1b185b.png"/>
In Tesla's Optimus supply chain, Sanhua and Tuopu serve as first-tier suppliers providing actuator assemblies, while Beite, XCC, Best, and several other companies provide ball screws
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/e5607cba-2e75-4321-8576-ced4a14484f9.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/e5607cba-2e75-4321-8576-ced4a14484f9.png"/>
<h2>Figure AI's Helix model initiates &#34;efficiency revolution,&#34; humanoid robot market size to reach $73 billion</h2>
The intelligence of humanoid robots relies on efficient training methods. The report points out that in traditional imitation training, it takes about 500 hours of data for robots to master a new skill (such as folding clothes), and each new skill requires repeating the same process, making large-scale completion difficult.
However, Figure AI's Helix VLA model takes a different approach by connecting visual language models (VLM) and robotic action models, achieving more efficient training. This model requires only 500 hours of supervised dataset training to enable robots to master multiple tasks without the need for individual fine-tuning for each task. This means humanoid robots can learn new skills faster and adapt to different application scenarios.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/31be7a2b-f526-4822-9241-4578376e1308.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/31be7a2b-f526-4822-9241-4578376e1308.png"/>
However, HSBC also pointed out that for humanoid robots to achieve longer battery life, they typically rely on cloud computing, which can increase latency. Therefore, battery technology remains a key bottleneck for the commercialization of humanoid robots.
In addition, the cost and market size of humanoid robots are also focal points for investors. HSBC conducted scenario analysis on the mass production of humanoid robots and predicts that the market size will grow to $73 billion by 2034 (base scenario), with a compound annual growth rate of up to 63%.
The report believes that as production scales up, the cost of humanoid robots will rapidly decline. It is expected that by 2032, the cost of each robot will decrease from about $58,000 to $20,000, with an average annual decline of 11%. This decline is consistent with the historical cost reduction trends of industrial robots and electric vehicles.
<blockquote>
&#34;As production increases, strong economies of scale will quickly reduce costs.&#34;
</blockquote>

TSLA

HSBC pointed out that in the bill of materials for humanoid robots, hardware components such as drives, motors, reducers, ball screws, and bearings account for as much as 55% of the value share, far exceeding sensors (17%) and AI chips/software (13%). The analysis predicts that under the baseline scenario, the market size for humanoid robots will grow from USD 900 million in 2025 to USD 73 billion in 2034

- The humanoid robot market is projected to grow from $900 million in 2025 to $73 billion by 2034, according to HSBC analysts.  
- Key components in the supply chain include hardware like motors and gears, which account for 55% of the value, while advancements in training efficiency, particularly by Figure AI's Helix model, are expected to accelerate commercialization.  
- However, challenges such as battery life and production costs remain critical, with costs potentially decreasing from $58,000 to $20,000 per robot by 2032 as production scales up.

Infographic of the Hottest Robot Suppliers - Tesla, UTree, and Figure Chain