CoinW Research Institute
Summary
As AI transitions from the digital world to the physical world, humanoid robots are becoming a new focal point in global technological competition. Tesla's Optimus, Figure AI, and Unitree Technology represent three different paths, racing around AI models, manufacturing systems, and application ecosystems. In this race, Unitree Technology, leveraging China's supply chain, cost control, and rapid iteration capabilities, shows a unique path for China's robotics industry. The development of the robotics industry is also driving market attention toward core links such as reducers, sensors, and precision components, further extending to emerging directions like AI infrastructure and DePIN. When physical AI Agents intersect with DePIN infrastructure, a new funding narrative is also forming.
1. Three Videos, One Signal: AI is Gaining a Body
Recently, three things collided on social media. Unitree Technology released a new video of its G1 humanoid robot running on complex terrain, with comments like "this is no longer science fiction" flooding the comment section. Almost simultaneously, Figure AI updated its footage of training in a BMW factory, showing robots learning to place parts into racks. Meanwhile, Musk retweeted a new clip of Optimus walking in Tesla's factory on X, once again mentioning the mass production timeline. These three events happened in the same week; it is not a coincidence. It illustrates a consensus rapidly forming: large models give AI the ability to understand the world, while robots provide AI with a "body" to enter the physical world. As the AI competition in the digital realm becomes intense, the battlefield for tech giants is silently shifting to the physical world.
2. Three Strong Players: Three Routes, One Anxiety
Currently, in the global humanoid robot field, Tesla, Figure AI, and Unitree Technology represent three different development routes. Tesla follows an ecosystem integration path, relying on the large-scale real-world data, AI training capabilities, and mature manufacturing systems accumulated from its autonomous driving business, giving Optimus strong software and hardware synergy advantages. However, robots still face challenges in moving from technical demonstration to large-scale production, as Tesla has repeatedly adjusted its product timelines, and the market remains cautious about its commercialization pace. Figure AI represents the technological route of "large models driving robots." Backed by AI industrial capital and a large model ecosystem, Figure aims to upgrade from traditional automation equipment to general intelligent robots by enhancing the understanding and decision-making capabilities of robots. At the same time, the company is collaborating with manufacturing enterprises like BMW to accelerate the exploration of industrial scene implementation. On the other hand, high valuations and rapid fundraising also mean that the market places higher demands on its future commercialization capabilities, with time needed to validate technology breakthroughs and practical applications.
In contrast, Unitree Technology's development path emphasizes productization and engineering optimization, utilizing China's mature supply chain, cost control capabilities, and rapid iteration advantages to accelerate the landing of robotic products. With the G1 humanoid robot and four-legged robot as representatives, Unitree has already formed a certain application foundation in research, exhibitions, education, and other scenarios. However, it should be noted that the current intelligence of humanoid robots is still in the early stages, and the Scaling Law of the VLA (vision-language-action) model has not yet been fully validated. The industry still faces challenges in data accumulation, model capabilities, and the expansion of commercial scenarios. Three routes, three anxieties: Tesla worries about timelines; Figure worries about matching valuations with real progress; and Unitree is anxious about the intelligence gap in high-end scenarios. However, for the secondary market, a more critical observation is that the selection of core components such as reducers, torque sensors, and precision lead screws among these three companies is highly converging. The key to future robot competition lies not only in who can manufacture robots but also in who can simultaneously solve the issues of intelligence, cost, and commercialization. For the industry chain, the opportunities brought by robotics development are not limited to whole machine enterprises. As different technical routes gradually advance, core components such as reducers, torque sensors, precision lead screws, and robotic control systems will continue to be market focal points. However, it is essential to recognize that the current robotics industry is still in its early stages, and the release of value by upstream companies will depend on the downstream commercialization progress. As companies like Tesla Optimus, Figure AI, and Unitree Technology continue to push product iterations, the market is watching which component companies will benefit first from the robot scale trend. For the secondary market, the focus of the robotics industry chain is not just on one whole machine company but on the long-term ecological opportunities formed around core components, intelligent manufacturing, and AI infrastructure. Yet, moving from technological breakthroughs to productive applications in robotics will still take time, and the pace of industrial development and order fulfillment remains a key focus for future observation.
3. Re-evaluating the Robotics Industry Chain: Finding the Future's "Shovel Seller"
Unitree Technology has reduced the price of humanoid robots to under 100,000 yuan, promoting productization and cost reduction in robotics. Looking back at the history of consumer electronics and new energy vehicles, price drops are typically an important driving factor for industry maturation, but humanoid robots are still in the exploration stage of commercialization. The core issues currently facing the industry are not only manufacturing costs but also the level of intelligence in robots, the expansion of application scenarios, and the verification of return on investment. Compared to exhibition and research scenarios, truly entering high-value scenarios such as factory production, logistics transportation, household chores, and companionship still requires further validation. From the current supply chain disassembly perspective, the execution layer occupies the largest share of the BOM cost for humanoid robots. Harmonic reducers, planetary roller lead screws, frameless torque motors, and hollow cup motors are the core components supporting robotic movement. In the perception layer, six-dimensional force sensors, IMU inertial measurement units, and visual modules determine whether robots can operate stably in real environments. The commonality of these links is high technical barriers and significant space for domestic substitution, and once whole machine manufacturers start to scale, performance elasticity will be directly reflected in the financial statements. However, it is essential to point out that the current robotics industry is still in its early stages, and the value release of supply chain companies will depend on the downstream commercialization pace. As manufacturers like Tesla Optimus, Figure AI, and Unitree Technology continue to advance product iterations, the market is focusing on which component suppliers can benefit first from the scaling trend of robots. For the secondary market, the points of interest in the robotics industry chain are not solely on one whole machine company but on the long-term ecological opportunities created around core components, intelligent manufacturing, and AI infrastructure. However, the transition from technical breakthroughs to productive applications in robotics still requires time, and the rhythm of industrial development and order fulfillment is still a focus for future observations.
4. From Robots to AI Agents: A New Interface in the Physical World
If we raise our perspective from the industry chain, humanoid robots are gaining a brand new definition: they are AI Agents in the physical world. Over the past year, discussions around AI Agents in the cryptocurrency market and tech circles have mainly focused on the digital space, such as automated trading, content generation, and on-chain task execution. These Agents handle information and operate on servers and blockchains. However, humanoid robots are different; they deal with objects, terrain, and collaborative relationships in the physical space. Large models provide them with the ability to understand instructions, visual and tactile sensors give them the ability to perceive their environment, and actuators enable them to complete tasks. When these three elements come together, AI is no longer just a dialog window on a screen but a physical intelligent agent capable of entering factories, warehouses, and logistics centers. For participants in the cryptocurrency market, this narrative is not unfamiliar. In the second half of 2023 to 2024, concepts of AI Agents represented by Virtuals and ai16z once sparked a wave of fervor. Now, as the hype around digital AI Agents enters a phase of differentiation, funds are instinctively seeking the next carrier for AI. Physical AI Agents, also known as embodied intelligence, precisely embody this expectation.
5. DePIN and Robotics: Financialization of Physical World Data
The development of the robotics industry is also giving rise to a cross-cutting area that is easily overlooked: the financialization of data from the physical world. Training a robot capable of autonomous action in real environments requires a vast amount of real-world data, including grasping actions, walking postures, collision recovery, and scene interactions. This data is scattered across factories, warehouses, and testing fields around the world, making acquisition costly and lacking effective rights confirmation and circulation mechanisms. This is exactly where blockchain and DePIN infrastructure could potentially intervene. Based on current industrial logic, at least three intersections are worth observing.
The first is the crowdfunding and confirmation of training data. If factories and developers around the world can upload robot training data through a decentralized network and receive token rewards for their contributions, the cost curve for acquiring data will be significantly flattened, akin to a DAO in the data field. The second is the distributed supply of simulation computing power. Training robots not only requires real data but also extensive simulation in virtual environments, and the DePIN network can organize idle GPU computing power to provide inexpensive training resources for small and medium robot enterprises. The third is the on-chain identity of the Internet of Things for robots. When robots are deployed on a large scale in factories, the operational data, maintenance records, and task logs of each device can become a reliable data source for supply chain finance and asset leasing. From this perspective, the value of the robotics industry will not be confined to the hardware itself but will extend to a whole new set of infrastructure around computing power, data, and the confirmation of physical assets.
6. Funding Landscape: Three Streams of Capital, One Direction
Where the money goes has always been the most honest signal. The current surge of funding in the robotics sector can be observed from at least three levels. In the primary market, Figure AI has experienced several rounds of valuation jumps over the past year, while Unitree Technology has backing from top institutions like Sequoia China and Meituan Longzhu, and "grabbing quotas" in the VC circle has become an open secret. The enthusiasm in the primary market often leads the secondary market by half a step; when these funds cluster around whole machine manufacturers, they naturally spill over to tradeable supply chain segments. The secondary market's reaction has already been quite evident. The scale of domestic robotics-themed ETFs has been continuously expanding over the past few months, and the trading volume center for reducers, torque sensors, and precision components has significantly risen, with institutional research frequency hitting new highs in recent years. These movements are not necessarily short-term speculation but rather funds positioning themselves ahead of a potentially years-long industrial trend. More intriguing for crypto users is the migration of capital at the cryptocurrency market level. The market capitalization share of AI segments in the cryptocurrency space is undergoing structural changes, with funds spreading from early "pure text AI concepts" to "physical AI" and "DePIN infrastructure." As Silicon Valley venture capital bets on the robotics field and A-share funds compete for component sectors, smart capital in the cryptocurrency market is also seeking different expressions of the same grand narrative. Three streams of capital, three languages, but pointing in the same direction.
7. Consensus Runs Faster than Facts
Unitree Technology standing at the forefront essentially signifies an important spatial shift in the AI industry, moving from the digital world to the physical world. This shift will not be accomplished in a single day, but the consensus it forms often races ahead of the facts themselves. For traders, the critical aspect is not precisely predicting which whole machine manufacturer will ultimately prevail but identifying where funds begin to congregate at certain nodes and which segments of the industry chain have already secured their entry tickets. Whether it is the re-evaluation of the supply chain in the traditional financial market or the experimental intersection of physical AI and DePIN in the crypto world, humanoid robots are becoming the central narrative for the next stage. The first chapter of this story has already opened, and the hidden lines on the funding landscape are becoming increasingly clear.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
