In recent years, Embodied AI is becoming one of the most concerned technology trends in the field of artificial intelligence, and its application prospects in industrial manufacturing, service robots, special operations and other fields are promising. With the continuous influx of capital, the heat of the industry continues to rise. Recently, Sprit AI announced the completion of 528 million yuan of Pre-A round of financing, becoming another star enterprise in the field of embodied intelligence in China. This round of financing was led by Prosperity7 Ventures, a subsidiary of Aramco Ventures, followed by China Merchants Venture Capital, GF Xinde and other capitals, bringing the company's cumulative financing amount to more than 700 million yuan.
In the context of the humanoid robot market size of 38 billion US dollars (about 276 billion yuan) predicted by Goldman Sachs in 2035, can the blessing of capital really accelerate industrial breakthroughs? Can a Tsinghua-backed technical team become a key variable in leveraging the market? China Exportsemi will conduct an in-depth analysis from the four dimensions of capital, technology, market and commercialization path.
Accelerated capital layout: embodied intelligence ushered in a golden period?
Since its establishment in February 2024, Sprit AI has completed four rounds of financing in just over a year, which is inseparable from the capital market's long-term optimism about embodied intelligence. This is not just a single business phenomenon, but an industry-wide trend.
Goldman Sachs noted in its latest report that the humanoid robot market is expected to ship 1.4 million units by 2035, driven by advances in AI technology, declining hardware costs, and growing demand for automation in the industrial and service sectors. According to PitchBook data, since 2023, the global financing scale in the field of embodied intelligence has exceeded 2 billion US dollars, among which Figure AI, Agility Robotics and other American companies have successively received large financing, and the domestic market is also heating up rapidly.
Sprit AI 's technical route is highly similar to Figure AI, a star company in Silicon Valley, that is, it is based on the combination of a general humanoid robot platform and an embodied large model. Compared with solutions such as Tesla's Optimus, this model emphasizes the robot's autonomous perception and decision-making ability of the environment, which is expected to reduce the cost of adaptation to specific scenarios. Because of this, the company has become a key target for capital chasing.
Figure: Sprit AI announced the completion of 528 million yuan of Pre-A round of financing
Technological breakthrough: from the lab to commercialization
The core challenge of embodied intelligence is how to get the robot to truly "understand" and adapt to the physical world, rather than just relying on pre-programmed tasks. Sprit AI 's recently announced self-developed VLA Spirit v1 algorithm realizes long-range operation of flexible objects, and in the official demonstration video, two robotic arms can smoothly complete the task of folding clothes, which is regarded as one of the most challenging tasks in the field of robotics.
At the same time, its first commercial-grade humanoid robot, the Moz1, was officially unveiled. Moz1 uses 26 degrees of freedom (excluding dexterous hands) and is equipped with the world's highest power density integrated force control joint, which is 15% higher than Tesla's Optimus. At the level of intelligent control, Moz1 introduces reinforcement learning and large-scale video data training, so that it can autonomously adapt to the environment and perform tasks without explicit instructions.
However, technological breakthroughs do not mean that commercialization is smooth sailing. Goldman Sachs analysis believes that the production cost of humanoid robots is still as high as more than $100,000, and supply chain bottlenecks make it difficult to reduce the manufacturing cost of key components (such as high-precision force-controlled joints and dexterous hands). In addition, there are still challenges in the adaptation of AI control software, especially the optimization of natural language interaction and self-learning capabilities.
Tsinghua team's academic and industrial integration advantages
One of the core competencies of Chihiro Intelligence is its deep academic background. The company's co-founder, Yang Gao, is an assistant professor at the Institute for Interdisciplinary Information Sciences at Tsinghua University and studied with Pieter Abbeel at the University of California, Berkeley, a leading global expert in reinforcement learning and embodied intelligence. The ViLa algorithm previously proposed by the Gaoyang team has been adopted by Figure AI, which has become an important milestone in embodied intelligence technology.
From the perspective of technology path, the team is exploring the deep combination of embodied intelligence and cognitive AI, so that the robot can understand high-level task requirements, rather than just executing low-level instructions. For example, one of the R&D goals of Sprit AI is to enable Moz1 to undertake tasks such as housework and elderly care in the home environment, which requires not only the improvement of hardware capabilities, but also the leapfrog progress in environmental perception, autonomous decision-making, and natural language processing.
Market potential and commercialization path
According to Goldman Sachs' forecast, humanoid robots are expected to be the first to enter the industrial manufacturing field in 2024-2027, and then gradually penetrate the consumer market between 2028-2031. Currently, the manufacturing and logistics industries have the most urgent demand for humanoid robots. For example, Agility Robotics' Digit robot has been piloted with Amazon for warehouse handling tasks.
In terms of the domestic market, national policies are actively promoting the development of the robot industry. The "14th Five-Year Plan for the Development of the Robot Industry" released in 2023 clearly states that it is necessary to strengthen the core technology research of humanoid robots and promote relevant application demonstrations. If start-ups such as Sprit AI can take the lead in the industrial manufacturing scene, they will have the opportunity to seize the opportunity.
However, it is worth noting that the commercialization of embodied intelligence still faces many challenges:
Cost control: At present, the cost of hardware is high, especially for core sensors and power systems, and it is difficult to reduce it significantly in the short term.
Application scenario landing: Although the market demand is large, the actual landing scenarios are still relatively limited, and enterprises need to clarify the market segmentation strategy.
Closed-loop data optimization: Robots need to learn in real-world scenarios for a long time, rather than relying only on laboratory data, which puts forward higher requirements for data collection and optimization.
Conclusion: The triple test of capital assistance, technology cashing, and market landing
Sprit AI 's financing of 528 million yuan is another increase in the capital market's commitment to the embodied intelligence track, but what really determines the development of the industry is still the maturity of the technology landing ability and business model. From the perspective of global trends, embodied intelligence is entering a critical stage from the laboratory to industrialization, and how to strike a balance between technology, market, and cost is the core challenge for all enterprises.
For the Tsinghua team, its academic background is undoubtedly an advantage, but whether it can transform cutting-edge research results into products that can be implemented on a large scale will determine the future development height of Sprit AI. In the face of the humanoid robot market with a scale of 154 billion yuan, this is not only a contest of technology, but also a test of time. In the context of the accelerated evolution of global AI technology and industrial landscape, if Chinese embodied intelligence companies can seize the window period, they may have the opportunity to occupy a place in the global market.
