The 2025 New Automotive Cooperation Ecosystem Exchange was held in Suzhou from December 5 to 6. Song Yang, Founder and CEO of IMOTIONTECH, delivered a keynote speech.
Song Yang noted that in the short term, China's annual automobile production stands at 30 million units, while embodied intelligent robots reached only 500,000 units last year—one-sixtieth of the automotive volume. In the long run, autonomous vehicles are essentially wheeled robots, a subset of robotics, and their numbers are expected to grow significantly. The integration and development of these two industries present both immediate and long-term challenges.
He pointed out that large language models, as foundational models, face intermediate leap issues whether applied to autonomous driving or robotics. Taking multimodal VLA as an example, actions performed in one room struggle to generalize to another scenario, requiring costly data collection. Similarly, adding dimensions like gravity to world models drastically increases computational demands and costs, alongside challenges like power consumption and heat dissipation—key industry pain points.
However, Song Yang remains optimistic. He believes that by leveraging China's vast industrial base and diverse scenarios, these environments and data can propel AI development. Through an industry-driven AI approach, China is poised to achieve more breakthroughs in artificial intelligence ahead of others.
Comments