Chery-led $100M round values China's PsiBot at $1.48 billion

The reason a two-year-old Chinese startup closing a $100 million round is worth pausing on, and not just filing under 'another AI unicorn', is who is leading the check. According to Bloomberg's reporting, the lead investor in PsiBot's new round, which values the company at $1.48 billion, is the carmaker Chery Automobile. Not a marquee VC and not a tech giant's strategic arm, but a car company. Alongside Chery, Lens Technology, a sensor maker for Apple and Tesla, is also in.

PsiBot, also known as Lingchu Intelligence, has now raised about $300 million since its 2024 inception. It sells 'world models', which its founder Viktor Wang describes as AI that gives machines a sense of the physical world rather than just the ability to answer questions. Wang has a specific pitch: he told The Next Web that embodied AI is about two years from its 'GPT-2 moment', the point at which the technique visibly starts to work. He also freely admits that 'even the frontier AI labs in Silicon Valley such as OpenAI and Meta don't have good data' for the physical world, which is a candid way of framing why he is raising to build his own pipeline of custom gloves and humanoid rigs.

The Chery-as-lead detail is the part I keep coming back to. Chinese carmakers are increasingly buying strategic positions in the AI layers they need before merchant vendors price them out. If world models really do become the substrate for both robots and self-driving cars, as Wang argues, Chery is effectively pre-buying the software brains its next generation of vehicles and logistics partners will need. Lens Technology's participation, given who it already ships sensors to, hints at the same logic on the hardware side.

The honest caveat is that this is a Bloomberg scoop about a round that is close to being finalized, so take the specific numbers as reported, not settled. What the reporting doesn't give you is any exclusivity terms with Chery, how much of PsiBot's data pipeline is actually running today versus on a slide, or the burn rate implied by raising roughly $300 million inside two years. Wang's own admission that even the American frontier labs lack good embodied-AI data is itself an argument that nobody in this race is close to a defensible moat yet.

The forward-looking piece is who benefits if Wang is roughly right. Chinese logistics and fiber-optic cable firms, where PsiBot says it is already testing, get first pass at a home-grown world-model stack. Chery gets an early strategic foothold in autonomy before the price of admission climbs further. And the broader signal, that the interesting AI capital in China is now flowing into the physical stack rather than another chatbot, is the part worth watching.