Wheels over legs AI2 raises 735M at 3B valuation

A wheeled humanoid just closed a 735 million dollar round.

AI² Robotics, the Shenzhen based hardware and AI developer behind AlphaBot, raised fresh capital that pushes its valuation above 50 billion RMB, roughly 2.8 billion dollars. The funding comes from a diverse mix of state backed funds, corporate players, and financial institutions, including the National Small and Medium Enterprises Development Fund, Sino Biopharmaceutical, the Moutai Group, CICC Capital, and GSR Ventures. Deployment data shows the round cements AI2’s status as a leading force in China’s fast moving physical AI sector, where the speed of investment often tracks government policy and enterprise demand for automation in public and semi public spaces.

AI² Robotics builds a wheeled mobile manipulator with a humanoid torso and five fingered hands. The design leans away from legged locomotion in favor of mechanical simplicity, durability, and lower production costs. The robot features more than 34 degrees of freedom and a custom waist leg lifting mechanism that can reposition the upper torso across a range of poses. The company emphasizes that while the platform cannot climb stairs or traverse rugged terrain, it is cheaper to produce, mechanically robust, and faces fewer regulatory hurdles when deployed in public spaces. The AlphaBot operates on Alpha Brain, AI²’s vision language action stack that enables real time spatial reasoning, environmental understanding, and action selection.

For plant managers, CFOs, and field operators weighing automation investments, the funding round signals more than a flashy demo. The case for wheeled humanoids rests on a clear ROI proposition: lower upfront production cost, lower ongoing maintenance, and easier regulatory paths compared with more complex legged designs. In AI2’s framing, the wheel based platform can be deployed in controlled public environments with fewer compliance frictions, while offering the dexterity of a humanoid upper body to interact with tools, touch screens, and distributed sensors. The Alpha Brain VLA stack is central to that capability, tying perception to action through a compact, on board compute pipeline designed to support real time tasks without constant cloud dependence.

Yet the execution realities are worth watching. The absence of published cycle times or throughput metrics means operators will need to anchor expectations to integration plans and field trials rather than claims from a capability brochure. The company’s hardware choice implies certain operational constraints: the wheel base favors steady, repeatable motion on smooth surfaces but sacrifices the stair climb and rough terrain capabilities that some use cases require. Deployment data shows that the tradeoff is intentional: speed to deployment, simpler maintenance, and smoother regulatory engagement in public spaces can shorten time to value for pilot programs, while longer term ROI will hinge on how quickly sites can integrate AI2’s Alpha Brain with existing automation stacks, sensors, and safety systems.

From a practitioner perspective, several constraints and tradeoffs emerge. First, the mobility choice is a design decision tied to cost, durability, and deployment speed, not ultimate versatility. Second, the 34 DOF and the lifting mechanism suggest a capable upper body but a need for disciplined calibration and safety testing when used near human workers or in crowded spaces. Third, realization of ROI will depend on integration requirements: how easily Alpha Brain VLA can be wired into current workflows, the data interfaces, and the acceptability of on site processing versus cloud based inference. And finally, as deployment expands beyond prototyping, operators should monitor how regulatory environments evolve for public space robots and how the business model aligns with enterprise grade service agreements, spare parts availability, and long term support.

The funding round underscores a broader industry trend: capital flooding into physical AI systems that promise tangible, near term ops improvements. If AI2 can translate its architectural strengths, robust wheels, flexible upper body, and a unified AI stack, into reliable, scalable deployments with measurable uptime, cycle time reductions, and predictable throughput, it will be a meaningful proof point for automating routine public interactions and light industrial tasks.