Astribot Humanoid Robots Score a Billion-Yuan Raise

A Shenzhen startup building Astribot humanoid robots has landed one of the largest robotics raises of the year, announcing in October 2026 that it secured about RMB 1 billion (roughly $149.4 million) in fresh financing at an RMB 15 billion valuation. According to TechStartups' daily funding roundup, the round was led by Shenzhen Investment Holdings-linked capital, with Hongtai Fund, the Ninghai Xingtaihe Fund, Shanheng Capital, and returning investor Xianmu participating. The financing is the company's fourth close in six months — an unusually fast cadence even in China's red-hot embodied AI sector.

The company, known internationally as Astribot and in China as Stardust Intelligence, builds cable-driven humanoid robots aimed at commercial services, research, and education, with households as the long-term goal. Its pitch is that AI for the physical world cannot be trained on a screen alone: the robot body, the embodied operating system, and the AI models must be designed together. Astribot calls this its "Design for AI" philosophy, and investors appear to be buying it — repeatedly.

Why cable-driven robots are having a moment

Most humanoid robots move on rigid joints: motors bolted directly at each joint, stiff and powerful but unforgiving in contact. Astribot takes a different route. Its robots transmit force through cables running from motors through the limbs, a bit like tendons in a human arm. The result is a machine that yields to bumps, grips gently, and handles soft or fragile objects more naturally. For robots expected to fold laundry, sort parts, or assist in a lab, that compliance matters more than raw strength.

The approach got its biggest stage yet at IROS 2026, the international robotics conference held in Pittsburgh in late September and early October, where the company introduced its T1 platform to North America for the first time. According to industry coverage of the debut, the wheeled humanoid stands about 1.55 meters tall, weighs roughly 66 kilograms, has 23 degrees of freedom excluding its grippers, and can carry up to 5 kilograms per arm. The T1 launched domestically at 89,900 yuan, with U.S. pricing starting at $18,000 and units available for immediate delivery.

The T1 pairs that body with Astribot's full software stack: its Lumo family of embodied AI models and a purpose-built operating system with developer tools, APIs, and VR-based teleoperation. Equipped with the newer Lumo-2 base model, the robot has demonstrated autonomous tidying of miscellaneous items into a backpack, according to the company's announcement. Developers are already experimenting on the platform: at a company hackathon in Beijing that concluded September 21, fifteen teams built and demoed more than ten working applications on the T1 in thirty-six hours, the announcement said.

The funding frenzy — and the catch

Astribot's raise is a symptom of a much larger boom, and the trajectory has been quick: founded in 2022, a Series B totaling more than a billion yuan at a valuation above ten billion yuan reported in June 2026, a North American debut at IROS in late September, and now the new round. Caixin reported on that June wave, which also saw rivals such as Spirit AI pull in about 1.5 billion yuan in a Series A+ round. Industry data suggests the market is moving from demos toward shipments: AgiBot, another Chinese contender, led global humanoid robot shipments in the first half of 2026, with the overall market approaching roughly twenty-five thousand units. The contest is not limited to China: Agility Robotics is taking its humanoid robot toward public markets, and Mecka banked a Series B round to teach robots human motion.

The company itself claims orders involving thousands of robots and manufacturing capacity measured in the tens of thousands of units. Those figures should be read carefully: capacity is not output, and an order book is not a delivery record. Still, the direction of travel is clear — investors are betting that humanoids are crossing the line from laboratory showcase to deployable product.

There is a serious counterpoint, though. As ainvest argued in an analysis of the company's financing, the real constraint on this industry may not be capital but compute. Training the next generation of generalist robot models requires enormous GPU resources, and domestic training capacity in China is a potential bottleneck. The involvement of state-backed investors could mean compute access is bundled into the deal, but the company has not disclosed specific GPU or training-infrastructure commitments. Until it does, that gap stays open.

What this means beyond the balance sheet

For developers and researchers, the news that matters most may be the price tag rather than the valuation. Astribot humanoid robots in the low tens of thousands of dollars, sold with open developer tools, puts serious physical-AI experimentation within reach of labs and startups that could never afford six-figure research platforms. That accessibility is exactly what the company is selling: a robot designed from the start to collect the interaction data its models need to improve.

For everyone else, Astribot's raise is another data point in the biggest hardware bet of the AI era: that the next frontier is not a smarter chatbot but a machine that can tidy a room, assist a technician, or keep a production line moving. Whether cable-driven bodies become the standard or a footnote will be decided not in funding announcements but on factory floors and in living rooms. The money pouring into Shenzhen suggests the company intends to be there when the answer arrives.