Lingbo Technologies, the robotics subsidiary of Ant Group (HK: 6688) — the fintech giant behind China's dominant mobile payment platform Alipay — has initiated its first external funding round with a target of raising 1.5 billion yuan (approximately $206 million). The company also plans a second funding round before the close of 2026, which would mark the fastest fundraising trajectory of any startup in China's robot-brain sector.

Strategy Centered on Robot "Brains"

Since its inception, Lingbo has been entirely funded by its parent company Ant Group. The decision to now seek outside capital comes as Ant faces growing budget pressure on computing resources while simultaneously investing in AI initiatives across other business units. Analysts note that Ant has already invested more capital into Lingbo than a typical top-tier startup would raise in its early funding rounds.

Lingbo's core focus is developing the software and AI models that serve as the "brain" for robots — the systems that enable machines to perceive their environment, make decisions, and execute actions. The company's strategy, which it calls "one brain, many machines," centers on building a single powerful cognitive system compatible with a wide range of robot bodies. This approach mirrors that of U.S.-based Physical Intelligence, a company valued at over $11 billion despite manufacturing few robots of its own. The parallel is notable: both companies are betting that the highest-value layer in robotics will be generalized intelligence software rather than hardware, much as operating systems and apps outweighed device manufacturing in the smartphone era.

Lingbo adapts technology originally developed for Alipay, Ant Group's mobile payment platform. These systems, designed for high-speed transaction processing and real-time decision-making under uncertainty, are being repurposed to help robots interpret and interact with the physical world.

LingBot Model Stack and Industry Traction

Lingbo's model stack, known as LingBot, spans vision, depth perception, mapping, and video understanding. Its flagship model, LingBot-VLA 2.0 — where VLA denotes Vision-Language-Action, an architecture that fuses visual perception with language understanding to generate physical commands — was trained on approximately 60,000 hours of real-world robot task data. The model is currently compatible with more than 20 different robot configurations from 17 brands, including Unitree and AgiBot.

On GitHub, Lingbo's models have attracted nearly 30,000 stars — roughly 2.7 times more than the second-ranked domestic team.

Investor interest has been robust. According to reports, some investment firms reached out before the funding round was officially launched. At the World Artificial Intelligence Conference in Shanghai, investors visited Lingbo's booth directly to evaluate the technology and express interest.

Approximately half of all capital flowing into embodied intelligence is now directed toward companies building perception and decision models — the "brain" — rather than physical hardware, and that proportion continues to rise. This capital split reflects a broader industry conviction that hardware is increasingly commoditized while adaptable AI models remain the scarce, defensible asset.

The Chinese government has designated robotics and "embodied intelligence" as top priorities in its 15th Five-Year Plan, which provides billions of dollars in subsidies and industry support.

Ant Group itself has led funding rounds for at least 12 robotics companies since early 2025, including a $73.58 million round for humanoid startup Zeroth.

Valuation Hinges on Proven Scalability

Despite rapid growth and mounting investor interest, Lingbo's Chief Scientist Shen Yujun has been candid about the challenges ahead. He stated that robotics has not yet experienced its "GPT-1 moment" because the field still lacks the millions of hours of data required for a genuine breakthrough. For context, the largest language model training runs draw on trillions of tokens of text — orders of magnitude more data than exists for any single robotics AI today.

Following the fundraising, Lingbo will remain Ant Group's primary business in embodied and physical AI, with the parent company continuing to provide support in infrastructure, data, and deployment.

According to insiders, the company's final valuation will depend on multiple factors, including whether LingBot models can demonstrate clear cross-platform capability across diverse robot types and tasks. The company must also prove readiness for large-scale deployment and establish a functional "data flywheel" — a cycle in which increased usage generates more data, improving model performance and driving further adoption. Finally, Lingbo will need to transition from an internal Ant Group initiative into a globally recognized platform for physical AI.