Xianglu Robotics Raises Funding to Scale AI Chefs

Xianglu Robotics announced a RMB 300 million (about USD 43.5 million) funding round on April 5, 2026, which the company frames as validation from state-owned capital for industrial robotics commercialization. CTO Jiancheng Yang described the raise as accelerating the company’s shift from technical validation to large-scale, international deployment of AI cooking robots.

Xianglu claims to be the first company to implement and scale AI-generated recipes within a closed-loop technology chain: real-world kitchen data → AI recipe generation → robotic execution. That pipeline combines data engineering (real kitchen telemetry and process data), ML-driven recipe synthesis, and tightly coupled robotics control and actuation to reproduce culinary results at scale. The stacks implied are cross-disciplinary: perception/control for execution, model-driven recipe generation, and operational systems for deployment and lifecycle support.

Xianglu reports deep penetration across three verticals: fast food, formal dining, and group catering. Manufacturing has been scaled into new lean facilities with a stated peak capacity of 10,000 units per month, supported by two intelligent production bases in South China and East China. The company has established 90 regional rapid-delivery warehouses within China and assembled a team of nearly 100 AI chef coaches, engineers, and trainers to support full-lifecycle operations. Internationally, Xianglu says equipment is deployed in dozens of countries, including the United States, Australia, France, Singapore, Japan, and South Korea, and that a digital Chinese recipe library helps overseas restaurants maintain consistent quality.

This is a concrete example of ML models moving from lab proofs-of-concept into integrated industrial systems where data pipelines, model outputs, and real-world actuators must be engineered together. The claimed closed-loop approach underscores the importance of production-grade data collection, domain-specific model tuning, and operational teams (coaches/trainers) to close the human-in-the-loop gap. For ML engineers and robotics teams, Xianglu’s scaling highlights challenges around reproducibility of sensory-driven tasks, packaging ML into deterministic processes, and supporting distributed fleets.

Verify third-party deployments and performance benchmarks outside China; examine the recipe-generation models’ safety, variability control, and cultural/localization handling; and monitor how state-owned capital involvement affects international expansion dynamics.