Meituan’s Strategic Gambit
The relentless pursuit of the next frontier in artificial intelligence has propelled Meituan, China’s delivery and retail behemoth, into a high-stakes game of diversification within the AI robot sector. Its recent exclusive lead investment in Shenzhen-based Ziyuanliang Robotics – injecting hundreds of millions into the A-round funding – signals more than just capital deployment; it underscores a calculated strategy to avoid missing the next breakout star like Yushu Technology, while navigating the still-fluid technological pathways of embodied intelligence.
Scattering Bets Across a Fractured Landscape
Unlike traditional sectors with established blueprints, the AI robot domain, particularly humanoid robotics, lacks a definitive technological consensus. Meituan’s investment portfolio reflects this ambiguity. Within months, it backed Ziyuanliang Robotics, championing an end-to-end “universal embodied large model” (WALL-A) trained primarily on real-world data for complex tasks like zipping jackets and folding clothes. Concurrently, it doubled down on Galaxy General Robots, a firm heavily reliant on synthetic simulation data migrated to physical machines – a starkly different approach.
This multi-pronged strategy is deliberate. “Meituan is essentially placing dual bets,” revealed an executive at a rival robotics firm. “With the technological path still ‘a hundred flowers blooming,’ and no undisputed leader in embodied intelligence, especially for humanoid AI robots, spreading investments mitigates the risk of backing the wrong technical horse and missing monumental market opportunities.” The ghost of Yushu, initially underestimated before its rise, looms large. “Everyone wants to find the next Yushu,” the executive added. “Not putting all eggs in one basket is prudent for investors who don’t want to miss that potential.”
The “Scenario Empowerment” Imperative
Beyond technological hedging, Meituan’s AI robot investments are tightly bound to its core “retail + technology” strategy and real-world operational needs. “Meituan currently prioritizes the coupling potential between AI robots and its existing business scenarios,” shared a source close to the company, emphasizing CEO Wang Xing’s pivotal role in these investment decisions via Meituan Strategic Investment and Meituan Longzhu. While practical AI robot applications remain largely in the proof-of-concept phase, Meituan aggressively seeks opportunities where these machines can enhance efficiency within its vast ecosystem – from last-mile delivery and warehouse logistics to in-store services.
This “scenario empowerment” philosophy drives post-investment collaborations. Following its backing, Galaxy General Robots partnered with Meituan to develop the world’s first humanoid AI robot-operated smart pharmacy solution, piloting two 24-hour unmanned pharmacies in Beijing with ambitions for 100 nationwide. Similarly, hotel AI robot specialist Yunji Technology integrated with Meituan’s food delivery platform, using robots to solve the notorious “last 100 meters” delivery challenge within hotels. Wang Qian, CEO of Ziyuanliang Robotics, echoed this focus post-funding, stating plans for “open service scenario deployments in China, exploring service closed-loops across different environments.”
Divergent Data, Converging Ambitions
The core technological schism lies in data acquisition and utilization – a critical hurdle for AI robot advancement. Ziyuanliang’s WALL-A model emphasizes learning from real-world interactions, employing multi-task training (like pouring water and changing clothes) to force the model to extract underlying cross-task principles. This, Wang Qian argues, fosters crucial few-shot or zero-shot generalization capabilities – the ability for an AI robot to perform novel tasks without extensive retraining. He dismisses pure simulation reliance: “Robotic reinforcement learning needs exponentially growing data, and simulated data struggles to transfer to reality due to physical interaction complexity.” Ziyuanliang claims over 90% success rates in complex, long-sequence tasks for its AI robots.
Galaxy General’s Wang He presents the counterpoint, highlighting the prohibitive cost and scarcity of real-world training data. He cites Tesla’s experience: months of data collection by a 40-person team were needed just to train its humanoid AI robot for battery sorting. “Enormous, costly data burdens lurk behind humanoid AI robot productivity,” Wang He asserted, making synthetic data a pragmatic, scalable alternative for his company. This fundamental disagreement on data strategy exemplifies the field’s unresolved technical direction.
Building an Ecosystem, Not Just a Portfolio
Meituan’s AI robot investment spree isn’t new; it’s an evolution. Before Yushu became a marquee holding (Meituan now holds 8% as its second-largest shareholder), the company invested in cleaning AI robot maker Gaoxian Robotics (2021), collaborative AI robot firm Faoiwei (2022), and backed hotel AI robot provider Yunji Technology ahead of its IPO. The shift towards exclusive bets, like the Ziyuanliang deal, indicates growing confidence in specific approaches, yet the overarching strategy remains ecosystem-driven.
“For hard tech with high confidentiality or prior major internal investment, Meituan keeps it in-house,” the source close to Meituan explained. “But in fragmented scenarios, or where technology lacks consensus, Meituan prefers empowering the ecosystem through investment.” In a sector lacking unified standards, leveraging capital to foster a diverse network of AI robot innovators poised to serve Meituan’s vast logistical and retail needs represents a calculated, lower-risk path than attempting to pioneer all breakthroughs internally.
The Long Game for AI Robot Dominance
Meituan’s 2025 strategy explicitly commits to deepening its “retail + technology” focus, amplifying investments in frontier technologies like AI, drone delivery, and automated distribution. Its AI robot investments are central to this vision. While the debate between real-world data versus simulation, or end-to-end models versus modular systems, rages on, Meituan is ensuring it has stakes across the board.
The goal transcends merely owning pieces of promising startups; it’s about integrating adaptable, intelligent automation into every facet of its operations – from warehouses sorted by AI robots to pharmacies run by AI robots and hotels serviced by AI robots. By funding diverse technical paths and mandating scenario-focused deployments, Meituan is not just betting on the future of AI robots; it’s actively architecting the ecosystem where these machines will prove their value, hoping that within its broad portfolio lies the next Yushu, or perhaps, the defining AI robot platform of the coming decade. The race for embodied intelligence supremacy is far from won, but Meituan is ensuring it has multiple horses running.