Alibaba and Tencent Lead China's Pivot from Chatbots to Embodied AI Robotics

China's AI race has entered a new, physical dimension. The country's leading internet companies are aggressively pivoting from the world of digital chatbots to the tangible realm of embodied artificial intelligence, where AI models serve as the brains for autonomous robots. This transition is being led by Alibaba and Tencent, who have launched a series of new AI models, restructured their businesses, and committed billions of dollars to the sector over the past year .

The shift targets a future where AI doesn't just answer questions but actively interacts with the real world—planning tasks, navigating spaces, and manipulating objects. For Alibaba, this means building the foundational software for robot control. For Tencent, it involves a capital-intensive strategy to power the entire ecosystem.

Alibaba's Model Offensive: From Chat to Robot Command

Alibaba has made the most direct technological push, releasing multiple AI models designed explicitly to bridge the gap between software and physical action.

Qwen3.7-Max: The Agentic Foundation

Unveiled at the Apsara Conference on May 20, 2026, Qwen3.7-Max is Alibaba's flagship model for the "agent era" . Unlike a standard chatbot, this model is a foundational platform for building AI agents capable of using external tools. Its core capabilities directly support robotics :

• Physical Robot Control: The model features robust "tool-calling" capabilities, allowing it to act as a digital brain that orchestrates physical actions in a robot, including navigation, obstacle avoidance, and task planning .
• Long-Horizon Autonomy: In a landmark demonstration, the model ran autonomously for 35 hours, making 1,158 sequential tool calls to optimize the software for a self-developed AI chip . This showcases the system's ability to handle the extended, multi-step tasks typical in real-world robotics.

RynnBrain: An Open-Source Brain for Robots

In February 2026, Alibaba's DAMO Academy deepened its commitment with the release of RynnBrain, an open-source embodied foundation model based on the Qwen3-VL vision-language model . Designed as a dedicated "brain" for robots, RynnBrain is built on a framework of egocentric cognition and spatiotemporal understanding .

The model introduces several key innovations for robotics :

• Spatio-Temporal Memory: Robots can record and recall historical states during a task. For instance, if a robot is interrupted while performing Task A to handle Task B, it can remember its precise spatial and temporal status to seamlessly resume Task A .
• Physics-Aware Reasoning: The system can understand its environment, predict object trajectories, and navigate complex, cluttered spaces like kitchens or factory floors .
• Open-Source Release: Alibaba released the full series of seven models, including dense versions (2B and 8B parameters) and a powerful 30-billion-parameter Mixture-of-Experts (MoE) variant, along with specialized models for manipulation planning (RynnBrain-Plan) and vision-language navigation (RynnBrain-Nav) .

Organizational Restructuring

These technical releases are backed by concrete business moves. In January 2026, Alibaba formed a dedicated business unit merging its robotics and embodied AI efforts under veteran executive Chen Jun . This followed an earlier revelation in October 2025 from Qwen chief Lin Junyang, who confirmed he had set up a small, specialized robotics and embodied AI team within the Qwen group .

Tencent's Strategy: Investing Billions, Not Building Bots

Tencent's approach to embodied AI is markedly different. Rather than releasing its own robotic control models, the company is positioning itself as a capital and infrastructure powerhouse for the entire industry.

A Massive Capital Commitment

Tencent has committed to a dramatic increase in its AI spending, framing it as strategic, front-loaded investment akin to capital expenditure . The company spent RMB 18 billion (US$2.6 billion) on new AI products in 2025, with RMB 7 billion spent in the fourth quarter alone. For 2026, Tencent President Martin Lau has pledged to "at least double" this amount, putting its annual AI investment at over RMB 36 billion .

The Infrastructure Play

Tencent's leadership has been clear that the company does not plan to build its own robots. CEO Pony Ma stated that Tencent's robotics lab has shifted its direction in recent years to focus on the company's strengths in cloud services, big data, and applications. The goal is to be a "collaborator and supporter for all robotics manufacturers rather than a competitor" . This infrastructure strategy was also outlined at the World Artificial Intelligence Conference, where Tencent and JD.com both confirmed they are integrating their software and cloud capabilities into robotic platforms .

Startup Funding Spree

In line with this strategy, Tencent has been a major backer in a wave of funding for embodied intelligence startups. It participated in a massive RMB 2 billion funding round for Simplexity Robotics, a Chinese embodied intelligence startup . Alongside other giants like Alibaba, JD.com, and Meituan, Tencent has been intensifying its investments in projects that wire AI directly into physical machines, with the goal of slashing logistics costs and re-engineering retail and service industries .