Why is Alibaba ditching chatbots for robots and AI agents?

Alibaba is moving deeper into robots and AI agents, marking a sharp turn in the artificial intelligence race that was dominated only recently by chatbots.

The Chinese e-commerce and cloud giant on Tuesday unveiled its first full suite of AI models built for robots, a move that signals where large technology companies now see the next commercial prize.

Chatbots helped consumers talk to machines and Alibaba now wants machines to act in the real world.

The shift signals a broader change for investors, developers and businesses, as AI’s centre of gravity moves from conversation to execution.

Why chatbots are not enough anymore

The chatbot boom was built around one powerful idea: ask a question and get a useful answer.

That changed search, customer service, coding and office work. But it also left limits as most chatbots still wait for a human prompt.

They respond, explain, summarise or draft. They rarely complete an entire job on their own.

AI agents are designed to go further. They can plan, use tools, call other software, remember steps and complete multi-stage tasks with less supervision.

In simple terms, chatbots answer questions; agents run workflows. That could mean booking a flight, preparing a sales report, managing supplier orders, updating spreadsheets, or coordinating a factory process.

This is why Alibaba’s pivot matters. It is not just adding another model to a crowded chatbot market, but trying to build AI that can plug into commerce, logistics, cloud services and industrial systems.

Marc Einstein, research director at Counterpoint Research, told CNBC that AI agents could “upend traditional Internet business models,” warning that “if this happens the consequences for those who are not prepared will be severe.”

Alibaba is not alone as ByteDance, Zhipu AI, Baidu and other Chinese AI players are also pushing beyond chatbots, showing that this is becoming an industry-wide reset rather than one company’s experiment.

Alibaba’s robot offensive: What it actually built

Alibaba’s new robot AI models are aimed at giving machines a better understanding of the physical world.

That means helping robots identify objects, understand space, plan movements and carry out tasks in environments such as kitchens, warehouses and factory floors.

The push builds on earlier work from DAMO Academy, Alibaba’s research arm, including RynnBrain, an embodied AI model designed for physical reasoning, navigation and task planning.

In simple words, this is AI that is not limited to text on a screen, but is meant to help machines see where things are, understand what they are for, and decide what to do next.

Alibaba has also been strengthening the software and hardware stack around this strategy.

Its Qwen3.7-Max model, introduced in May, was built for the “agent era” and is designed to handle long, complex tasks.

Alibaba said the model sustained a 35-hour autonomous run involving more than 1,000 tool calls, a sign that the company is trying to improve reliability over long workflows rather than just chatbot fluency.

The company has also unveiled the XuanTie C950, a 5-nanometre RISC-V processor designed for agentic AI workloads.

That matters because agents are more demanding than chatbots as they need memory, coordination and repeated interaction with tools and data systems.

Alibaba’s broader pitch is that it can operate across the whole AI chain: chips, cloud infrastructure, foundation models, platforms and consumer or enterprise applications.

That gives it a route to monetise AI in more places than a standalone chatbot app.

CEO Eddie Wu has framed the opportunity in sweeping terms, arguing that there may one day be more agents and robots than people.