Tsai emphasized that Alibaba’s strategy is not to pick a single winner in the AI stack but to invest across the entire value chain. The full stack spans proprietary chips, cloud infrastructure, foundation models, model-as-a-service (MaaS), and applications. He argued this integrated approach is essential because no one can predict with certainty where long-term value will concentrate.
Alibaba CEO Eddie Wu has echoed this, stating the company aims to become “the world’s foremost full-stack AI service provider.” The full-stack strategy is central to Alibaba’s stated five-year goal of surpassing $100 billion in combined cloud and AI external revenue, including MaaS.
Alibaba is funding its AI ambitions with cash flow from its core e-commerce business. On top of that, the company announced a three-year investment of at least RMB 380 billion — approximately $53 billion — in AI and cloud infrastructure. This sum exceeds Alibaba’s total AI and cloud spending over the past decade.
Tsai said the company is “all in” on AI, investing across the stack and figuring out the winners later. At the same time, he has warned that AI data center construction risks overbuilding, with some projects lacking clear customers and demand.
A central theme of Tsai’s VivaTech appearance was the role of open-source AI. He positioned Qwen — Alibaba’s family of large language models — as the world’s most used open-source model. The logic is straightforward: open-sourcing Qwen democratizes AI, proliferates applications, and drives demand for Alibaba Cloud’s inference and training compute.
Tsai made a direct sovereignty pitch to European customers. “Stop putting all your eggs in one basket,” he told the audience, positioning Qwen as an alternative to an American “kill switch.” He argued that Alibaba’s open-source and cloud offering represents a “second cloud basket” for customers seeking more choice and control over their data and technology stack.
This message is tailored for European governments and enterprises wary of over-dependence on U.S. providers.
Tsai described AI agents as “virtual knowledge workers” that can perform many tasks humans do today. He estimated that the global economy is about $110 trillion, with labor representing roughly 60% of that — and approximately two-thirds of labor tied to knowledge work.
AI agents, in his view, can address this $44–50 trillion segment by taking on tasks that require logical reasoning, autonomous planning, and long-term memory.
This vision extends beyond simple chatbots. Tsai sees agents evolving from personal assistants into autonomous virtual employees, reshaping the white-collar economy.
Tsai’s broader vision is human-centric. He argued that the point of AI is not to create the most advanced models but to ensure AI applications are widely used and beneficial to society. AI should create tangible value by augmenting human productivity rather than being judged only by benchmark performance.
The VivaTech speech was Alibaba’s strongest “all in” AI message yet — combining a massive market thesis, a multi-billion-dollar investment commitment, a full-stack technical strategy, and a sovereignty-minded pitch to European partners facing U.S. dependence.
Comments
0 comments