PARIS – Alibaba co-founder and chairman Joe Tsai said artificial intelligence represents a market opportunity far larger than enterprise software and defended the massive infrastructure spending underway across the industry, arguing that AI is beginning to unlock meaningful productivity gains.
Speaking at the VivaTech conference in Paris, Tsai outlined Alibaba’s AI strategy, its commitment to open-source models and its growing efforts to position itself as an alternative technology partner for European companies concerned about digital sovereignty.
Tsai framed AI not as another software category but as a technology capable of generating “units of human intelligence and human productivity.”
“If you ask me how big is the market, what is the total addressable market (TAM), I would tell you that it’s much, much bigger than anybody’s IT budgets,” Tsai said. “It’s much, much bigger than the software market.”
He estimated that more than half of the world’s roughly $100 trillion gross domestic product is tied to human intelligence and productivity, creating what he views as an enormous opportunity for AI technologies.
“That is the TAM of AI,” he said at the tech summit, which sponsored this journalist’s trip. “That’s why we’re all in on AI.”
Need more clues? Ask the Sherlock chatbot in the lower right corner to summarize this story, explain technical concepts or answer other questions.
Searching for AI’s ROI
The comments come as technology companies worldwide continue pouring hundreds of billions of dollars into AI infrastructure, even as investors and corporate customers debate whether current returns justify the spending.
Tsai acknowledged that many chief executives have yet to see dramatic productivity improvements. “A lot of corporate CEOs will tell you that their engineers are burning a lot of tokens, cost goes up,” he said.
Still, he argued that companies are approaching an inflection point where experimentation begins translating into measurable gains. At Alibaba, he said, engineers using AI coding tools are expanding beyond their traditional responsibilities and exploring new projects, accelerating innovation across the company.
“We’re at the cusp of real productivity gains right now,” Tsai said.
Alibaba’s AI ambitions build on nearly two decades of cloud computing investments. Tsai said the company began developing cloud infrastructure 17 years ago out of necessity as its e-commerce operations generated massive amounts of data.
Rather than relying on third-party database and storage providers, Alibaba developed its own technology stack, which eventually evolved into one of China’s largest cloud computing businesses.
AI model companies: Flash in the pan?
Today, Alibaba participates across nearly every major layer of the AI stack, from chips and cloud infrastructure to foundation models and applications.
Tsai highlighted the company’s Qwen family of large language models, which have become among the world’s most widely used open-source AI models. He argued that Alibaba’s full-stack approach provides strategic flexibility at a time when it remains unclear where the greatest value in AI will ultimately accrue.
“Right now, the model companies are very hot,” Tsai said. “But over time that may not be the case.”
The strategy is supported by Alibaba’s highly profitable e-commerce business, which Tsai said generates approximately $25 billion in annual free cash flow. “That’s what allows us to make future investments,” he said.
Tsai also dismissed concerns that the industry’s AI infrastructure buildout is creating a speculative bubble. He acknowledged the massive spending among U.S. hyperscalers, and annual investments could exceed $1 trillion in the coming years.
While the numbers appear staggering, Tsai said they are justified by AI’s long-term economic potential. “We’re trying to tackle a total addressable market of $50 trillion,” he said.
Benefits of open-source models
A major theme of Tsai’s European tour has been digital sovereignty, a topic that has gained urgency across the continent amid concerns about dependence on foreign technology providers.
Tsai argued that open-source AI offers one pathway toward greater technological independence and data control.
Two things drive sovereignty concerns, he said. “Technology independence” and “data privacy.”
He said organizations can download open-source models such as Qwen and run them inside their own infrastructure, reducing reliance on external providers while keeping sensitive data behind corporate firewalls.
“You can take your own data and train it further, fine-tune it, do post-training of the model, and keep your entire data private within your own firewalls,” Tsai said.
He contrasted that approach with closed-source models accessed through application programming interfaces, where users may have limited visibility into how their data is handled. Leading U.S. AI model companies such as OpenAI and Anthropic offer mostly closed models.
China fears
The discussion turned to geopolitical tensions after the moderator asked whether European organizations should worry about becoming dependent on Chinese technology in the same way some fear dependence on American providers.
Tsai acknowledged the concern, saying bluntly that there’s no guarantee there won’t be future restrictions on access.
But he argued that diversification itself provides value. “Right now all of your eggs are in one basket,” Tsai said, referring to U.S. partners. “Why not get a second basket?”
Alibaba is already working with major European industrial companies including BMW, Siemens and Bosch through its cloud business in China. Tsai said manufacturing may become one of the most important sectors for enterprise AI because of the high-quality proprietary data these companies possess.
That data can be used to improve design, testing, quality control and production processes through customized AI models. “We think that this is in the future going to be a very interesting segment,” Tsai said.
He pointed to Bosch’s autonomous driving initiatives as one example of AI’s growing role in industrial applications that require large amounts of computing power.
People play while AI agents work
Looking further ahead, Tsai offered an optimistic vision of AI-powered agents working continuously on behalf of individuals.
Standing in Alibaba’s new Paris office overlooking a café, he said he watched people sitting outdoors enjoying the afternoon and suggested that such scenes may represent the future.
“They may be drinking coffee and having a good time,” Tsai said. “But the reality is they have deployed agents that are doing the work.”
Rather than eliminating human purpose, he argued, AI could free people to spend more time with family, attend concerts, watch sporting events and pursue entertainment.
“When people spend less time in the office, where do they want to go?” Tsai asked. “They don’t want to just sit at home.”
The remarks reflect a growing divide within the AI industry between those who view automation primarily as a productivity tool and those who see it fundamentally reshaping the relationship between work and leisure. For Tsai, the latter vision appears increasingly plausible as AI systems become more capable.
“We believe,” he said, “that artificial units of intelligence will be able to add value to human intelligence.”
