- Wu Yongming says AI is entering a new phase in which machines become the main force behind thinking and intelligence becomes a scalable commodity
- Alibaba is betting on models, chips and cloud infrastructure as the foundations of an AI economy that could eventually dwarf today’s technology industry
Alibaba Group CEO Eddie Wu Yongming (吴泳铭) has said machines could eventually account for more than 1,000 times the total amount of “thinking” done by humans, arguing that AI is entering a phase comparable to the Industrial Revolution.
Speaking at the 2026 Apsara Conference in Hangzhou on September 22, Wu said the key shift is that machines are becoming the main force behind thinking, while intelligence is becoming a scalable commodity.
‘Thinking’ goes to scale
Wu compared the transformation to the Industrial Revolution, when steam engines, internal-combustion engines and electricity turned physical power into something that could be produced and deployed at massive scale.
“This time, it’s thinking’s turn,” Wu said.
Wu estimated that the total amount of machine “thinking” today is equivalent to only about 3% of human thinking, leaving room for growth of at least tens of thousands of times.
He said the eventual volume could exceed human thinking by more than 1,000 times.
But Wu cautioned that AI has yet to produce a defining product for the machine-intelligence era. Today’s AI coding tools, he said, are more like the electric light bulb of 1882: they replace existing work but have yet to create an entirely new era.
Building the AI grid
The implication, he said, is that AI will first require an infrastructure buildout on a scale comparable to the power industry.
Just as the widespread adoption of electricity required power plants and transmission grids, the next phase of AI will require vast amounts of computing infrastructure.
Alibaba plans to invest across three pillars — AI models, AI chips and AI cloud computing.
Its Qwen team plans to train a new model with 5 trillion to 10 trillion parameters as it moves toward artificial superintelligence, or ASI.
Alibaba’s semiconductor unit T-Head also unveiled the Zhenwu V900, which the company said delivers three times the computing power of its predecessor and can scale to a cluster of up to 500,000 chips.
The gigawatt race
Alibaba Cloud, meanwhile, aims to expand its global data-center capacity to more than 20 gigawatts by 2032, according to Wu.
The strategy marks a further shift in Alibaba’s AI narrative from competing primarily on model capabilities to building the infrastructure needed to make AI ubiquitous.
The company has pledged more than 380 billion yuan ($56.7 billion) in AI and cloud infrastructure investment over three years and raised about $10.2 billion through a Hong Kong share sale in August.
Alibaba’s bet
The infrastructure race is also becoming a global competition measured in gigawatts.
Microsoft has said it plans to reach 38GW of data-center capacity by 2032, while Amazon Web Services has previously disclosed plans for several gigawatts of additional capacity.
For Alibaba, the bet is that models and applications will continue to change rapidly, while the underlying infrastructure — chips, data centers and cloud platforms — will remain essential as AI scales.
Header image credit: Alibaba Group
