Alibaba signage. Photo by Leon Lee via Flickr. Alibaba Group Chairman Joe Tsai has raised concerns about a possible bubble in data center construction for AI.
Speaking at the HSBC Global Investment Summit in Hong Kong, Joe warned that the current boom in server farms could overload real demand. This expansion is moving fast globally, from the US to Asia.
“A major movement here was a push towards AI, which was being starved years of R&D during the early 2000s boom,” said Bill Theobald, an AI engineer and senior lecturer at the University of Illinois at Urbana Champaign.
Tech giants worldwide are pouring money into AI by building large data centers. These centers contain powerful processors that run AI models, but many are being constructed without a guaranteed customer.
Nvidia and the Korean chipmaker SK Hynix are cashing in as Microsoft, SoftBank, and Alibaba spend billions on their AI chips. Alibaba has committed to investing more than 380 billion yuan ($52 billion) in AI over the next three years.
New data center projects are being developed all over Asia, with places like India and Malaysia quickly becoming sizable AI infrastructure centers.
The scale of spending is even more dramatic in the United States. One thought project, Stargate, imagines a half-trillion-dollar investment in AI infrastructure.
However, Tsai cautioned that not all those projects were being built based on real demand. He was worried about data centers being built with no guaranteed users, as some investment funds were raising hundreds of millions or even billions of capital for such projects.
Alibaba’s stock drops as investors question AI spending boom
After Tsai’s comments, Alibaba’s stock fell more than 3% in Hong Kong. His worries reflect a broader skepticism among investors who fear that spending on artificial intelligence infrastructure will not pay off anytime soon.
An increasing number of analysts are asking if such rapid expansion is necessary. There’s been a new cause for those doubts: the Chinese startup DeepSeek recently revealed an open-source AI model that purportedly competes with US technology but was developed for a tiny fraction of the cost. It begs the question of whether large companies are overpaying for AI infrastructure without considering less expensive, more appropriate options.
Also, AI has not yet demonstrated its profitability at scale. And while it can transform sectors, practical implementations are sparse. Tsai noted that many data center projects are being launched based on optimistic future projections rather than real demand in the present.
He pointed out that many investment funds are raising billions for AI infrastructure without “uptake” agreements—contracts that would guarantee customers for these services. This raises the question of whether some data centers will ever be fully used.
US tech giants pour billions into AI but face growing skepticism
The largest investments in AI are being made in the United States, with companies investing staggering sums of money.
Amazon, Alphabet (Google’s parent company), and Meta have all committed to spending $100 billion, $75 billion, and up to $65 billion, respectively, on AI infrastructure this year.
According to some estimates, Microsoft, one of the behemoths of AI, will invest about $80 billion in AI data centers this fiscal year.
However, in the face of such widespread commitments, there are signs that even the biggest companies are starting to rethink their strategies. In February, reports indicated that Microsoft had canceled some data center leases, prompting concerns the company might be purchasing more AI computing capacity than it needs.
Microsoft executives have sought to allay such worries, arguing that the company is investing at record levels to keep ahead in AI. However, they have also indicated that spending growth will slow in the next fiscal year, which starts in July.
Tsai was still amazed by the figures mentioned in the United States and was skeptical about the size of investment in AI. He said many companies are counting on future demand that may not come as expected.
Cryptopolitan Academy: Coming Soon - A New Way to Earn Passive Income with DeFi in 2025. Learn More
