Fuelled by excessive spending on artificial intelligence-linked commodities and computing capacity, AI infrastructure investment has become a global bubble that is unlikely to burst in 2026 or 2027, according to Viktor Shvets, head of global and Asia-Pacific strategy at Macquarie Group.

“There are many parts of AI … from the underlying technologies to infrastructure – the things that are going to run on the back of the infrastructure of models; automation, quantum computing, biotech, gene slicing and sequencing, all the rest of it … is going to run on top of that,” said Shvets at the Macquarie Asia Conference 2026, on Tuesday.

“The AI bubble absolutely [exists] at the low-end infrastructure end,” he said, adding that investors were piling money into a “transformative technology” long before revenue can catch up, referring to areas such as data centres, chips and related hardware.

The five hyperscalers – Alphabet, Amazon, Meta, Microsoft and Oracle – were expected to spend a combined US$805 billion in capital expenditure in 2026 alone, according to Morgan Stanley’s latest estimates. In contrast, the total revenue from generative AI was only expected to reach US$161 billion in 2026, according to a report published by Fortune Business Insights earlier this year.

The mismatch between investment and monetisation has already triggered concerns. At the HSBC Global Investment Summit in Hong Kong last year, Alibaba Group Holding chairman Joe Tsai told delegates that he was starting “to see the beginning of some kind of bubble”, warning that many data centres were being built “on spec” while lacking either secured clients or “uptake” agreements.

Alibaba is the owner of the South China Morning Post.