Mark Cuban Says 'We're in a Bubble' and Warns AI Data Centres May Not Need Today's Vast Resources

Billionaire investor Mark Cuban has warned that the artificial intelligence boom could be creating a market bubble, with heavy borrowing and rapid technological advances posing risks for investors.

Speaking on The Diary of a CEO podcast on 26 October 2026, he questioned whether today's expensive AI infrastructure would remain necessary as the technology becomes more efficient.

Cuban's warning centres on two concerns: investors may be prioritising market excitement over business fundamentals, while technology companies are taking on substantial debt to expand their AI capabilities.

He has previously joked that some data centres could eventually become 'pickleball courts', suggesting that infrastructure built for today's technology might lose value as requirements change.

Why Mark Cuban Thinks the AI Market Is a Bubble

Cuban made his fortune during the dot-com era, selling Broadcast.com to Yahoo in 1999. Speaking to host Steven Bartlett, he compared aspects of today's market with that period while acknowledging important differences.

His concern is that investors are trading shares based on anticipated news and what another buyer might pay, rather than focusing on the underlying value of businesses. Such behaviour, in his view, can cause prices to move sharply in response to relatively small developments.

'We're in a bubble,' Cuban said during the interview. He also acknowledged that he did not know what would cause the bubble to burst.

The comments featured in the 26 October episode of The Diary of a CEO. The discussion also covered AI's limitations, including its ability to predict patterns without necessarily understanding consequences.

Cuban's argument is not that artificial intelligence has no value. Instead, he is questioning whether current investment levels and market valuations can be justified over the long term.

Debt and Data Centres Add to the AI Bubble Risk

The financial stakes extend beyond share prices. Amazon, Alphabet, Meta and Oracle have issued substantial amounts of debt as technology companies expand infrastructure to support AI development.

According to Reuters' analysis of LSEG data, the four companies issued approximately $194 billion (£145 billion) in bonds through 7 July 2026. That was about $86 billion (£64.5 billion) more than the roughly $108 billion (£81 billion) issued throughout 2025, representing an increase of approximately 79 per cent.

The figures illustrate the scale of borrowing, but they do not establish that the companies will struggle to repay their debts. Their ability to manage those obligations will depend on financing costs, cash flow and the returns generated by their investments.

Cuban's data-centre warning raises another question. If advances allow AI systems to operate with less computing power or electricity, some facilities could prove less necessary than their builders anticipated.

That remains a possibility rather than a certainty. Demand for AI computing could also continue to grow, meaning efficiency improvements would not automatically make existing infrastructure redundant.

Cuban has described himself as supportive of AI and pointed to Amazon's survival after the dot-com crash. The challenge, he suggested, is identifying which companies will ultimately succeed.

His advice to investors was straightforward: 'Just be patient.' He also recommended keeping some cash available while waiting for market conditions to change.

That approach is not a guarantee against losses, and holding cash carries its own opportunity costs if share prices continue rising.

The Federal Reserve's 2025 household survey found that 63 per cent of US adults could cover a hypothetical $400 emergency using cash or its equivalent. That measure includes savings and a credit card paid off at the next statement.

For households with limited financial reserves, preparing for a downturn may be harder than it is for wealthy investors. Cuban's warning is ultimately about uncertainty: neither the timing of a market correction nor the eventual winners of the AI boom can be known with confidence.