A massive disconnect is emerging in the AI industry. On one side, a projected $3 trillion is being poured into building datacenters. On the other, a recent MIT study found that 95% of organizations are getting zero return from their investments in generative AI pilots. This has ignited fierce debate about whether the spending spree is a sustainable boom or a colossal bubble.
Despite the market’s “current positivity,” with tech giants like Nvidia and Microsoft hitting record $5tn and $4tn valuations, skepticism is growing. The revenue expectations underpinning this boom are “lofty,” with Morgan Stanley projecting generative AI revenues to grow from $45bn last year to $1tn by 2028. But the MIT research raises serious questions about whether businesses are actually willing or able to pay for it.
The Uptime Institute, which rates datacenters, adds to the skepticism. Its executive director of research, Andy Lawrence, warns that many of the announced multi-trillion-dollar projects are “speculative.” He states, “Many of the datacentres… either will never be built, or will be built and populated only partially.”
Even industry insiders are sounding the alarm. In March, Alibaba’s chair, Joe Tsai, warned he was seeing “the beginning of some kind of bubble,” pointing to projects raising funds without any customer commitments. This is the “speculative” side of the market that worries analysts, where unproven projects are being funded by risky debt.
Tech companies are betting that business and public demand will eventually catch up to the hype, justifying the trillions in expenditure. But with 95% of early business pilots failing to produce a return, the $3 trillion question remains: is the industry building essential infrastructure or castles in the sky?