While headlines celebrate Nvidia’s $5tn valuation and Google’s $100bn quarter, a growing number of experts and studies are questioning the reality behind the $3 trillion AI datacenter spending spree. They warn that “exuberance” may be blinding investors to significant risks, speculative projects, and a lack of real-world returns.
A devastating report from MIT in August provided a major reality check, showing that 95% of organizations are getting zero return from their investments in generative AI pilots. This directly contradicts the “lofty revenue expectations” that project the generative AI market to hit $1 trillion by 2028, which is the assumption justifying the $3tn in datacenter spending.
Industry insiders are also raising alarms. Andy Lawrence of the Uptime Institute, which rates datacenters, stated that many of the projects announced “with a fanfare” are “speculative.” He warned that many “will never be built, or will be built and populated only partially.” Alibaba’s chair, Joe Tsai, echoed this, saying he sees “the beginning of some kind of bubble” in projects raising funds without customer commitments.
This “speculative” boom is being financed by an influx of risky debt from the private credit market. Gil Luria of DA Davidson worries that lenders, caught in the AI hype, are funding “unproven” projects. He warns that if this debt, potentially reaching hundreds of billions, goes bad, it could pose a “structural risk to the overall global economy.”
The industry is betting that ChatGPT’s 800 million weekly users are a sign of unstoppable demand. But skeptics point to the MIT data and the speculative builds as evidence that the $3 trillion AI boom could be a bubble waiting to burst.