The $3 trillion AI datacenter boom is being built on “lofty revenue expectations,” with Morgan Stanley projecting the generative AI market to hit $1tn by 2028. But a hedge fund founder’s warning about depreciation, combined with an MIT study on “zero returns,” casts serious doubt on whether this “incredible” spending will ever pay for itself.
The warning from Harris Kupperman is stark: datacenters will “depreciate twice as fast as the revenue they generate.” This means that for every dollar of revenue, two dollars of asset value are lost, making profitability a steep uphill climb.
This is especially troubling for the $1.5tn “speculative” side of the boom, which is being funded by “private credit.” These lenders are backing “unproven” projects, and if Kupperman’s math is right, the “quickly depreciating assets” they are using as collateral will be worth far less than the loans.
This is compounded by the MIT study, which found 95% of organizations are getting “zero return” on their generative AI pilots. If businesses aren’t seeing value, they won’t generate the “lofty revenue” needed to overcome the high depreciation costs.
While “hyperscalers” like Google and Microsoft can absorb these costs, the “speculative” projects “without their own customers” cannot. The $3tn bet is that revenue will grow faster than depreciation—a gamble that looks increasingly risky.
Picture Credit: www.rawpixel.com
