A new report from McKinsey & Company has laid out just how much money will be needed to ensure data centers are equipped to handle the compute demands needed for an AI-fueled future and let’s just say, it is far from cheap.
In an April report, the management consulting firm estimates that nearly $7 trillion in capital outlays will be needed by 2030 to support the capacity demands of AI data centers.
The math behind the money. McKinsey partners back up their reasoning for the 13-digit estimate in the report by predicting $5.2 trillion in capex will be needed to support data centers that are fit to handle AI processing loads, while another $1.5 trillion will be needed to fund data centers powering “traditional IT applications.” That amounts to $6.7 trillion needed to keep facilities up to speed with compute demand.
The figures account for a scenario where demand for AI continues at a similar momentum. However, the global consulting firm also calculated the amount of capex needed to handle AI processing loads in the event that there is an accelerated or constrained demand for compute power in the next five years. In an accelerated demand environment, where 205 incremental gigawatts of AI-related data center capacity is added by 2030, McKinsey predicts $7.9 trillion capex will be needed to support loads. In a more restricted environment, where only 78 incremental gigawatts are added in the next five years, the firm estimates that $3.7 trillion in capex will be required.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
Fact or fiction? IT Brew caught up with a few experts to see if those figures sound accurate when talking about the price of the AI race. Wes Cummins, CEO of Applied Digital, told IT Brew the $7 trillion estimate appears to be “on the high end.”
“If we’re talking just the next five years, I think that’s a hard number to hit, just from the practicality of building and finding enough power,” Cummins said.
Steven Lim, an SVP of marketing and go-to-market strategy at NTT Global Data Centers, attempted to do some calculations of his own to validate the report’s estimate. To Lim, the large figure is not that far-fetched.
“If we have an idealized situation and a consistent demand curve that we see today, then $7 trillion is not out of bounds,” he said.
Lim did raise a few concerns around the five-year timeline. He questioned if there is enough skilled labor to build out the infrastructure needed to support compute demands within the timeframe. He added current geopolitical variables—such as the on-again, off-again tariffs—can also potentially derail things.
“If everything happens, then yeah, we could invest that much and we could see that level of deployment,” Lim said. “I just think there’s so many other factors that could change that.”