Data center energy demand is skyrocketing, primarily driven by AI and digitalization, with global consumption projected to double by 2030 from 2024 levels. In the U.S., demand could triple by 2028, with forecasts showing substantial growth from 4% to over 10% of total U.S. electricity by 2030.
Behind these forecasts is an assumption that public utilities and other power generating facilities will be able to build the systems to deliver such power. For almost a decade, until the last year or so, power generation facilities showed minimal growth, but to meet data center and other demands will require an expansion of power output/generation of more than 25%, according to studies by the Deloitte Research Center and others.
Data center demand alone is projected to take a fivefold jump from 2024. Industrial electrification from increased manufacturing facilities will increase power demand by 2030, on top of growth in household and commercial consumption.
At the same time, new supply is not coming online fast enough. The energy mix is shifting toward renewables, which accounted for 93% of new capacity through July 2025, with solar and storage making up 83%. But the pace of connecting these new energy sources has lagged. Two terawatts of capacity are stuck in interconnection queues, almost twice the currently installed capacity.
In addition, the power grid faces other challenges, particularly from extreme weather events. In just 2024, there were a record 27 extreme weather events that cost more than US$1 billion each. Such events have risen steadily since the 1980s, when there were on average 3.3 disasters per year that cost US$1 billion or more (inflation-adjusted).
All this growth will require investment, and investment has to come from somewhere. The electric power sector’s traditional funding avenues – filing rate cases and issuing debt and equity – may not suffice. Customer electricity bills rose 23% between 2019 and 2024, with residential prices climbing by nearly 26%, and state power commissions may be reluctant to allow massive price increases.
Significant additional federal funding is problematic, and additional tax breaks for power companies might have a political downside, especially in the 2028 election.
So… who will pay for the coming AI power-demand… and how?





Did we ask for this? Nope.
Paul Krugman has a recent substance discussion of the AI/ data center issue. That ‘boom’ appears to be self-sustaining and may in fact be a bubble. As well, a lot of the financing of these data center projects are bound up in non-transparent activities (like local councils approving but not being able to discuss the financial risks associated with that project). He reports that some of the financial vehicles allow the data center operator to write off a business loss to the local community or rate payer.
There’s also some discussion that these data centers are being built to the exclusion of other business opportunities in the US. Nearly all the capital being spent in 2025 is going to data center construction.
I’m sure many of the blog readers will also contribute to this discussion. But keep in mind that Trump’s admin has cancelled most of the infrastructure clean energy projects that would have made some of this worl palatable.
Sarcastically, why don’t we ask the AI how to solve the problem? Or has the hype out paced reality again?