How the data center boom is exposing weaknesses in US power grids, and a look at what fixes the electricity infrastructure may need to become fit for the future
The AI buildout is bottlenecked by energy. But America has the electricity to power its data centers; the problem is getting it to them.
Works in Progress MagazineChris Gillett
Context & Ripple Effects
Related coverage has tracked the same constraint from several angles: rising AI and crypto-linked data-center demand has collided with limited land, grid capacity and utility delivery capability. Reliability warnings and the move by some operators toward on-site generation show that the issue is not simply aggregate electricity supply.
This story advances that arc by focusing on the transmission and distribution problem: connecting available power to new, concentrated loads. That makes grid infrastructure a gating factor for AI deployment rather than a background utility concern.
First-order effects
Data-center developers face slower or more uncertain project deployment where grid connections and local delivery infrastructure cannot support requested load.
Utilities and grid planners must prioritize upgrades and connection planning for large new loads, while data-center operators may continue to seek alternatives to waiting for grid service.
Second-order effects
On-site power can become a practical workaround for operators, but it shifts more infrastructure responsibility and permitting exposure from utilities to data-center projects.
Shortages of grid capacity intensify competition for viable sites, electrical equipment and skilled construction labor, adding constraints beyond chips and data-center buildings.
Third-order effects
If concentrated AI load continues to outpace interconnection and delivery upgrades, location and power-access strategy will increasingly shape which operators can expand and where computing capacity is built.
The pattern points toward tighter coordination between utility planning, transmission and distribution investment, and large-load siting; without it, private self-supply may grow alongside—not replace—the grid.
The trend: AI infrastructure is turning reliable, deliverable electricity—and the grid buildout behind it—into a central determinant of data-center growth.
Accommodating lots of renewables on a grid means building grids for much larger peaks than they were once built for. On the one hand, this can be quite expensive. Today, two thirds to three quarters of UK bills are fixed and grid costs, and the country is spending £70 billion
@s8mb It's funny that the world mostly knows about the Texas grid because of the outages, given that it's probably the most efficient and best-designed grid in America. [image]
Excellent article complementing the work that @alasdairpr @SamWinterLevy & I have done for @CarnegieEndow and showing how to address the time to power problem
Well said, connect&manage. 76 gigawatts of new loads could be added if they were willing to disconnect for just 22 hours per year, in non-consecutive chunks of a few hours each
The main energy constraint on data centre buildout is not the price of electricity, it is the time it takes to get connected to the grid. Fixing this would cost billpayers and taxpayers nothing, reduce electricity prices, and make it vastly faster to build new data centres in
My article on US grid capacity for @WorksInProgMag is out! I argue that grids can grow faster by prioritizing connecting the highest-value power plants and flexible power users. https://worksinprogress.co/...
“Between 1990 and 2024, the price of electric motors declined by 97.5 percent. Power electronics fell 99.5 percent in price, processors built into devices by nearly 99.9 percent, and batteries 98.8 percent.” @WorksInProgMag https://worksinprogress.co/... [image]
Texas FTW “The solution is to let the power plant come online before the upgrades are constructed on the condition that it agrees to turn off during peaks. The power plant doesn't contribute to the reliability of the system since it won't be available when the system is under [im…
The electricity grid is a public good. Connections are currently given out at cost, letting developers capture their scarcity value. Auction grid connections to: get rid of fake queues; accelerate the AI boom domestically; and capture public value for the public.
The US data center buildout is the world's largest infrastructure project. But it faces one bottleneck: a slow and centrally planned electric grid. Twenty years ago, a power plant waited 20 months for interconnection. Now it is 55 months. In a new Works in Progress piece, we [ima…