Original title: AI for AI: Optimizing Additional Infrastructure Build-out to Power Artificial Intelligence Data Centers
Alexander Crosier, Kyle Onghai, Ronnie Sircar · September 2026
Artificial intelligence data centers need huge amounts of power. This rapidly growing demand is driving up electricity prices because power companies are not building new supplies fast enough to match it. To see why, researchers created a model linking data-center power growth, available power generation, and market electricity prices. They simulated how demand and supply expand over time while facing real-world randomness, which produced a range of possible price outcomes. They also modeled an investor trying to maximize revenue by deciding when to build new power. The study found that even though fast-growing power demand demands new energy, companies might hesitate to build quickly. Investors worry about uncertain forecasts, construction risks, and lowering their own profits by building too much supply.
Imagine a town that suddenly needs more water because everyone is filling huge swimming pools. The water company wants to build a new reservoir, but they don't know exactly how many pools will be built or how fast. If they build too much, they waste money and lower the price per gallon. If they build too little, water gets expensive and scarce. In this paper, the town is the electricity grid, the swimming pools are AI data centers, and the reservoir is new power plants.
AI data centers need huge amounts of electricity, and their rapid growth is driving up power prices without enough new supply being built. This affects everyone who pays an electricity bill, because utilities and investors are uncertain about how much generation to build. If supply doesn't keep up, prices stay high or get worse.
The abstract does not provide empirical data, real-world case studies, or specific policy recommendations. It presents a theoretical framework and models without testing against historical data or validating the stochastic assumptions. The analysis focuses on investor incentives rather than grid reliability or environmental impacts.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
AI for AI: Optimizing Additional Infrastructure Build-out to Power Artificial Intelligence Data Centers — Alexander Crosier, Kyle Onghai, Ronnie Sircar
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.