As data centres accelerate demand and grid upgrades lag behind, US energy crisis looms

Rising energy costs and surging data centre power consumption threaten to strain the aging U.S. electricity grid, prompting calls for urgent infrastructure investment and innovative solutions amid environmental trade-offs.

Everyone – politicians and the public – are talking about energy costs and the strain that burgeoning data centres, driven by artificial intelligence, are placing on an ageing U.S. electricity system. Rising residential prices, projected demand increases and slow-moving grid upgrades have combined to turn what was a technical planning challenge into a major political and economic issue. ^[1][7]

Residential electricity prices rose sharply between September 2024 and September 2025, increasing by about 7.4% from 16.8 to 18 cents per kilowatt-hour, while government analysts expect prices to continue to outpace inflation into 2026; with household incomes broadly flat in real terms, those increases bite hardest at low-income families and filter through the wider economy via higher business costs. ^[1]

The scale of anticipated demand is enormous. Industry and government estimates show data centre energy use has already tripled over the past decade and could double or triple again by 2028; by 2030 some analyses project U.S. electricity demand rising roughly 25% and data centres consuming as much as 11–12% of national electricity, figures that would dramatically reshape planning needs for generation and transmission. According to the Department of Energy’s 2024 report, data centres consumed about 4.4% of U.S. electricity in 2023 and could account for between 6.7% and 12% by 2028. ^[1][7]

Systemic capacity shortfalls are already materialising. The PJM Interconnection, the nation’s largest grid operator, recorded record-high capacity auction prices in December 2025 as supply growth lagged demand, with auction results falling short of reliability targets by some 6,600 megawatts, enough to power nearly 5 million homes, and sending capacity prices sharply higher to incentivise new generation. Those price signals have pushed some utilities and producers into windfall-like positions while raising acute consumer affordability concerns. ^[2]

Part of the problem is the slow pace of long-lead infrastructure projects, from generation orders that face multiyear delivery times to transmission builds that are often delayed by permitting battles and fragmented jurisdictional processes. The Brattle Group has estimated the cost of necessary new and upgraded equipment at between US$760 billion and $1.4 trillion over 25 years, a price tag that reflects decades of underinvestment and the cumulative impact of project delays. ^[1]

Big Tech and hyperscalers are responding with an “all of the above” energy strategy: accelerating renewables and storage where practicable while securing faster-deploying gas capacity and locking in nuclear long-term via power purchase agreements and investments in small modular reactors. Reuters reporting finds utilities and tech firms are combining renewable projects with new gas-fired plants to meet immediate reliability needs, even as companies maintain public commitments to decarbonisation and pursue carbon capture, SMRs and other low-carbon options for the 2030s. ^[5][3]

That energy-first pivot carries its own environmental and resource trade-offs. Rapid data centre expansion also amplifies water demand for cooling, particularly in water-stressed regions such as Phoenix where additional facilities could meaningfully increase local water stress. Operators are testing waterless and closed-loop cooling technologies, yet those approaches can raise energy intensity, exposing a difficult trade-off between water conservation and electricity demand that remains underreported. NGOs and investor groups are pressing for better disclosure of the full water and energy footprints. ^[4]

Policy reactions are emerging at state and regional levels as consumers and officials confront bill shocks. Several U.S. governors and state actors have intervened to temper price increases or pause electrification mandates that would further raise demand, while regional regulators consider measures, ranging from temporary price controls to catch-up capacity auctions, to stabilise markets and protect consumers. At the same time, industry forecasts from financial and consulting houses warn of a near-term balance that may stay tight: Goldman Sachs outlines scenarios in which AI-driven loads push occupancy and power use to stress points through the late 2020s, even as slower growth or a demand shock could produce oversupply. ^[1][2][6]

There are no quick fixes. Experts and regulators emphasise a blended approach: accelerate deployment of renewables and storage where possible, streamline and standardise transmission planning and permitting, add fast-build gas capacity with strict emissions mitigation as a bridge, and pursue long-duration low-carbon options including nuclear, geothermal and enhanced grid management. Equally important are demand-side measures, time-sensitive pricing, programmable devices and AI-enabled load management, that can shave peaks and reduce the need for costly capacity builds if stakeholders cooperate. ^[1][3][5]

How the United States reconciles ambition on AI and decarbonisation with the immediate imperatives of reliability and affordability will define energy policy for the next decade. The choices made now, about who pays for grid upgrades, how projects are permitted and how corporate power demands are integrated with public policy, will determine whether rising demand becomes a catalyst for rapid modernisation or a source of persistent price and resource stress. ^[1][2][5]

📌 Reference Map:

• ^[1] (The Conversation) – Paragraph 1, Paragraph 2, Paragraph 3, Paragraph 5, Paragraph 8, Paragraph 9
• ^[7] (U.S. Department of Energy / LBNL report) – Paragraph 3
• ^[2] (Reuters) – Paragraph 4, Paragraph 9
• ^[5] (Reuters) – Paragraph 6, Paragraph 9
• ^[3] (Reuters) – Paragraph 6, Paragraph 9
• ^[4] (Reuters) – Paragraph 7
• ^[6] (ITPro / Goldman Sachs summary) – Paragraph 8

Source: Fuse Wire Services