US Data Center Boom Faces a New Political Reality
Starting in January, Democratic governors will lead states holding 77% of the 300 GW of data center capacity announced for 2027–2030. The five biggest markets—Texas, Virginia, Pennsylvania, Ohio and Georgia—are among them. This gives the incoming governors influence over where most planned growth takes place. Most winning candidates campaigned on making data centers pay their grid costs, rather than blocking construction. States and utilities can require developers to make long-term commitments and provide upfront financial security to reserve grid capacity. Those terms can affect whether projects proceed, especially speculative ones. The article expects the election results to change the terms of growth more than its pace. Its risked forecast expects 121 GW to be added during 2027–2030, with many projects already under construction or otherwise advanced. Policy changes could still delay projects awaiting approval, but the article says those pauses affect a relatively small share of expected capacity.
What political change is coming to the states that host most of the planned US data center capacity?
Starting in January, Democratic governors will lead states holding 77% of the 300 GW of data center capacity announced for 2027–2030. The five biggest markets—Texas, Virginia, Pennsylvania, Ohio and Georgia—are among them. This gives the incoming governors influence over where most planned growth takes place.
Most winning candidates campaigned on making data centers pay their grid costs, rather than blocking construction. States and utilities can require developers to make long-term commitments and provide upfront financial security to reserve grid capacity. Those terms can affect whether projects proceed, especially speculative ones.
The article expects the election results to change the terms of growth more than its pace. Its risked forecast expects 121 GW to be added during 2027–2030, with many projects already under construction or otherwise advanced. Policy changes could still delay projects awaiting approval, but the article says those pauses affect a relatively small share of expected capacity.
How much data center capacity is planned for 2027–2030, and how much does the article expect will actually be added?
The article reports 300 GW of data center capacity announced for 2027–2030. That is the headline pipeline, not a promise that every project will be built. The article’s risked forecast expects 121 GW to actually be added during those years.
The gap reflects how the forecast treats projects with different levels of risk. It weights projects by their operator, customer and power supply, while giving high-risk or speculative projects little or no weight. It also excludes cancelled projects when defining capacity. For the later years, about 49 GW of the 191 GW planned for 2029–30 is included in the forecast.
The forecast therefore describes a more cautious, risk-adjusted expectation than the announced total. The article says permitting pauses are not the main limit: much planned capacity is already under construction or roofed, while slow construction pushes nearly a third of expected 2027–28 capacity past its planned year. Political and grid-cost changes may alter which projects move ahead.
What could a new Texas permitting pause delay, and why might construction delays have a bigger effect?
A new Texas permitting pause could hold up data center projects that still need approval. In the article’s base case, a pause from 19 January until new rules take effect in September 2027 would delay about 3 GW. The article also estimates 1.5–3 GW delayed if developers act early to secure stormwater permits; without a new Texas pause, about 1 GW would be delayed nationwide.
The key mechanism is approval timing. A pause mostly affects projects still waiting for permits, while projects already underway can continue. About half of planned 2027–28 capacity is under construction or roofed, and only 16% of expected capacity still needs approval.
Construction delays may have a bigger effect because they affect projects beyond those caught by a permitting pause. Slow building already pushes nearly a third of the capacity expected for 2027–28 past its planned year. If pauses persist for another year, over 10 GW could be pushed back; most delayed capacity is expected online by the end of 2030.
What does data center capacity measured in gigawatts mean?
A gigawatt, or GW, is a unit of power equal to one billion watts. Power describes the rate at which electricity can be supplied or used at a given moment. The article defines data center capacity as facility power, excluding cancelled projects.
That means a figure such as 300 GW describes the combined power capacity of announced data center facilities, not a count of buildings or a measure of electricity used over a year. Energy use is measured over time, such as in gigawatt-hours; capacity and energy use are related but not interchangeable. Actual demand can depend on how much of a facility’s capacity is being used.
The distinction matters when comparing the article’s announced total with its risked forecast of 121 GW to be added. Those figures concern expected facility power coming online in 2027–2030. They do not tell readers exactly how much electricity the facilities will consume in each hour or year.
Why are utilities asking data center developers for long-term commitments and upfront financial security to reserve grid capacity?
Reserving grid capacity for a data center can involve a long commitment and money secured upfront. The article says utilities increasingly require commitments of more than a decade and upfront security. These rules matter because they set the terms developers must accept to obtain grid capacity.
Virginia offers a concrete example: contracts signed from January will require 14-year commitments and collateral of up to $1.5 million per megavolt-ampere. The article says such terms weigh most on speculative projects, because developers must commit for years before knowing whether the capacity will be used. The rules can therefore discourage projects with uncertain customers or plans.
Rules already cover states with over half of planned capacity, and pending or promised rules could extend them to about three-quarters. The article’s risk assessment had already given speculative projects little or no weight, and the midterm results strengthen the odds that these projects will not progress. The article presents grid-cost terms, rather than outright bans, as a lasting policy change.
What does it mean for a data center to generate power behind the meter, and why might it still need grid and air permits?
Behind-the-meter power is electricity generated onsite and used by the data center without first being supplied through the wider grid. The article describes fully behind-the-meter sites as well as hybrid facilities that combine onsite generation with a grid connection. This arrangement can reduce how much power a facility draws from the grid.
But onsite generation does not mean a project avoids permits. In Texas, about half of planned data center capacity will be in facilities generating their own power. The state environmental regulator’s pause applied to these projects too: they need air permits for generators, as well as the stormwater permits required for all projects. Air permits take longer than the two-month window described in the article.
A hybrid campus still draws on the grid, so ERCOT’s hold on approvals for new large connections still applies to it. Over time, higher grid costs and rules allowing grid operators to curtail large loads may strengthen the case for onsite power, but they do not eliminate these permitting or connection issues.
How does the electric grid keep supply and demand balanced when a very large new customer connects?
The article does not lay out the technical process, so this explanation uses general electricity-system knowledge. A grid must keep generation and electricity use closely matched from moment to moment. If supply and demand drift apart, grid frequency can move away from its normal operating level, so operators coordinate resources to maintain balance.
In general, operators forecast demand and arrange for generators and other available resources to supply power as needed. When a very large customer wants to connect, its expected demand matters for planning whether the network can serve it. The article gives one example of that process: ERCOT paused approvals for new large data center connections pending an audit report and a state regulator meeting.
The article also says grid operators may curtail large loads, meaning they can require major customers to reduce electricity use under certain rules. Such arrangements can help manage pressure on the grid, but the article does not specify how ERCOT balances supply and demand or the detailed requirements for a new connection.
This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.
Read more in the JupiteX app
Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.
Or read more news on the web