Data Centers Want Power. Regulators Say Wait.

Google’s Eemshaven, Netherlands, data center. Source: Google.

Bringing online enough power for new data centers is not just an engineering challenge; it’s also a regulatory and bureaucratic gauntlet, especially in the United States. The U.S. electric utility system faces well-known hurdles: lengthy permitting processes for new power lines and plants, congested interconnection queues, fragmented oversight across federal, state, and local authorities, and communities resistant to new infrastructure. These factors can make it slow and complex to deliver power to data centers, even when companies are ready to invest. For example, in Northern Virginia – home to the world’s densest cluster of data centers – the local utility (Dominion) has seen lead times of 3–5 years to energize new large data center campuses due to the need for new substations and transmission upgrades. Permitting a high-voltage transmission line often involves multi-year environmental reviews and public hearings; one such line to support data centers in Loudoun County (VA) sparked opposition because it cut through neighborhoods, illustrating local permitting challenges. As a result, in areas like Loudoun County, power availability has become a gating factor for data center construction, forcing some projects to pause until grid upgrades catch up.

Another pain point is the interconnection queue backlog – the waiting list of generation projects seeking to connect to the grid. In the U.S., over 2,600 GW of new generation and storage is stuck in interconnection queues as of end-2023, more than double the total existing U.S. generation capacity. This backlog affects data centers in two ways: (1) it delays new renewable and other power projects that data centers might be counting on for supply, and (2) in some cases data center operators have proposed building their own on-site generation (like gas turbines) which then must also navigate interconnection studies. U.S. grid operators are straining to process the surge in requests; PJM (the Mid-Atlantic grid) noted that its interconnection applications skyrocketed from a few dozen a year to thousands in recent years, overwhelming legacy processes. Permitting and regulatory delays are thus a serious bottleneck. As analysts warn, adding the required grid capacity will need upto $2 trillion of grid spending through 2030 in the U.S., and transmission projects can take 5–10+ years to approve and build, potentially bottlenecking data center growth if regions aren’t proactive. In the interim, some hyperscalers have even had to throttle back expansion plans or seek creative workarounds due to power constraints.

(Image credit: LBNL)

By contrast, some other countries offer more streamlined or coordinated approaches. For instance, in Scandinavian countries (Sweden, Norway, Finland), governments actively attracted data centers by guaranteeing relatively quick grid interconnections and tapping ample existing capacity from hydroelectric and wind resources. These countries treat large data centers almost like heavy industry, integrating them into national power planning. Land and grid planning is often centralized, which can reduce bureaucratic friction. Ireland, on the other hand, provides a cautionary tale: after a data center building spree in the Dublin area (where data centers jumped to ~14% of Ireland’s electricity use by 2022), the Irish grid operator EirGrid declared a de facto moratorium on new data center connections around Dublin until grid reinforcements are built. This was an extreme step taken to protect grid stability, essentially “no new data centers until we upgrade our wires.” Ireland’s energy regulator is now proposing strict rules for any future data center connections, such as requiring them to have on-site generation or load flexibility, and to locate in regions with available capacity. The Irish example highlights how some countries are aggressively managing data center growth to avoid overstressing the grid. However, it also shows the risk: EirGrid recently warned the government of a potential “mass exodus” of data center investors if they cannot get connections. The best practice internationally appears to be early planning and clear rules: regions like Singapore temporarily paused data center development in 2019, then re-opened it in 2022 with a carefully managed quota and criteria (favoring highly efficient designs and requiring use of renewables/imported clean energy). This ensured new data centers aligned with Singapore’s limited grid capacity and climate goals. The Netherlands similarly instituted a temporary ban on mega-data centers in 2019–2020 to develop a policy framework; it now requires stricter sustainability and spatial rules for large facilities (for example, waste heat reuse and renewable energy sourcing are encouraged/mandated for permits in some Dutch provinces).

The U.S. regulatory environment, being decentralized, means responses vary by state. Virginia, with its data center hub, has so far taken an accommodative approach – offering tax incentives and expedited local permitting – but is now confronting community backlash over land use, noise, and power infrastructure. In Texas, policymakers are actively debating how to handle a wave of AI-driven data centers on the ERCOT grid. Texas Senate Bill 6, which includes a “kill switch” allowing grid operators to temporarily disconnect data centers that consume more than 75 megawatts during grid emergencies, was signed into law by Governor Abbott in June 2025 and is now being implemented. The law also requires data centers to share the costs of necessary grid upgrades. The rationale is that ordinary Texans should not bear the full cost of new infrastructure needed solely to serve data centers, and that these facilities must act as responsible “grid citizens.”

Travis Kavulla, Vice President of Regulatory Affairs at NRG, explained: “The law’s intent is to make sure large loads pose as little reliability risk to the system as possible and are not drinking the milkshake of all other Texas power customers.” Ensuring fair cost distribution and reliability is the goal, though the industry warns overly strict rules could drive away investment. Other states like Georgia and California are likewise considering measures such as special rate classes for data centers (to allocate grid costs) and requirements for using renewable energy or diesel backup limits.

From a broader perspective, international best practices emphasize proactive planning and collaboration. Some strategies include: designating specific zones for data center development where the grid is strong (as Japan has done, channeling new data centers to regions with excess power or underutilized plants), requiring “grid impact” assessments as part of data center approval (so that the cost of needed upgrades is identified and funded upfront), and encouraging co-location with power generation. Co-locating data centers near existing power plants or renewable farms can minimize new transmission needs. This is being explored in parts of Europe and the U.S. (for instance, putting new data centers on sites next to decommissioned coal plants that still have grid interconnects and transmission capacity). In the U.S., regulators like the Federal Energy Regulatory Commission (FERC) are starting to factor large new loads into grid planning scenarios; a shift from the past when only generation was the focus. FERC in 2023 ordered reforms to require transmission planners to consider future load growth (including large industrial/data center additions) in their long-term plans. This should help anticipate needed upgrades sooner.

Yet, despite these efforts, bureaucratic delays remain a top concern for industry leaders. A recent McKinsey analysis noted that just procuring the electrical equipment (transformers, switchgear) for a new data center can take two years due to supply chain backlogs, and getting utility approvals can add years more. The bottom line: in the U.S., red tape and grid bottlenecks are emerging as perhaps the biggest check on unfettered data center growth. Other countries have sometimes been more agile. For example, Sweden built new substations in record time to accommodate Facebook’s Luleå data center, and Norway fast-tracked grid links for data centers to use surplus hydro power, but even they face constraints on how much new load can be added without significant grid investments. As one industry expert put it, the challenge is not only building enough power plants for data centers, but also “wiring” that power to the right places in time. Creative solutions and regulatory flexibility will be needed to break these bottlenecks while maintaining reliability.