Strategic Thinking on Electric Transmission in the Age of AI

Introduction

Artificial intelligence (AI)–fueled electric demand growth, combined with industrial reshoring and economy-wide reshoring require the U.S. power sector to grow by upwards of 20 percent in the coming decade. At stake is U.S. leadership in the most strategically vital technologies and industries of the future. Given these stakes, federal policymakers need to assess the role of the federal government in ensuring transmission planning and investment serve the strategic national interest of meeting demand growth while maintaining globally competitive standards of reliability, affordability, and emissions intensity.

Three recent developments in the mid-Atlantic power market of PJM provide a case study on the interaction between demand, generation, and transmission. These developments—a boom in datacenter demand in northern Virginia, a $14 billion surge in PJM capacity costs, and the planned restart of the Three Mile Island nuclear reactor—are literally linked together by the high-voltage transmission system in the region. A close look at this case clearly identifies the strategic value of an expanded transmission system and should compel federal policymakers to approach the issue as a key strategic priority.

A Story in Three Parts

Northern Virginia, just outside of Washington, D.C., has rapidly become the single largest datacenter market in the world. Nicknamed “datacenter alley”, this small corner of Virginia hosts over 400 datacenters, which based on planning reports from Dominion, the electric utility serving the region, represent nearly 4 gigawatts (GW) of demand in 2024. Booming demand sourced from cloud services and AI is expected to drive demand to over 13 GW by 2038.

Booming demand has consequences. Demand growth from datacenter alley (combined with market reforms to accurately reflect generator reliability) led to massive price increases in PJM’s July 2024 capacity market results. Prices cleared at $270 per megawatt (MW) per day up from $29 per MW per day in the prior auction. Unsurprisingly, the Dominion zone which hosts datacenter alley cleared even higher at $444 per MW per day. These prices are paid by demand-side consumers to generators on a forward basis to ensure that sufficient capacity is available to meet future demand peaks. All told the yearly capacity cost to customers rose to $14.7 billion up from $2.2 billion in the previous auction.

Price increases like this are not without political consequences. Governors from five states have written PJM asking for changes to capacity market rules that would bring down prices in future auctions, though these same changes would clearly undermine the viability of the market construct as a whole. PJM has delayed its next auction by at least six months to assess potential changes and avoid another sky-high pricing result that would intensify political backlash. Ultimately, the market construct is delivering a price signal that industry and policymakers should pay attention to: new generation resources are desperately needed.

As if on cue, in September of 2024, Constellation, a major independent power producer in PJM, announced the restart of the Three Mile Island nuclear reactor. The roughly 800 MW reactor is due back online in 2028. The restart is made possible by a new power purchase contract with Microsoft, which as one of the hyperscaler technology firms, is one of the largest consumers of electricity in PJM and includes a large footprint in Virginia’s datacenter alley.

The return of Three Mile Island to the resource mix is an important first step to mitigate capacity prices and meet surging demand. But a single large investment in a generation does not resolve the issue. For one, PJM’s scenario modeling suggests far more capacity, something on the order of 50–100 GW, is needed in the coming decade. And two, generation alone cannot solve the challenge facing the mid-Atlantic power grid.

The Role of High-Voltage Transmission

The transmission system acts as the connective tissue delivering supply to demand and is the physical infrastructure on which the “market” for electricity depends. The Three Mile Island nuclear plant is situated in southeastern Pennsylvania. If this supply is to serve datacenter demand in northern Virginia it must travel over the high-voltage transmission system. But the transmission system that links these two regions is inadequate for the task, a fact empirically borne out by massive capacity price spreads between the Dominion zone (home to data center alley) and the rest of PJM.

Broadly stated, the PJM grid suffers from poor integration, which limits the ability of both existing and potential new generation resources to cost-efficiently and reliably serve demand. For example, PJM hosts a large coal-fired power fleet in states like Pennsylvania, Ohio, and West Virginia. These plants have in recent years run at low utilization rates and many were slated for early retirement on economic grounds. Now, these power plants could play a role in serving growing demand, but not if there is insufficient transmission capacity to move their output over the Appalachians into the demand center of Northern Virginia and the D.C. area.

The same holds true for new-generation resources. So many solar, wind, gas, and storage projects are backlogged in PJM’s interconnection that a new project entering the queue today may not receive interconnection until 2030. Long queue delays and project-killing network upgrade fees are symptoms of a transmission system with little to no spare capacity. An underbuilt high-voltage transmission limits the ability of existing and future generation capacity to efficiently and reliability serve demand growth and bring down prices in PJM’s capacity and energy markets.

PJM and its member utilities are aware of the problem and planning has identified numerous transmission projects to boost transmission capacity in the region. For example, the Maryland Piedmont Reliability Project is roughly 70 miles of new 500 kilovolts (kV) high-voltage transmission that will move power north to south from Pennsylvania into Maryland and northern Virginia; in effect, connecting Three Mile Island to datacenter alley. But the project faces headwinds having activated local opposition and drawn in Maryland’s governor who is expressing “concerns” about the project’s benefits to Maryland. Given that approval from the Public Service Commission of Maryland is required to proceed, prospects are poor for the rapid construction of this critical infrastructure.

Historical Headwinds

These same headwinds have for decades killed off similar projects that would be immensely valuable to serving current grid demands. A typical example is the Transource project, first proposed in 2015, which sought to build a series of high-voltage 230-kV transmission lines from Pennsylvania into Maryland to help relieve transmission system congestion and serve demand in the northern Virginia area. The project was classified as an economic efficiency project because it was determined as not strictly necessary to maintain grid reliability. Multiple rounds of analysis demonstrated that net benefits to consumers far exceeded project costs, but despite this clear value to the grid, the Pennsylvania PUC denied approval of the project in 2021 citing a lack of benefits for Pennsylvania ratepayers. Though federal courts overturned this decision in 2023 in all likelihood the project will never proceed.

The history of failed transmission expansion in this region stretches back even further. In 2007 PJM approved a utility proposed plan to build a 765 kilovolt (kV) transmission line stretching 275 miles from West Virginia to Maryland. The PATH Project would have connected the 765 kV transmission system that runs through the western portion of PJM, and all the generation assets in that region, to demand centers on the East Coast. Another example was the proposed MAPP Project, a 500 kV transmission line running 230 miles from New Jersey to Virginia.

Both projects faced opposition from local landowners, permitting delays, and opposition from states worried about benefits to their ratepayers. Eventually, both projects were scrapped. When proposed these projects would have hurdled cost/benefit ratios and improved overall system efficiency in the near term. A decade plus later, these “backbone” transmission projects would provide significantly more slack in the system, allowing for emergent sources of demand and supply to connect as they arise rather than face delay.

Past policymakers cannot be judged harshly for failing to foresee the current AI boom and the demands it creates for new transmission. Their choices were made in the context of decades of rapidly slowing electric demand growth and a broad economic slowdown following the 2008 global financial crisis. But the past presents lessons on the value proposition of strategic-scale transmission. In the present, policymakers must calibrate for the new era of AI, computational scaling, and electricity-intensive industry. Policy should rapidly advance no-regrets transmission investment that delivers both near-term value to ratepayers and long-term flexibility to the system.

Thinking Bigger on Transmission

The PJM case study demonstrates two important characteristics of transmission that are overlooked and unaccounted for in current policy. First is lifespan: high-voltage transmission systems last up to 80 years or longer, while the rights-of-way hosting these lines effectively last forever, providing reusable space for future generations to accommodate new higher-capacity transmission or other energy and public infrastructure of the future.

The long lifespan of transmission enables its second key feature, flexibility. The mix of generation resources on the supply side and the mix of consumers on the demand side changes over time and grows very uncertain beyond medium-term horizons. Once built, transmission capacity creates flexibility for new commercial demands and new generation resources to be added as needed, without delay. The failed projects above present a counterfactual example of this value. A realized example is the American Electric Power (AEP) 765 kV backbone transmission system in Indiana, Ohio, and West Virginia. Built in the 1970s and 1980s to deliver coal and nuclear-fired generation to serve growing demand; that same network today plays a central role in attracting datacenter demand to the AEP service territory. In an era of demand growth, high voltage transmission capacity ensures the grid acts as an enabler rather than a constraint.

The long view of the U.S. energy landscape understands that the most strategically vital industries and technologies of the future are uniquely electricity intensive. These commercial and technology trends drive a macro result, which is a growing electricity intensity of GPD for the first time in decades. contextual factors suggest a strategic value for transmission that is not adequately captured in current planning models or current policy. Crucially, state policymakers who today act as the primary gate on transmission investment have no incentive, authority, or jurisdiction to procure transmission based on this strategic value; it is simply beyond their remit.

In short, there is a clear role for federal authority and federal dollars in ensuring transmission system investment in line with the strategic national interest. Federal siting authority should be expanded to ensure valuable transmission projects are not held up by narrow state political interests; this could be done through expanded and automatic back-stop siting authority for projects authorized by regional planning processes or over a certain voltage. This would place electric transmission closer to the standard for natural gas transmission siting, recognizing that all energy transmission infrastructure is firmly in the national interest and crucial to a U.S. strategic advantage in energy on the global stage.

Federal investment should be expanded to reduce the near-term cost impact on ratepayers of a massive scaling of investment in the sector underway in generation, transmission, and distribution systems. Federal cost-sharing effectively buys down the state-level cost-allocation debates which heavily constrain high-voltage transmission investment. Department of Energy funding for SPP-MISO seam projects demonstrates the ability of federal dollars to deliver vital grid-expanding investments that otherwise are bogged down in planning and cost-allocation politics. Federal cost-sharing should be directed into transmission projects that demonstrate increased speed-to-power for load and increased speed-to-interconnection for generation. Crucially, this support should extend to grid technology and reconductoring solutions that could deliver transmission capacity increases in the near term at low investment cost, creating breathing room on the grid while larger grid expansion projects proceed over longer timelines.

Conclusion

How should federal policymakers think about the role of transmission given the context of surging electric demand growth?

The broad bipartisan agreement that exists on national security and technology leadership implications of AI must be extended to solving policy challenges in the electric power sector. Pressing issues abound in the sector including leadership in nuclear energy, the gas-midstream, and gas-electric coordination issues. But transmission policy must not be deprioritized given the clear strategic value presented by this case study. To let the dysfunctional status quo on transmission persist would be a strategic failure.

Cy McGeady is a fellow with the Energy Security and Climate Change Program at the Center for Strategic and International Studies in Washington, D.C.