Abundance, Not Additionality, Will Meet the Energy Demands of AI

American electricity consumption is growing for the first time in generations. And though low-carbon technologies such as solar and wind have scaled impressively over the past decade, many observers are concerned that all this new demand will provide “a lifeline for more fossil fuel production,” as Senator Martin Heinrich put it.

In response, a few policy entrepreneurs have proposed novel regulations known as “additionality” requirements to handle new sources of electric load. First suggested for electrolytic hydrogen, additionality standards would require that subsidized hydrogen producers source their electricity directly from newly built low-carbon power plants; in a Heatmap piece from September, Brian Deese and Lisa Hansmann proposed similar requirements for new artificial intelligence. And while AI data centers were their focus, the two argued that additionality “is a model that can be extended to address other sectors facing growing energy demand.”

There is some merit to additionality standards, particularly for commercial customers seeking to reduce their emissions profile. But we should be skeptical of writing these requirements into policy. Strict federal additionality regulations will dampen investment in new industries and electrification, reduce the efficiency of the electrical grid through the balkanization of supply and demand, and could become weapons as rotating government officials impose their views on which sources of demand or supply are eligible for the standards. The grid and the nation need a regulatory framework for energy abundance, not burdensome additionality rules.

After decades of end-use efficiency improvements, offshoring of manufacturing, and shifts toward less material-intensive economies, a confluence of emerging factors are pushing electricity demand back up again. For one, the nation is electrifying personal vehicles, home heating, and may do the same for industrial processes like steel production in the not-too-distant future, sparked by a combination of policy and commercial investment. Hydrogen, which has long been a marginal fuel, is attracting substantial interest. And technological innovation is leading to whole new sources of electric load — compute-hungry artificial intelligence being the most immediate example, but also large-scale critical minerals refining, indoor agriculture like alternative protein cultivation and aquaculture, and so on.

In recent years, clean energy has seemed to be on an unstoppable path toward dominating the power sector. Coal-fired generation has been in terminal decline in the United States as natural gas power plants and solar and wind farms have become more competitive. Flexible gas generation, likewise, is increasingly crowded out by renewables when the wind is blowing and the sun shining. These trends persisted in the context of stable electricity load. But even as deployment accelerates, low-carbon electricity supply may not be able to keep up with the surprisingly robust growth in demand. The most obvious — though not the exclusive — way for utilities and large corporates to meet that demand is often with new or existing natural gas capacity. Even a few coal plants have delayed retirement, reportedly in response to rising demand and reliability concerns.

Given the durable competitiveness of coal and especially natural gas, some form of additionality requirement might make sense for hydrogen production in particular, since hydrogen is not just a nascent form of electric load but a novel fuel in its own right. Simply installing an electrolyzer at an existing coal or natural gas plant could produce hydrogen that, from a lifecycle perspective, would result in higher carbon emissions, even if it displaces fossil fuels like gas or oil in final consumption. Even so, many experts caution that overly strict additionality standards for hydrogen at this stage are overkill, and may smother the industry in its crib.

Likewise, large corporate entities and electricity customers adopting additionality requirements for their own operations can bolster investment in so-called “clean firm” generation like nuclear, geothermal, and fossil fuels with carbon capture. In just the past month, Google announced plans to back the construction of new small nuclear reactors, and Microsoft announced plans to purchase electricity for new data centers from the shuttered Three Mile Island power plant, the plant made famous by the 1979 meltdown but which only closed down in 2019. Three Mile Island’s $100-per-megawatt-hour price tag would have been unthinkable just a few years ago but is newly attractive.

Notice the problem Microsoft is trying to solve here: a lack of abundant, reliable electricity generation. Outdated technology licensing, onerous environmental permitting processes, and other regulatory barriers are obstructing the deployment of renewables, advanced nuclear energy, new enhanced geothermal technologies, and low-carbon sources. Additionality fixes none of these issues. Of course, Deese and Hansmann propose “a dedicated fast-track approval process” for verifiably additional low-carbon generation supplying new sources of AI load. Yet this should be the central effort, not the after-the-fact add-on. The back and forth over additionality rules for the clean hydrogen tax credit is a case in point. The rules for the tax credit will (likely) be finalized by January, but lawsuits already loom over them. Expanding this contentious additionality requirement to apply to broad use cases will be even more contentious without solving the actual shortage data center companies care about. Conversations about additionality are a distraction and misplace the energies of policymakers and staff.

Substituting one regulatory thicket for another is a recipe for stasis. Instead of adding more red tape, we should be working to cut through it, fast-tracking the energy transition and fostering abundance.

With such broad requirements, what’s to stop future administrations from expanding them to cover electric vehicle charging, electric arc furnace steelmaking, alternative protein production, or any politically disfavored source of new demand? Could a second Trump Administration use additionality to punish political enemies in the tech industry? Could a Harris Administration do the same? What if a future administration maintained additionality standards for new sources of load, but required that the electricity come from fossil fuels instead of low-carbon sources?

Zero-sum regulatory contracts between sources of electricity supply and demand are not simply at risk of becoming a tool for handing out favors on a partisan basis — they already are one. Two pieces of model legislation proposed at the July meeting of the American Legislative Exchange Council, an organization of conservative state legislators that collaborate to write off-the-shelf legislative measures, would require public utility commissions to prioritize dispatchable generation and formally discourage intermittent renewable sources like solar and wind. One of the proposals suggests leaning on state attorneys general to extend the lifespans of coal plants threatened with retirement.

These proposals did not move forward this year, but it is unlikely that the motivating force behind them is exhausted. And whatever one thinks of the relative merits of intermittent versus firm generation, ALEC’s proposals demonstrate just how easily gamed regulations like additionality could be and the risks of relying on administrative discretion instead of universal, pragmatic rules.

This is not how the electric grid is supposed to work. The grid is, if not an according-to-Hoyle public good, a shared public resource, providing essential services to customers large and small. Homeowners don’t have to sign additionality contracts with suppliers when they buy an electric car or replace their gas furnace with an electric heat pump. Everyone understands that such requirements would slow the pace of electrification and investment in new industries. The same holds for corporate customers and novel sources of load.

The real problem facing the AI, hydrogen, nuclear, geothermal, and renewables industries is an inability to build. There are more than enough clean generators queueing to enter the system — 2.6 terawatts at last count, according to the Lawrence Berkeley National Laboratory. The unfortunate reality, however, is that just one in five of these projects will make it through — and those represent just 14% of the capacity waiting to connect. Still, this totals about 360 gigawatts of new energy generation over the next few years, much more than the predicted demand from AI data centers. Obstacles to technology licensing, permitting, interconnection, and transmission are the key bottlenecks here.

Would foregoing additionality requirements and loosening regulatory strictures on technology licensing and permitting increase the commercial viability of new or existing fossil fuel capacity, as Deese and Hansmann warn? Perhaps, on some margin. But for the foreseeable future, the energy projects and infrastructure most burdened by regulatory requirements will be low-carbon ones. Batteries, solar, and wind projects make up more than 80% of the queue added in 2023. Meanwhile, oil and gas benefit from categorical exclusions under the National Environmental Policy Act, while low-carbon technologies are subject to stricter standards (although three permitting bills recently passed the House, including one that waives these requirements for new geothermal projects).

Consider that 40% of projects supported by the Inflation Reduction Act are caught up in delays. That is $84 billion of economic activity just waiting for the paperwork to be figured out, according to the Financial Times. Additionality requirements are additional boxes to check that almost necessarily imply additional delays. Permitting reform makes them redundant and unnecessary for a cleaner future.

This underscores perhaps the most essential conflict between strict additionality requirements and clean energy abundance. Ensuring that every new policy and every new source of demand allows for absolutely zero additional fossil fuel consumption or emissions will prove counterproductive to global decarbonization in the long run. Natural gas is still reducing emissions on the margin in the United States. Over the past decade, in years with higher natural gas prices, coal generation has ticked up, indicating that the so-called “natural gas bridge” has not yet reached its terminus. Even aggressive decarbonization scenarios now expect a substantial role for natural gas over the coming decades. And in the long term, natural gas plants may prove wholly compatible with abundant, low-carbon electricity systems if next-generation carbon capture technologies prove scalable.

The United States is the world’s energy technology R&D and demonstration laboratory. If policies to prune marginal fossil fuel consumption here stall domestic investment and scaling of low-carbon technologies — as current permitting regulations already do, and proposed additionality requirements would do — then we will not only slow U.S. decarbonization, but also inhibit our ability to export affordable and scalable low-carbon technologies abroad.

Environmental progress’s surest path is in speeding up. For that to happen, we need processes that allow for rapid deployment of clean energy solutions. Expediting technology licensing, fast-tracking federal infrastructure permitting, and finding opportunities for quicker and more rational interconnections should be first and foremost.

The real solution lies in building a regulatory environment where energy abundance can flourish. Clearing the path for clean energy development, we can achieve a future where energy is affordable, reliable, and abundant—a future where the United States leads in both decarbonization and economic growth. It’s time to stop adding barriers and start speeding up progress.