From American Affairs Journal, Volume X, Number 2 (Summer 2026):
American public discourse has rather suddenly become electric. A confluence of factors in technology, policy, and economy has rendered energy availability and affordability a pressing debate. Several trends have pushed these issues to the fore. On one hand is the imperative of meeting growing energy demand at any cost, driven in part by the intriguing prospects of AI. Another strain of technologists wants to deploy new options for the electric system, promising sundry benefits. One example is the arrival of cost-effective batteries making electricity storage viable, which could potentially lead to less need for expensive but infrequently used generation capacity. Customers, especially residential ones, are deeply concerned about rising bills. Swings in policy undermine long-term investment incentives, with maximalist and divergent platforms pushed through by narrow margins. Adding to the boil is the prevailing insistence that all this and more must be done yesterday. Utilities and others who manage the electricity grid are caught in the crossfire, trying to determine how best to deliver a growing amount of reliable, affordable power.
Through it all, the common thread underlying each element is the infrastructure—generators, wires, transformers, and other gear—that keep the lights on. Some advocate for their favorite kit or defame their most loathed competitor, while many commentators hardly understand the complicated machinery that comprises U.S. electric grids. Informed and careful thinking is needed because today’s decisions have profound and longevous effects. The deeper issue is simpler: while everyone agrees that an electricity buildout is needed, everybody thinks somebody else should pay for it.
The main forcing factor is load growth. For nearly fifteen years, there was little aggregate growth in electricity demand, but now the intermission is over and expectations of growing demand have returned. To deliver that energy reliably, investments are needed across the system in generation, transmission, and distribution. The problem is sufficiently complex that a single solution, such as adding only generation or transmission, is not viable. The fragmented nature of the U.S. electric system complicates the problem. End users will ultimately pay for those investments, but many avenues are possible and cost-shifting is a key strategy for the interested parties.
Contextualizing the American Energy Buildup
Understanding how we got here is crucial for developing an appropriate strategy to address today’s challenge. The grid has been built and expanded over the course of many decades, matching the steady growth in demand. During the twentieth century, utilities could undertake ambitious, long-term projects and let customers grow into them. Public utility commissions signed off on these plans, rolling the cost of long-term investments into the rate base and charging customers to recover the outlays. Then, for over a decade, the aggregate growth stopped. After the Great Recession, aggregate electricity consumption in the United States hardly changed. Underlying shifts were not uniform, as some states never saw a stagnation in electricity consumption. Texas and Florida stand out as states that have seen and supported continually growing demand. More states, however, saw declines: California, Illinois, Kentucky, and Missouri foremost among them.
The aggregate trend hides some important intensive changes. Since the Great Recession, manufacturing became more energy efficient, with output growing despite lower electricity use. That trend can continue as profit-maximizing manufacturers seek to make the most efficient use of costly energy inputs. Another part of the manufacturing story is a pivot by the sector toward relatively inexpensive and abundant natural gas. But these shifts have been made and are unlikely to continue. The portions of the manufacturing sector that were convertible have largely been accounted for, and those that remain have good reasons to resist further fuel shifting.
In addition to manufacturing, there have been substantial efficiency gains in other areas. One notable area is in lighting. Switching illumination to more energy-efficient technologies like light-emitting diodes (LEDs) has led to substantial reductions in the amount of energy needed to provide the same level of illumination. Here, like the manufacturing shift to natural gas, the conversions have already been made, and one can expect few reductions in load growth from this quarter. These shifts underscore the one-off nature of the period of limited load growth that we experienced and are now realizing is over. While the above gains managed to offset the effects of population and economic growth, they will not reliably continue.
Over this period with limited sales growth, it should come as no surprise that infrastructure investment has been weak. Two explanations are immediately obvious. The first is that if the system is providing a stable amount of final sales, there is little motivation to expand. The second is that infrastructure is ultimately paid for by ratepayers, whether directly by accumulated revenues or through financing dependent on collateral from future sales. If sales are constant, there is no scope to pay for expanding the system. Now, however, there is tremendous momentum to expand the system, hence the present energy friction.
The current fear about energy infrastructure investment is that it will increase system costs. It most certainly will. Only the introduction of new, cheaper technologies can reduce the cost of provision. It is heroic to presume that expanding the system with currently available technologies is going to capture unrealized economies of scale and lower costs for all users; that prescription implies that our current system is massively inefficient. We will have to incur substantial costs; the challenge is who will pay for the system upgrades.
In classic political fashion, the answer to that question has thus far been “not me but thee.” Following the tradition of “users pay,” a salient answer might see firms investing heavily in AI bearing the lion’s share of the cost. Prominent firms like Google, Meta, Microsoft, and OpenAI are seen as creating the load problem. Some of those firms are historically valuable, so obviously, in the minds of many, they are fattened calves to sacrifice. Never mind that AI may not play out along the lines of the most optimistic projections, or that an ultimate dominant player may not be today’s incumbent behemoth. Many data centers have nothing at all to do with AI, providing a wide range of services from web hosting or cloud computing to cryptocurrency mining. Most data centers are relatively small, though the public imagination latches onto the largest hyperscaled facilities.
This has been the basis of the Trump administration’s proposal to “bring generation with you.”1 The notion that hyperscalers are going to supplant an entrenched industry like electric utilities is rather heroic. Neither party wants that to happen. If tech firms wanted to be in the electricity business, they’ve had plenty of opportunity to opt into a capital-intensive and heavily regulated, low-margin business. But they haven’t and probably won’t. Hyperscalers want to buy energy and the reliability that the U.S. grid delivers. By the same token, utilities look at the disruptive landscape of technology and hail the importance of reliability.
Segments and Issues in the Current Energy Landscape
In a rare contemporary instance of grade deflation, the most recent annual ASCE infrastructure assessment awarded a D+ to American energy infrastructure.2 The low grade is misleading, however, as it relies heavily on earlier analyses geared toward policy aspirations for emissions reduction and electric transitions within the transportation system. Now, in an environment facing growing demand after an about-face in policy priorities, the current U.S. energy infrastructure deserves a closer look....
....MUCH MORE
