The Grid Can’t Wait for Washington to Figure This Out

There’s a version of the AI story that gets told almost exclusively in the language of software, through terms such as models, agents, tokens and parameters. It’s a compelling narrative, but it’s also incomplete.

Behind every large language model or newly built data-center campus, there’s an infrastructure question that’s considerably harder to answer than the technology question: Where does the power come from? How does it get there? Who pays for what it takes to deliver it?

These are not abstractions—they’re problems the engineering industry is uniquely positioned to address.

Our Position

The American Council of Engineering Companies (ACEC) Energy Committee recently released a position statement on what must happen to ensure our nation’s infrastructure is sufficient to meet the rapidly escalating demands of modern data centers. First, and above all else, we must be honest about the stakes. The Electric Power Research Institute (EPRI) now projects that data centers could consume between 9 and 17 percent of all U.S. electricity generation by 2030. That’s more than double current use, and 60 percent higher than EPRI projected just two years ago.

The five largest tech companies plan to spend somewhere between $660 billion and $690 billion on AI infrastructure this year alone. To put that in perspective: as a share of GDP, that dwarfs peak federal investment in the Interstate Highway System by a factor of five. That context matters. This isn’t a niche policy question. It’s a defining infrastructure moment for our country, with real implications for our economic competitiveness and national security.

Our position isn’t that growth should be slowed or redirected. Rather, it’s that growth of such magnitude requires the type of disciplined, systems-level thinking that tends to get crowded out in policy debates where enthusiasm dominates one side and anxiety dominates the other. Large, concentrated power loads must be carefully matched to verified transmission capacity. Interconnection policy needs to be grounded in grid physics and supply chain realities. Costs associated with major infrastructure upgrades should be allocated transparently in ways that protect existing ratepayers while still enabling responsible expansion.

The challenge isn’t just scale—it’s timing. The grid isn’t ready, the permitting system isn’t ready, and the time has come to reckon with how consequential that gap actually is.

The bottom line on policy is straightforward: lengthy, sequential permitting processes are the largest structural barrier to getting this infrastructure built. The solution isn’t to abandon community input or environmental review. That’s a false choice and, frankly, a distraction from the real work. The solution is a better-organized review process in which multiple regulatory agencies move concurrently with clear and enforceable timelines, and genuine interagency coordination.

Move with SPEED

The SPEED Act reflects exactly this type of commonsense, bipartisan framework. It also reflects what our industry has said for years about how to actually move projects through the system without gutting the environmental and community protections that give infrastructure its social license. The U.S. Senate should pass it.

At their core, most policy debates are really arguments about distribution: who captures the gains and who absorbs the costs. This one is no different. When a massive new load comes onto the grid, someone pays for the infrastructure upgrades such a load requires. Cost-causation principles exist for a reason. They ensure existing ratepayers aren’t quietly subsidizing billion-dollar tech companies. Transparency here isn’t just good policy—it’s a practical prerequisite for sustaining public support over the long haul.

Another Major Moment

America has faced defining infrastructure moments before: the transcontinental railroad, the interstate highway system, rural electrification, buildout of the internet backbone. In each case, our country succeeded when it matched political will with engineering expertise and moved forward with urgency.

This is another hinge moment for infrastructure. The private capital is committed. The demand is accelerating. The national-security implications are as high as they’ve been for any infrastructure challenge in a generation. What we’re advocating isn’t a radical agenda—it’s the minimum required to remain competitive in an AI-defined century.

Read the ACEC Energy Committee’s position paper on data center growth and energy infrastructure here.

The original version of this column appeared in the June 2026 Future Forward section of Informed Infrastructure.