Tracking the Next Phase of Energy Demand

As demand for power accelerates alongside AI infrastructure, William Blair’s Jed Dorsheimer, group head of the energy and power technologies sector, joined Yahoo Finance’s Market Catalysts to discuss his inaugural Racks & Stacks Monthly, a new report tracking data center buildout, power dynamics, and industry sentiment.

The Racks & Stacks Monthly report argues that reliable, dispatchable power is becoming a constraint on U.S. economic growth. What’s driving that shift, and why is it becoming such an important issue right now?

“Dispatchable” power refers to electricity sources that grid operators can turn on or off to match consumption needs. Reliable, dispatchable power is becoming a constraint because the global system that allowed the U.S. to outsource energy-intensive production is starting to break down. For decades, the U.S. benefited from a model in which manufacturing moved overseas while it focused on higher-value services, supported by stable trade dynamics and foreign capital flows.

That’s now shifting. As geopolitical tensions rise and countries like China challenge the existing order, the U.S. is being pushed to reshore more industrial and strategic capacity, while also managing higher debt and less reliance on foreign buyers. That brings the focus back to the physical economy, and with it, the need for energy at home.

The challenge isn’t just generating more power—it's generating the right kind. Data centers and advanced manufacturing require constant, reliable energy, more like steel mills or refineries than traditional commercial buildings. That makes dispatchable sources, such as natural gas, nuclear, and hydro, more important than intermittent power alone.

We see this as a structural shift, not a cyclical one. As investment moves back toward physical assets and industrial capacity, reliable power is emerging as a key constraint on U.S. economic growth.

You estimated that data centers could account for more than half of U.S. electricity load growth through 2030. What does that mean for power systems as demand grows over the next decade?

Data centers are expected to account for more than half of U.S. load growth, but they’re really a proxy for something bigger: a broader resurgence in industrial activity. Whether it’s data centers, semiconductor fabs, or chemical plants, the common thread is a shift toward more energy-intensive, always-on demand.

That has a few important implications. First, load shapes are changing; these are large, concentrated users that require high reliability and operate around the clock. Second, it puts pressure on a grid that wasn’t built for this kind of demand after years of relatively flat growth. And third, it’s inherently inflationary. Expanding generation and infrastructure takes capital, labor, and materials, which creates both cost and timing challenges.

Capital is available, but investors need clearer signals—faster approvals, more certainty around offtake, and the confidence that projects won’t get stuck in regulatory limbo.

JED DORSHEIMER, Group Head of Energy and Power Technologies

For power systems, it means the era of complacency is over. As demand grows, supply—especially firm, dispatchable supply—has to keep up. If it doesn’t, you start to see higher prices, tighter reserve margins, and declining reliability. Over the next decade, the regions that move early to secure that capacity will be in the strongest position.

The analysis also points to a potential power supply shortfall by the end of the decade. What are some of the biggest obstacles preventing supply from keeping up with demand?

The biggest obstacle is that the grid is being asked to do two contradictory things at once. Demand is rising sharply, driven by industrial growth, while policy and market structures have often encouraged the retirement of reliable, dispatchable power in favor of a more variable supply that doesn’t fully address reliability.

On top of that, there are several bottlenecks. Permitting and siting take too long, delaying critical infrastructure. Interconnection queues are backed up and increasingly difficult to navigate. Equipment and labor are constrained, from transformers and turbines to skilled electrical workers. And in many markets, pricing doesn’t fully reward reliability capacity, or fuel security. This leads to underinvestment in the assets the system now needs most.

There’s also a disconnect with financing. Capital is available, but investors need clearer signals—faster approvals, more certainty around offtake, and the confidence that projects won’t get stuck in regulatory limbo.

The issue isn’t a lack of resources or capital. It’s that the system was built for a different era of demand and a different geopolitical backdrop, and it hasn’t fully caught up.

Nuclear power, particularly small modular reactors (SMRs), has received renewed attention. What role could nuclear play in helping close the supply gap?

Nuclear has an important role to play because it’s one of the few scalable sources of carbon-free, dispatchable baseload power. If the U.S. needs more reliable electricity to support reindustrialization, AI infrastructure, and advanced manufacturing, nuclear has to be part of the solution.

That said, there’s a difference between its long-term role and near-term reality. The existing nuclear fleet is incredibly valuable and should be preserved wherever possible. Uprates, relicensing, and restarts may be more impactful in the near term than people expect.

SMRs are promising because they offer more flexible, repeatable deployment and can better match demand at industrial sites or data center hubs. But they’re unlikely to solve near-term supply gaps at scale without faster progress in permitting, supply chains, and execution.

Nuclear should be viewed as both a bridge and a destination, extending the life of today’s fleet while building toward a more advanced nuclear foundation over time.

As you track these developments through your new Racks & Stacks Monthly report, what indicators will you be watching most closely?

One of the key indicators we track is sentiment through our Racks & Stacks survey. Earlier this year, for the first time, it dipped into contradictory territory, driven by concerns around capital availability, the ability of renewables to meet demand, and growing pressure on labor and project timelines. That may prove to be a warning signal, although it’s too soon to say whether it reflects a broader trend or a short-term reaction to geopolitical events.

It’s also important to separate sentiment from fundamentals. Despite that shift, both demand and supply remain near record levels. The system isn’t slowing—it's showing signs of strain as it adjusts to a new level of growth.

The underlying theme is simple: computing is becoming a powerful business, and power is becoming a central economic issue. The key question is whether the system is moving quickly enough to keep up.

JED DORSHEIMER, Group Head of Energy and Power Technologies

We’re also watching structural bottlenecks closely, including power pricing, interconnection timelines, permitting, and construction cycle times. These are the indicators that show whether the system is adapting to a new demand environment or still operating under outdated assumptions.

More broadly, we’re looking for signs that the market is starting to price reality. Are utilities and regulators treating this as a structural shift, or a temporary cycle? Are large energy users accounting for the true cost of reliability? And are policymakers prioritizing firm power in a way that reflects physical constraints?

At its core, Racks & Stacks is designed to track those dynamics in real time. The underlying theme is simple: computing is becoming a powerful business, and power is becoming a central economic issue. The key question is whether the system is moving quickly enough to keep up.

Accessing the Racks & Stacks Monthly Report

For more information on Dorsheimer's research, including the Racks & Stacks Monthly report and his broader coverage, please reach out to your William Blair representative.