Chris Seiple, vice chairman of Wood Mackenzie, a research and data company, joined Keith Martin in early December to talk about the consequences of rising US demand for electricity. The following is an edited transcript.

MR. MARTIN: Wood Mackenzie put out a report this fall called "Gridlock: The Demand Dilemma to Facing the US Power Industry.” We are going to dig into some of the consequences it predicts. How can our listeners find the report?

MR. SEIPLE: Just Google "Wood Mackenzie Horizons" or it can be found at this link: Gridlock: the demand dilemma facing the US power industry

Increasing Demand

MR. MARTIN: We went the last two decades with almost no growth in US demand for electricity. That is now changing thanks to data centers, more energy-intensive search engines, a manufacturing renaissance and electrification of transportation. How much is Wood Mackenzie predicting electricity demand will increase annually in the US over the next decade?

MR. SEIPLE: There is more load or demand that wants to connect to the system right now than can be supplied by the industry as a whole. Forecasts become less certain the more distant the projections. We are easily looking at a roughly 3% annual load growth over the next five years or about a 15% increase in total. That is 75,000 to 80,000 more megawatts or roughly 15,000 to 16,000 MW of baseload a year.

We are adding renewable generating capacity currently at a rate of about 12,000 MW a year if it were supplying power in the right locations and the right hours. Unfortunately, those capacity additions are not necessarily happening where we see a lot of the growth in demand, and it is not happening in the right hours and it is not baseload. There is a growing shortfall between demand and supply.

MR. MARTIN: How much of the load growth is attributable to new data centers?

MR. SEIPLE: We are forecasting roughly 50,000 MW of new data center capacity coming into the market over next five years, so that is about two thirds of the load growth.

There is uncertainty associated with that number. If you look at Wall Street and equity analyst forecasts of data center demand, it gets you to a lower number of roughly 25,000 MW of additions over the next five years. I said we are forecasting 50,000 MW of data center capacity. We track individual proposed data centers and we were tracking about 50,000 of proposed data centers as of this summer. Over the past six months, the number of proposed data center projects that we are tracking has grown to 100,000 MW of data center capacity.

MR. MARTIN: So we have 25,000 to 100,000 MW of potential data center load growth in five years. You are sticking to the lower end of that scale, and 50,000 MW is what you think is most likely.

MR. SEIPLE: To a mid-range. Correct.

MR. MARTIN: What is the breakdown among the other sources of load growth?

MR. SEIPLE: The bulk of the rest of the load growth is coming from energy-intensive manufacturing. That is the vast majority of the rest. A small portion of the remaining load growth is coming from electrification of transport. Most of that is in California.

MR. MARTIN: You don’t have bitcoin mining on the list. If Trump shifts US policy to promote greater use of cryptocurrency, will that be a significant contributor to load growth?

MR. SEIPLE: It will not be a significant source of load growth, but given how constrained the system is, growth of even another 2,000 to 3,000 MW becomes another constraint on the system. What I would have said if you asked me this question before the election is that the value of an interconnection to data center companies has become so high that a number of bitcoin miners were looking at converting their bitcoin mining facilities into data centers. Because of how scarce interconnection capacity is right now, that will limit how much additional bitcoin mining actually adds to the amount of capacity.

MR. MARTIN: Some renewable energy developers have been hoping green hydrogen producers would become customers for the electricity from their projects. What does Wood Mackenzie foresee for green hydrogen?

MR. SEIPLE: We see a lot of challenges with the economics of green hydrogen. Now that utilities are being inundated with interconnection requests for large loads to connect to their systems, green hydrogen production will just be another one of those large load sources that has to get in line to be able to connect to the grid. Given that they are some of the most complicated and uncertain projects to pull together, this has become yet another hurdle to scaling the green hydrogen business in the US. We do not have a particularly optimistic outlook for it.

Data Centers

MR. MARTIN: How large is the electricity load for the typical data center?

MR. SEIPLE: It is a moving target. From January 1, 2023 to July 2024, the average size of proposed data centers was increasing by about 8 MW of capacity per month. By July, the average size of proposed data centers in the United States had gotten to about 300 MW, but there are larger projects that are 1,000 MW in size trying to connect, as well.

MR. MARTIN: How is the data center demand concentrated geographically?

MR. SEIPLE: That’s an interesting question because the data centers historically have been in Virginia, Texas, Arizona, Nevada and northern California around San Francisco. There are still a large number of proposed data centers in those areas.

What we are seeing now, as interconnection capacity becomes scarce, is a wider geographic spread where data centers are trying to connect to the system. Some of the big emerging markets now are Ohio, Iowa, Indiana and then, in order, Mississippi, North Carolina, Wisconsin, South Carolina and Pennsylvania.

MR. MARTIN: We have a very congested power grid. Will it be able to accommodate the additional load in those new areas?

MR. SEIPLE: No is the short answer. That is probably the central issue that has to be solved for growth to occur in these areas.

Take New York State. New York State does not have a lot of data center activity, but New York State has two semiconductor manufacturing facilities that are under construction. Those two facilities alone could add as much as 1,500 MW of demand to the system. Just look at what happened in New York State over the past year. You have a struggling offshore wind industry. A number of such projects have been canceled. Invenergy and energyRe canceled their Clean Path NY transmission project that would have moved 3,800 MW of renewable energy from western New York to New York City.

In my prior career at GE Energy Financial Services, I was one of a large group of people that worked on permitting one of the last combined-cycle power plants to be built in New England. It took 20 years. Economic development officials in New York now are struggling to figure out how they are going to power these two semiconductor facilities.

Looming Political Issue

MR. MARTIN: This is a looming political issue if shortages of electricity become a brake on economic growth. You didn’t mention the potential 25% tariff that Trump is now threatening on all imports from Canada, including electricity imported on the Champlain-Hudson line.

MR. SEIPLE: Yes. I just returned from Washington where I met with officials from the Canadian embassy. To your point, being able to figure out how to power these facilities is at the center of economic growth and national security priorities. As much as the power industry has been focused since the election on what might happen to the Inflation Reduction Act, how we get all the new load connected is going to bubble up as one of the main political issues in the power industry.

MR. MARTIN: Grid operators like PJM and MISO have moratoria in place, some formal, some informal, on connecting new power plants to the grid. Generators cannot even get in line in some places. There has been a lot of talk about bloated interconnection queues, but the same attention has not been paid to customers trying to connect. Do you think we will see a shift in the interconnection debate?

MR. SEIPLE: When demand was not growing, it was easier for utilities not to have to prioritize and not to have a sense of urgency around connecting new generation to the grid quickly. As you said, this is bubbling up as a political issue.

Dominion Energy, which has had some of the biggest challenges in connecting new load, will be one of the first utilities in the country to do wide-scale deployment of dynamic line-rating technologies. These are technologies that enable utilities to get more capacity out of the existing system. Two utilities owned by AES are doing the same thing.

Dynamic line-rating technology will allow a battery manufacturing facility to open more quickly even though the local utility could not get the equipment needed in time to expand the grid. SPP and MISO are developing software systems that enable them to process interconnection queues more quickly. You are seeing little things like that. (For more on the potential role of grid-enhancing technologies to help, see this link: FERC Order 1920 on Transmission: How Big a Deal?)

In some parts of the country, there is a disconnect between the transmission planning being done to connect large loads to the system versus the transmission planning being done to interconnect generation. The two types of planning need to be synchronized, but it is not really happening.

MR. MARTIN: Maybe we will not see similar interconnection queues for customers like we see for generators because the customer facilities simply will not get built.

Consequences

MR. MARTIN: Let me move to some of the consequences of what we have been discussing. One is that frustration with the slow pace of interconnection is causing data center developers to look for ways to avoid utilities altogether by locating next to existing nuclear and gas-fired power plants. Amazon tried to arrange a direct power purchase from a nuclear plant owned by Talen. The Federal Energy Regulatory Commission nixed the arrangement after concluding that the effects on the grid were not properly addressed.

You say in your report, “A large prize awaits the renewable developer that finds a way to meet data center demand with a renewable, behind-the-meter supply option.” What do you have in mind?

MR. SEIPLE: I think FERC nixed this particular deal over how the type of transmission service was proposed to be paid by the data center.

It is not clear that there is any reason a grid-connected resource could not essentially exit the wholesale market solely to serve a data center, reducing the supply available to the wholesale power market.

At your annual conference last summer, we heard a hyper-scaler data center developer say it is not a price-sensitive customer. If you can deliver us firm, clean energy, we are willing to pay just about whatever it takes to be able to get that power.

Because the new data center capacity is unable to connect to the grid quickly enough, nearly everybody is looking at off-grid solutions that would supply power until they are able to connect to the electricity grid. They are searching for temporary solutions. They want to do it with clean energy. They would love to do it with renewables and storage. No renewable developer has solved the data center problem of how to deal with the need for around-the-clock power versus the intermittency of renewable generation.

If somebody can figure a way to do that, it will unlock a lot of opportunities. The vast majority of the off-grid solutions that are being proposed to date are gas-fired generation. Those are off-the-electricity-grid, behind-the-meter projects, but they will still have to connect to the gas grid, and there can still be constraints on the ability to deliver gas to these facilities, so it is not a complete solution to this problem either.

MR. MARTIN: Good explanation of the problem. Companies like Microsoft and Google are interested in small modular reactors. GE Vernova said it expects to deploy as many as 57 SMRs by 2035. Experts like Steven Chu, a physicist who was one of Obama’s energy secretaries and is now at Stanford, say that SMRs are unlikely to play a major role until 2050. What role does Wood Mackenzie see for them, and on what timetable?

MR. SEIPLE: We see nuclear energy, whether small modular reactors or large conventional nuclear reactors, needing to play an important role in power supply. Nuclear is expensive in Europe, and it is expensive in the US. There are other places, including OECD countries like Korea, that have shown it is possible to build nuclear power plants at a lower cost.

The tech companies are now starting to step up as financial sponsors to help get the nuclear business growing. Each one of them is working with a different design. This is all about building the same type of reactor again and again to get the learning and the economies of scale that are needed to bring the cost down within the US. Given where things are currently, we are looking at time frames that put us into the 2040s. Our view is closer to Steven Chu's. 

MR. MARTIN: According to your report, the US manufacturing renaissance seems to be concentrated in just three areas: batteries, solar equipment and semiconductors. Is that accurate, and what percentage of new factories measured by increased electricity load is in these three areas? 

MR. SEIPLE: Those are the three areas into which we have the most visibility and know the most about. It is accurate in the sense that this is where the bulk of the electricity consumption will be from the manufacturing renaissance. It does not cover other types of manufacturing. There is probably another 5,000 MW of electricity demand in other sectors.

We put wafer and cell manufacturing for solar at about 5,000 MW of electricity demand over the next five years. That assumes that only something like 15% of proposed cell manufacturing facilities in the US actually get built. If every cell and wafer manufacturing facility that is proposed gets built, it would be 30,000 MW of electricity demand.

This is part of the problem with which the electric utilities are having to deal. They are not seeing 5,000 MW of demand. They are seeing 30,000 MW of demand requests. They are having to figure out whom to prioritize, which facilities will actually be built, and how to address the risk that those facilities will not produce at the expected levels because of changing market demand for the products and overcapacity among global manufacturers. This is another level of complexity than has existed in the past when trying to forecast US electricity demand.

MR. MARTIN: You say in the report, “Deregulated markets are likely to be more volatile, with load growth potentially outpacing supply additions, boosting and unsettling power market prices.” Why is this true just of deregulated markets?

MR. SEIPLE: In the regulated markets, the utilities have more control because they are vertically integrated. They can decide only to connect the amount of load for which they have generating resources to support.

In the deregulated markets, and going back to the Talen-Amazon type of deal, it is not clear there is anything preventing generators from connecting directly to data centers, essentially taking that generating resource off the grid and reducing the supply of wholesale power. We are in a situation in deregulated markets where demand can grow faster than generation supply can grow.

This will be the first real challenge that we have had since we deregulated our electricity markets. The issue will be whether our deregulated electricity markets are able to ensure reliable supply. If not, we could see regulatory intervention in deregulated markets.

Merchant Financings

MR. MARTIN: What does the greater volatility in deregulated markets suggest about the ability to finance merchant power plants in those markets?

MR. SEIPLE: As you and I know and have talked about in the past, it is very rare that financing a new merchant generating plant has ever worked well for somebody over a long period of time. Most merchant plant financings have led either to write-offs or outright bankruptcies.

If you are thinking today about financing a merchant power plant, the most important issue you have to get comfortable with—and it is really challenging to do so—is what type of regulatory intervention we might see in markets and how it will affect power prices. There is a long history of regulatory intervention in markets that was not factored into financial projections.

MR. MARTIN: At the same conference you mentioned last summer that we hosted, Jan Smutny-Jones, who heads the California Independent Energy Producers Association, called what I am about to describe as the Pearl Harbor of the power sector. Rising electricity demand is expected to push up power prices, at least in the near term, because the power industry has been caught flatfooted and cannot increase output fast enough to satisfy demand. It takes time to build out new transmission and generating capacity.

How did this happen given all the integrated resource planning that is done by utilities and utility regulators?

MR. SEIPLE: It was a 70-year trend that got reversed basically in the space of one year. We had 70 years of getting less and less electricity demand growth out the same amount of GDP activity. The reversal of that 70-year trend was something that nobody saw coming. 

MR. MARTIN: Fair enough.

Transformers are a choke point. Fifty percent of utility procurement is done using Wood Mac software, so you have a fair amount of visibility into supply chains. What delays are you seeing currently, and is the situation improving?

MR. SEIPLE: We have a whole team that does supply chain consulting for renewable developers and utilities. I just checked with that team this morning. The average lead time to receive a transformer had been increasing. The situation does not appear to be getting worse. It is about three years currently for high-voltage equipment. For projects that are construction-ready in all other respects, transformers are one of the big choke points. They are not the only choke point, but they are one of the critical ones.

MR. MARTIN: Your report concludes with the following observation: “The last time the US electricity industry saw unexpected new demand growth like this was during World War II. Between 1939 and 1944, manufacturing output tripled and electricity demand rose by 60%. It was a closely coordinated national effort that brought together industry and policymakers to address the challenge. A similar effort is needed now." Do you see any sign of that happening?

MR. SEIPLE: Not yet. There are positive things happening, like the grid enhancing technology and FERC Order 1920 that move us in the right direction, but the movement is too slow to address the challenge.

World War II is an interesting time to look at because there was close coordination between industry and government, including on getting a lot of transmission built really fast.

Two things Trump said during the campaign are potentially hopeful signs.

He said, “We’re going to get energy produced fast or we're not going to be able to compete with China and other countries, so we're going to get it produced fast."

He also said, “I will declare a national emergency to allow us to dramatically increase energy production.”

We have a lot of renewable energy projects sitting in interconnection queues. Renewable energy projects are the most shovel-ready of all the projects. The question is what national policy we will have around how to get more transmission built quickly.

MR. MARTIN: Chris, it has been a very interesting discussion. We could go on for a couple hours based on the audience questions flooding in. Unfortunately, there will not be time to tackle them.