For the past year, it’s been clear to Andy Lubershane that the U.S. is entering what he calls the “electricity gauntlet.” Like many people who are immersed in the power sector, he saw the early signs — and now the rest of the world is starting to wake up to it.
Lubershane coined the term “electricity gauntlet” to describe the treacherous path between two big challenges for the power grid: On one side is growing demand, driven by new electric vehicles, factories, and data centers; on the other side, new power supply is stalled, slowed by interconnection queues, stagnant transmission capacity, and other bottlenecks.
“I feel like a year ago, I was getting some nods and some agreement that this is a concern that is certainly on the horizon,” he said. “I think in the past three to six months, the nods have gotten more and more vigorous to the point where, now pretty much everyone in the industry recognizes that we’re in the gauntlet,” said Lubershane.
Lubershane is a partner and the head of research at Energy Impact Partners, a climate-tech venture capital firm. In an appearance on Catalyst with host Shayle Kann, who is also an investor at Energy Impact Partners, Lubershane pointed to a recent spate of national press covering the issue, like the New York Times, the Wall Street Journal, and Washington Post.
One particular trend is driving attention to the issue: the energy intensity of artificial intelligence.
“The reason that it has become so, so much at the forefront of the public attention over the past three to six months is predominantly due to AI, which is hot and sexy in general,” Kann said.
Stuck in the middle are utilities, who are scrambling to find enough power to meet rising demand. Utilities like Georgia Power and Dominion Energy have recently released resource plans saying they will need much more generation than expected, even this year. Meanwhile, the National Energy Reliability Council has released an assessment showing that growing load growth raises reliability risk in new parts of the country.
The stakes for the climate are high. Utilities, including Evergy and First Energy are turning to carbon-intensive options, like extending the life of coal-fired power plants that were set to close and planning the construction of new natural gas power plants.
Although Lubershane is still “very bullish” on the growth of wind and solar, he believes variable generation won’t be enough. “I don’t see a way that we make it through this first phase of the gauntlet without adding a significant amount of more natural gas fire generation in North America.”
Lubershane and Kann talked through four technologies that could provide an alternative to traditional gas plants.
Time for a nuclear renaissance?
“There’s been a lot of people for the past decade, at least, that have been sort of crossing their fingers and hoping and asking this question, like, when will nuclear have its time?” Lubershane said. “Is there the potential for a real nuclear renaissance?”
Nuclear is an especially attractive solution to meet new load growth because of its constant supply of power. For example, Amazon purchased a nuclear-powered data center site in Pennsylvania from Talen Energy. The data center, positioned behind-the-meter, will be able to draw up to 960 megawatts from the 2,500-megawatt Susquehanna nuclear power plant.
“A shocking level of consensus within the community of power system planners and utilities and grid operators…believe that at some point, a nuclear renaissance will be necessary,” said Lubershane.
But unlike gas, new nuclear power cannot get built quickly. In the U.S., only one new nuclear project has come online in the past three decades, years late and billions over budget. New reactor designs promise to cut costs and construction times, but only one new kind of reactor has received a construction permit from regulators so far, the first in 50 years.
“Nuclear is not going to solve our problem for the next five years,” Lubershane said.
When will geothermal have its break-out moment?
Geothermal is a lot like nuclear: It’s a firm, zero-carbon power source that’s been around for decades — but has struggled to build new capacity. That’s because the shallow hydrothermal needed to power turbines is confined to only a few geographies.
But companies like Fervo are utilizing drilling techniques from the oil and gas industry to drill deeper and create artificial wells, potentially making geothermal power accessible in more places. Drill deep enough, and you can build a power plant just about anywhere and get what Lubershane calls a “nuclear fusion-esque resource.”
The problem is, drilling that far down — as far as 20 kilometers — is hard and expensive. The deepest hole ever drilled is only 12 kilometers.
“So the question is,” Lubershane said, “how much can we expand geothermal capacity beyond that sort of almost accidental resource that we have today?”
He pointed to intermediate solutions — places where new techniques could unlock geothermal heat but don’t require 10- or 20-kilometer depths.
“There’s a lot of interesting approaches out there that are doing medium-depth geothermal that could probably make geothermal a resource that’s viable across, let’s say, most of the Western half of the U.S.” he said.
In the western U.S., there’s enough heat for power generation just one to two miles down, according to the National Renewable Energy Laboratory. And so Lubershane is glad to see investment in new companies like Fervo that are exploring new geothermal technologies.
“I do think it’s another one of those resources we could come to rely on in the 2030s,” he said, adding however: “But unfortunately, yet again, this doesn’t solve our like three-to-five year problem for the most part.”
Get some GETs
One quicker set of solutions are grid-enhancing technologies, or GETs. While it’s notoriously hard to build new transmission lines, there are a few ways to send more energy through existing wires, such as dynamic line rating and power flow controllers.
Lubershane is especially bullish on dynamic line rating, which uses sensors and analytics to tell grid operators, in real time, whether a particular line can carry more energy.
Transmission lines have static line ratings that tell operators how much power they can safely carry, but those static ratings are conservatively safe, formulated to avoid danger in the most extreme scenarios.
But those extreme scenarios happen only a fraction of the time, meaning that most of the time, operators could send more power through the lines.
“If you have more data on the real-time status of the line, whether that’s temperature, wind conditions, the amount that the line is sagging, the way that it’s moving given the wind, that sort of thing, then you can make real time decisions about moving more power through the line,” said Luberhsane. “And, you know, this is a really, really compelling, elegant idea.”
Compared to building power plants, sensors and analytics software are much faster and cheaper to deploy.
“There are additional constraints that make it more difficult to realize the benefits of this sort of thing in practice than it may be sounds on paper,” he added. “But I am hopeful, and I do think that there are going to be increasing numbers of transmission lines and regions where adding dynamic line ratings makes more and more sense to wring as much value as is really physically possible out of the transmission system.”
Future-proofing natural gas
“For all of the gas that we are going to be building in the next five years, what is the best path for future proofing it?” Lubershane asked, rhetorically.
If natural gas is going to be built anyway, he argued that those plants should prepare for a low-carbon future, such as by using carbon capture and storage (CCS), hydrogen, or designs that allow for low capacity factor.
CCS would allow those plants to operate at high capacity while also minimizing emissions. Lubershane suggested a range of ways for a natural-gas plant to prepare for CCS.
“From the bare minimum of, like, site your plant somewhere that’s kind of near where you might be able to put carbon in the ground to starting to do some studies to actually designing the plant in such a way that would facilitate CCS retrofit later on down the road.”
Another option is to prepare natural-gas plants to run at a lower capacity factor as peaker plants that ramp up only during peak loads. Peaker plants may still be valuable even if cheap zero-carbon power displaces higher-capacity thermal generation. One way to prepare plants to operate at low capacity factor is to ready them to run on hydrogen made from zero-carbon power or a gas-hydrogen blend.
“Green hydrogen is not really an energy resource. It’s a storage resource in a way. It’s a way of taking renewable energy and later on, at some point, running it through a natural gas power plant, which you’re not going to want to do very often,” he said. “It’s a low capacity factor resource.”
One more option is to design gas plants to be smaller, distributed resources that can provide resilience services at the edge of the grid. “Basically, it’s also a backup power resource,” Lubershane said.
Are growing concerns over AI’s power demand justified? Join us for our upcoming Transition-AI event featuring three experts with a range of views on how to address the energy needs of hyperscale computing, driven by artificial intelligence. Don’t miss this live, virtual event on May 8.
