Listen: AI is changing the energy system, putting premium on flexibility

Dan Testa:

Hello and welcome to Energy Evolution, the S&P Global Commodity Insights podcast, bringing you weekly episodes featuring conversations with experts and industry leaders about the business of transitioning to cleaner energy. Now this episode is part of a special series, bringing updates from the floor of the S&P Global sponsored CERAWeek 2024 Industry Conference. My co-host, Taylor Kuykendall, is lucky enough to be reporting live from the show. Here's Taylor with today's guest.

Taylor Kuykendall:

All right. Welcome to Energy Evolution. We're here live at CERAWeek. I'm with president and CEO of EPRI, that's Electric Power Research Institute, Arshad Mansoor. Welcome to the show.

Arshad Mansoor:

Thank you. Thank you. Pleasure to be here.

Taylor Kuykendall:

Yeah. So, it's wild. This isn't your first CERAWeek I'm assuming.

Arshad Mansoor:

This is not the first rodeo. It's amazing. It's exciting. It's everybody that is anybody in the energy system is here today.

Taylor Kuykendall:

I think I've heard it referred to as the Energy Super Bowl, right?

Arshad Mansoor:

This is the Energy Super Bowl.

Taylor Kuykendall:

Right. And this-

Arshad Mansoor:

For five days.

Taylor Kuykendall:

Oh, and it's wild. It's crazy. We're here. There's tons of people just buzzing around. But yeah, let's dive in here. We want to talk a little bit about the electric grid. I want to say first off, kind of everybody's talking AI. So, let's start there, right? What are some of the implications that artificial intelligence has for energy systems and the transition to cleaner forms of energy?

Arshad Mansoor:

So, I think a landmark month was November of 2022. November of 2022, ChatGPT was, I call it democratized, which means it was officially released. So, anybody in the world could see the benefit of what large language model does. That was a huge impact, not just in the world of AI, but in the world of energy. Because as people found out what large language model does, the use of it, the application of it, we are just about a few steps in a 10K race.

So, now everybody's trying to make sure they have enough space, enough supply chain to build as many data centers as they can. So, all of a sudden after that last year, we started seeing data center need exploding, large 200 megawatts, 300 megawatt data centers, not just in Northern Virginia, but in every place else. It's because people don't know yet. We don't know yet how AI will impact us. But we want to make sure that anybody who has the infrastructure is going to win the AI race. So, that's why all of us certainly have seen a very rapid growth in the ask for power for AI data centers. But we don't know where we are going with this.

Taylor Kuykendall:

And so, one of the things we talk about, obviously, creates a lot of demand regarding all these data centers. But there's also some potential AI as a tool, right, to kind of manage the grid or for other services. What are some of the prospects you see there?

Arshad Mansoor:

So, it's an amazing prospect. I mean, like I said, we don't even know how it will impact society in a much broader way than just energy. But just on the energy side, the amazing opportunity with AI is how do we manage this grid? So, we are not designing the grid just to meet the demand for 1% of the time. So, let me give you statistics. There are 8,760 hours in a year. 1% of that is 87 hours. We build the grid.

We have enough generation resource of the grid. 10% of what we build is only for 1% of the time. As we do more EV, more heat pump, more electrification, we may have 20% of the grid there just for 1% of the time. What AI can do is help us to make load more flexible. So, my peak demand is not growing, but my energy consumption is growing. And I think that's the biggest impact AI could have is make our grid not just dispatch generation to meet forecasted load, but also dispatch load to meet forecasted generation because wind and solar is forecasted, not dispatched.

So, I feel AI changes the entire way the power system has operated over the last 130 years where we forecast the load and where dispatch generation. Now we will be dispatching load based on forecasted generation. But loads are millions of them. Without AI, without computation, without understanding how those millions of nodes of load works, you cannot do that. So, that may be the biggest impact of AI in the energy system, the way we operate the grid.

Taylor Kuykendall:

Yeah. No, that's incredible. Hopefully, AI is not going to replace podcast hosts anytime soon. But I want to talk about some other things that happened too because there's also cryptocurrency. And you mentioned electric vehicles. Seems like a lot of new sources of demand for electricity. Can you put that in perspective? Is this unprecedented growth? Where does it all set?

Arshad Mansoor:

So, a couple of unprecedented growth. What we are seeing in just the last two years, the electric companies, ISOs, are revising their forecast. And they're revising the forecast 2X, 4X, sometimes 8X more than what they forecasted just 18 months ago. And this is a combination of, we say that the one year, it's amazing how much difference just one year makes. And that year is from November 2021 to November 2022.

Why? November 2021, the infrastructure law was passed. In July of 2022, the CHIPS Act was passed, which is bringing chip factories to the country. August 2022, the Inflation Reduction Act was passed. Everybody knows the impact. And November 2022, ChatGPT was released. Just 12 months, four events is changing the energy world, in US initially, but it will have a global impact. And so, it is unprecedented. It is some would say feels like 1980s where some part of the region is seeing three, four, five percent year-over-year growth.

But we have to manage this in a different way than what we did in 1980. We have to manage it in a much smarter way so that energy is growing, but peak demand is not growing. I go back to if 1% of the time needs 10% of the grid infrastructure, then we need to utilize it much more effectively. And this is where load flexibility, some would call virtual power plant, some will call... How do we make sure that we are not... If there's a 200 megawatt data center that is asking for power, ideally, the 200 megawatt data center would not cause the peak load to go up by 200 megawatt.

Maybe the peak load will not go up at all because the data center is not going to be a flexible load. So, when the grid asks, "Hey, can you back off?" The data center will use their backup power, run by hydrogenated vegetable oil or renewable natural gas, and they will get off grid. And now, you have the perfect opportunity of a data center and a grid both working towards the same goal, which is to keep the lights on for everybody. We have to meet this load in a new way than we met the load in the '80s when demand was growing.

Taylor Kuykendall:

Yeah. No, that totally makes sense. Definitely a different time. One of the things I think of when we talk about all this demand coming on as a coal reporter that before I started doing all this, I can't help but think about what this means for those companies I used to cover. And I'm hearing a lot lately that they're talking about, "Hey, all this new demand, we're going to keep these plants open. We're running back over." How does fossil fuels kind of fit into all this? Did we shut things down too quickly, or what's going on?

Arshad Mansoor:

So, I think the key thing is we got to keep the lights on. Society, the global society, needs us to accelerate the transition to a clean energy future. The change in weather we are seeing across the world. So, we cannot lose focus on the big picture goal of go as fast as we can for a clean energy transition. But prudence is as important as pace. And prudence means if we cannot keep the lights on and if we cannot keep energy affordable, then that other good doesn't happen, which is going towards clean energy future.

So, with the rapid growth and with technologies like SMRs, hydrogen, they're all beyond 2030 scaling up. What do you do now? What you do now is you're seeing build out of gas. What do you do now? You are seeing in places to keep the lights on, the transition from coal is delayed. You have to be a realist in this ambitious goal for a net-zero future. And the realism is keep the lights on. So yes, will some coal plant retirement be delayed. We're already seeing this.

Are we going to be building more natural gas plant? We have to. But we have to double down so that we don't have to wait for another 10 years for SMRs, large nuclear power plants, hydrogen, long duration energy storage. We cannot wait 10, 15, 20 years for these technologies to be scaled up. And that's where you need the pace. Pace to shorten the time for innovation and deployment. Prudence to keep the lights on.

And to keep the lights on, you must have enough generation when the wind is not blowing and the sun is not shining. And the generation today you have that are dispatchable that are large, are coal, natural gas, nuclear, hydro. So, what can you build and what can you keep in the next two, three, four years? Those will be some very difficult choices that will have to be made.

Taylor Kuykendall:

So, you mentioned earlier this really productive year in terms of policy with the CHIPS Act, infrastructure, IRA. Do we have everything we need in policy-wise for the toolkits to make this new transition happen? If you were making a wishlist to accelerate the energy transition, what might you add to it policy-wise?

Arshad Mansoor:

So, as Electric Power Resource Institute, we focus on technology. We inform policy. So, if I have to inform policy, we got to build things. We got to build transmission. We got to build hydrogen infrastructure. We got to build all those things. To build you need licensing, you need permitting. So, we've been talking about licensing, permitting for a long time. There's some very encouraging news that we are seeing.

For example, DOE on eminent domain and federal land. They're trying to shorten the whole transmission permitting process to two years. Now, that will be amazing if you can do that. So, there's a wishlist of informing policy what needs to be done. I would say, shortening the time for licensing and permitting would be critical. Building infrastructure takes time. And if it takes 10, 15 years to do licensing and permitting, then the other good doesn't happen, which is building a transmission line will get me more renewables to the grid.

Taylor Kuykendall:

And so, you mentioned technology. I'm wondering what are some of those energy technologies out there that you're most excited about right now? Is there anything that's caught your eye recently?

Arshad Mansoor:

So, I would say the technology that this decade we can deploy at scale, there are two technology. One is grid-enhancing technology. Grid-enhancing technology is if I already have a grid, and if I already have a transmission line infrastructure, and I'm using it 50% of the time, why can't I not increase digitalization? Why can't I push more power on the transmission line that I already have or the right-of-way that I already have? Why can't I deploy advanced conductor in a significantly scaled up way? Why can't I do dynamic line rating?

These are all technologies that exist. We could double down, triple down on the application of those technologies. Similarly, why can we not make load flexibility a key part? Virtual power plant, some would say. Some would say, I would say, if you're a low to middle income customer and your water heater breaks, you should get a free water heater. It's not a freebie. That water heater is now part of the grid. So, when the grid is stressed, the water heater doesn't get any electricity. You still have hot water.

So, load flexibility and grid-enhancing technologies are technologies of today. We just need to scale them up significantly. Where innovation will have to come so that we can shorten the cycle is, I would say, excited about both advanced light-water reactor SMR, and Gen IV nuclear reactors. 2030 is a critical pivotal time because you have a number of these technologies where first of a kind deployment is going to happen by 2030. Very excited about long duration energy storage, 12, 24, 36 hours, mechanical, gravitational.

No longer lithium-ion with four, six hours. These are long duration. So, I would say all technologies have a role to play. Right now, next six years, double down on load flexibility or virtual power plant. And double down on grid-enhancing technologies, which is use the grid that you already have more than what we are using it today while we are building new grid. You still have to build. But building even with the licensing and permitting streamline will take time.

Taylor Kuykendall:

Yeah. No, that totally makes sense. I want to catch something that you mentioned, the water heater and that being a part of the grid and response. I was actually reading something the other day where they're doing something similar with a stove. It had a battery built into it and they were twirling back. Is there going to be a time when just almost everything in our house is feeding back electricity to the grid and being-

Arshad Mansoor:

I think not feeding back electricity to the grid, but taking electricity in line with grid's demand. So, for example, you have now countries with electrification. So, if you look at today in US, 20%, roughly 20% of the energy we use is electricity. If you look at couple of countries like Norway, Sweden, even China, 30 to 35% of the energy that they use is electricity. So, they have to significantly manage the electricity when you use electricity.

So, for example, if you don't charge your car from 4:00 to 7:00 PM, maybe that's best for grid management. So, 4:00 to 7:00 PM. If you want to charge the car 4:00 to 7:00 PM and you're willing to pay five times more electricity, go ahead and do it. But if you don't, then you get to charge the remaining 20 hours of the day at a much lower cost. I think that dynamics of real time and automated load management based on grid management is the future that we will go to. I think that's where if we don't go there, then this demand growth will cause us to build the grid where 20% of the grid is built to serve 1% of the time in a year. And that, it will not be affordable for the society.

Taylor Kuykendall:

And we've talked a lot about keeping the grid on and keeping the lights on, but in terms of load growth. But I wanted to talk about extreme events like when we're talking about hurricanes, weather like that. What do we need to do to make sure that the lights are on whenever disaster hits?

Arshad Mansoor:

Well, we launched an initiative, Climate READi. We spelled READi with an I not with a Y because as engineers we like to create acronyms. READi is RE resilience, AD adaptation, I is an initiative. So, I would say to your listener, go and type Climate READi, ready with an I. And you will see a largest collaboration that has ever been put together of not just energy companies, but NOAA, who knows the weather side, with NORC, and FAR, and Nauru, the regulatory people, the policy people, experts from national labs.

And what they're doing is they're coming up with a, I call it a transparent and comprehensive method. First, to downscale. What does Houston's weather looks like in 2045? How much flooding could we expect? How much heating could we expect? So, that's a climate modeling that needs to be done. There are many different models available, but it needs to be transparent and comprehensive. So, we are creating a methodology for climate downscaling.

Second, once you know what Houston's weather in 2045 looks like, how does it impact the way you make, move, and use electricity? How does it impact infrastructure? How does it impact planning, operations? We are putting together another exhaustive list of here are everything that could be impacted by that change. And then the third piece is, so what do you do about it now? Do you wait for 2045 to see another Harvey type flood and then take action? Or, do you take action now? What's the cost of taking that action? What's the cost of inaction?

So, we are putting together a whole list of solutions and benefit cost method. So, this work would be done by the end of this year. And we will make it available to the global, anybody can use it. It is the most comprehensive framework to understand how do you downscale climate to understand local weather in the future? How do you take that and figure out how does it impact my energy system? And then what are the things you can do today? What's the cost of doing it? What is the cost of not doing it?

So, I would say that transparent comprehensive method would be used for decades to come. Because however fast we move for our net-zero future, climate is already changing, weather patterns are already changing. So, adaptation needs to be as important as decarbonization.

Taylor Kuykendall:

That sounds like an incredible tool. Can't wait to check it out myself. Sounds like even as journalists, I think we could have some fun with that. Thank you again for joining us on the show today. For our listeners, we have a whole lot more interviews coming this weekend. It's going to be great and stay tuned for that. Arshad, thank you again so much for joining Energy Evolution.

Arshad Mansoor:

Yeah. Thank you for the opportunity, Taylor.

Dan Testa:

All right. That's going to do it for this episode. Now, before we sign off, I want to just give a shout-out to our other podcast team members, that includes my stalwart co-host, Taylor Kuykendall, Camilla Naschert, Camellia Moors, Christopher Coats, and Karen Willenbrecht, as well as our agency partner, The 199. Now to stay up to date with upcoming episodes, be sure to subscribe to Energy Evolution on your favorite platform.

And if you've got any ideas for future themes or guests, we'd love to hear from you. Please shoot us an email at energyevolution@spglobal.com. If you like what you heard today, consider sharing our podcast with others or leaving a review on your preferred platform. We look forward to bringing you more deep discussions in the future. Take care and we'll catch you next time on Energy Evolution.