Forecasting When the Grid Has No Margin for Error with Sean Kelly (Amperon)

In February twenty twenty one, power prices in Texas broke. During winter storm Uri, electricity prices hit nine thousand dollars per megawatt hour, nearly two hundred times normal levels. For some companies, that meant catastrophic losses. For others, it meant huge wins.

The difference wasn't better power plants. It was time, Specifically, having enough time to make a decision before it was too late. Shaun Kelley says his team was warning customers nearly two weeks before the storm hit.

Not that something might happen, but that it probably would and soon. That idea, forecasting is really about buying time, sits at the heart of this conversation.

Sean is a cofounder and CEO of Amperon, a company that forecasts electricity demand, wind, solar, and prices across power markets. Before that, he was a trader. He run power plants, and he lived with the consequences when forecasts were wrong.

In this conversation, he explains why forecasting has become harder, not easier, in a world where extreme weather events are becoming more and more commonplace.

He also talks through how demand is growing again after decades of stagnation, driven by data centers and electrification, and why even small forecasting misses now translate into huge financial and reliability risks. So this is a story about uncertainty. It's about what happens when businesses, like generators, utilities, and traders, don't see trouble coming early enough to react. I'm Alejandro Adiego, and welcome back to transmission.

Sean, thank you very much for joining us today to transmission by Motor Energy. It's great to have you here. We're going to start with you telling us who you are, what your role is in the energy industry, and what does Amperon do, your company.

Thanks so much for having me and for working on my schedule at a time that I'm in New York. And so with that, I'm Shaun Kelley, cofounder and CEO of Amperon. And Amperon is AI powered forecasting for the energy transition. We're at the intersection of data and energy, and help our customers make the best decisions.

And thank you for sharing that. There are a lot of things that you could forecast in the energy industry. What are some of the use cases and some of the data streams that you forecast with your product?

So we started with electricity demand. We knew demand was important. We knew that there was a lot more coming around meters and meter data that we were excited about.

But then about five years into the company, which will be eight years old in January, we realized that demand's not the entire story. It's net demand. So that's when we began forecasting wind and solar. And then lastly, in the ERCOT market, Texas, we forecast price as well.

And and when I say that we forecast demand, we do it across a number of meters. We do it across a number of portfolios, about forty million meters in the entire portfolio, and we do that on a fifteen day cadence. There's a forty five day model. There is a seven month model, and there is also a five year model.

So different different different but similar methodology that goes in all of those. And then wind and solar, we'll forecast the ISO, which is independent system operator, I e, the market as we sit here in New York ISO, and I live in ERCOT. No ISO in there, but the Texas market. And then we also will do it for your individual utility scale asset, whether it be wind or solar.

So one of your typical customers would be a utility. What other usual customers would you have?

Yeah. We started with traders and retail energy providers.

Retail energy providers, if you live in a regulated state, is there's fourteen states that you can choose your electricity provider. New York is one of them. Texas is one of them.

And so that's where we started. We did not work with utilities for the first five years until we got strong product market fit. Now we work with more than two dozen utilities. We've also gone from there, worked with public power, which municipalities and cooperatives, and even some commercial industrial clients, and as I mentioned, IPPs, which is independent power producers.

Okay. Giving a step back and going back to the origins of the company, why did you start the company? What problem did you see that was not solved yet?

Yeah. So I guess it it steps back into my career. And so when I got out of Texas A and M University in two thousand five, walked right on to a trade floor two weeks afterwards. And so went to, Tanaska and then started an eleven year trading career, that spanned a handful of companies, a number of markets, focused in the United States, traded the eastern all of the eastern markets as well as Texas, and and touches of Canada.

So that's what really helped me see the problem is I was a trader. I also dealt with a like, I dealt with retail portfolio hedging early early days when I was at a great shop called Eagle Energy Partners. They had a a retail energy provider called Champion. I was involved in that.

And then I also ran over three dozen power plants.

And in as we're here in New York, at EDF, I helped with the integration, or actually ran the integration of nine mile in Guinea to nuclear assets and then ran three other nuclear assets. So everyone wants to talk about nuclear. So just a really interesting just an interesting kind of background. And then I have my first company, which helped large commercial and industrial clients with their electricity spend.

So here, my anchor client was five five five Fifth Avenue. Really great address. They had a number of properties about eight throughout Manhattan, worked with a grocery store across Long Island that had thirty locations. So I kinda looked at having a really weird career that had seen everything power.

And as we sit here, power markets are essentially twenty five years old kinda since circa two thousand, and I've caught twenty one of those years. So that's what I that was my background.

So I was living in New York at the time. Started the company here in New York, and the thesis was twenty seventeen was awesome time to be in New York. Walked around. Everyone you met is a cofounder disrupting, working in Google or Facebook or Amazon, and it was just a vibrant, vibrant startup scene.

We didn't and don't exactly have that in Houston. New York, tons of startups. Houston, where I live now and where I grew up, energy. Everyone's in energy.

Your dad works in energy. Your mom works in energy. Like, everyone across the entire city works in energy. New York people weren't paying attention to that.

So I love arbitrage. And so I looked at, wait, energy, I e, the only thing I know how to do, and where I live, tech, haven't actually ever met each other.

What if we do this? So I went to New York energy week, now New York climate week, and I volunteered and got to see a bunch of great stuff. Someone who's now become a friend, Jigger Shaw. I got to see him speak.

And there was another volunteer named Abe Stanway. That's my cofounder. So we're a a New York energy match. And then we just realized that if he built it and I sold it, we'd be onto something.

So those are the early roots of just seeing the problem that energy transition's hard, and we were getting better meter data, fifteen minute intervals instead of once a month reads. We were then getting more wind. We were getting more solar. It was getting harder.

The turbines were getting bigger. The solar installations were getting larger, So that means the misses are bigger and more impactful from a p and l, like profit and loss standpoint. And then weather got weird. I mean, in my early days, you looked at hurricane season and you would pay attention because it would come into the Gulf Coast and all oil was being drilled out in the ocean and it would make oil prices like or gas prices go through the roof.

Now you're coming and you had just crazy winter storms that just wiped out everything. Winter storm Uri, winter storm Elliott during Christmas. We'll probably dive into those a little bit more later. But you have a one in a hundred year event every, like, month or two that mathematically is hard to get around whereas the first, like, ten years of my career, there's kinda nothing to see. So with all those problems that we solve, if I could hire these people that I was meeting here in New York and hire those best engineers, data scientists to solve these problems, then we'd be onto something. And that's exactly what we've done.

And I can imagine that at that point when you were starting to think about your own company, the spectrum of all the potential problems that you could tackle, you had to narrow down the scope and focus on one first customer, one first pilot. Could you walk us through what that was and how it felt like?

Yeah. So we had a we had an incubator that was trying to give us money, and two guys just met, had an idea. And so we sat back and they oh, do you have any customers? Do you have any whatever letter of intent? Any I was like, oh, okay. And on a Thursday, I left and met with them on Monday, and I gave him two signed papers.

One customer that had thirteen thousand electricity meters in Texas, and the other one had twenty eight thousand electricity meters in Texas. And they're like, woah. Woah. Woah.

Just acquired forty thousand customers overnight? I was like, yeah. It it took four days. And so that's when we realized that, wait, we can continue with the retail energy providers.

And then going back to the product and a lot of the thesis, we our first blog that Abe wrote on our website is talking about smart meter data. So a lot of the thesis was around smart meters because we were getting much more granular data. Our iPhones are sitting here picking up to the millisecond, and we're super excited in energy about one data point every fifteen minutes. And so that was where the initial was.

And then I think the next thing that we did is we said no a lot. We did a ton of customer interviews. We talked to everyone who we possibly could. And the thing that I always say is if you want money, ask for advice.

If you want advice, ask for money. So we went in and asked for advice, and then we went back to the person. I said, Alex, hey. Look what I built.

And you said, that is a genius idea. And I go, I know, man. You told me six months ago to build it. And you're like, I did.

I mean, I would pay for this.

Perfect. And so it worked out exactly how I wanted it to, and that was the early stages of the company. It's just listening to the customer and still to this day, we do customer stories till we're blue in the face and no no one will still talk to us out of the one hundred and fifty plus that we have today.

Thank you for sharing those stories. And now drilling down into that product, which is the forecast, AI driven forecast, and would you start with load management, if I understood it correctly?

What does it make it different to other players in the field? Why is it so special, and what did you bring to the table?

Yeah. I mean, to an extent, it was a late mover advantage. The companies that we're competing against, one started in nineteen ninety two, one started in nineteen ninety seven, one started early two thousands. And so we had a late mover advantage.

We built on day one in the cloud and have done everything in the cloud. We've also run multiple AI powered machine learning models and they've retrained every single hour since November of twenty eighteen. Now as we sit here at the end of twenty five, that's a lot of that's a lot a lot of iterations on the model. And so that's what's really helped us and that's the advantage that we have is we were hiring against the first data scientist we ever lost was to Google DeepMind.

If you're losing your data scientist to Google DeepMind, you hired the right data scientist.

So all of your forecasts are AI driven? There's no fundamental analysis behind?

There is a number of models.

We start with linear regression, old faithful Yes.

And then AI gradient boosted trees, a number of secondary models, and then we ensemble them together. No one model is going to win all the time. Like we talked about at the beginning, the world is crazy, and so not one model is going to win the entire time. Not one weather forecast is going to win every time.

So we have six different in some way, shape, or form weather vendors that we use. And so that's where you wanna run an ensemble of all of these. And by ensemble, just a blending together to pull the strengths of each models and overcome the weaknesses. Yeah.

So when these first customers starting using your products, what were the results that they were seeing?

They were saving hundreds of thousands, if not millions of dollars. And the two that I was seeing that surprised me. And as a trader, I thought accuracy was the most important thing.

I would still argue it's the most important thing, but there's two other things that surprised me is time. How much time do you have in any given week? You and I and Elon all have the same amount of hours in any given week. We were putting back we had one customer say who had managed multiple portfolios. He goes, you put ten hours back in my week.

I would love ten hours back in my week. If I could have an additional ten hours, that would be a beautiful thing. That's what we gave away. And then the next piece that I didn't understand is that the interfaces and energy are terrible.

And so with that, we've had the front end engineer that I was I wasn't worried about losing him to Con Ed. I was worried about losing him to Facebook or Meta. And so that's what I was worried. So our front end was best in class, and you were able to ish and get things so easy through the API, download the CSV, use the user interface, go through one of our channel partners.

And you were able to look at it how you wanted to look at it, and that was what was so important. We were asking for comparisons, and I remember asking our first customer. I said, Matt, can you, like, download this to show the comparison? He goes, I can't.

I was like, why? And he goes, because you just asked me to download six months and each month takes two hours to download and I don't have twelve hours to spend on this. Yeah. And our model is sitting there refreshing every single hour and he's complaining that he can't download and ours takes, I mean, seconds.

And so that those were the two advantages I didn't realize that we had on top of accuracy.

And for our audience to to follow, to make sure that they are following, when you say that you saved hundreds of thousands or even millions of dollars in costs, what does this actually mean? How did you save them cost?

Yeah. So one of the early, I guess, mid points, the first really, really tumultuous event we dealt with was winter storm Miri. For those of you who don't know, that was Valentine's Day of twenty twenty one.

It was we lost power for about seventy two hours as I was living in Houston. Our power went out at two o'clock in the morning on February fifteenth that Monday. Prices then stuck at nine thousand dollars per megawatt. A normal price at that time would have probably been fifty, sixty dollars a megawatt.

So nine thousand dollars per megawatt is nine dollars a kilowatt hour. New York has really high prices. We were talking about this yesterday. It's fifty cents a kilowatt, and that's all in all of the fees.

This is the energy component only is nine dollars. In Texas, I pay, like, about twelve to fourteen cents depending on the time of use. And wait. Twelve to fourteen cents, and now I'm paying nine dollars?

From a budgeting standpoint, that's huge. And so we told our customers this was coming on February second. Wait. Hold on.

I just told you a disaster was gonna happen on February fifteenth on February second. You've got plenty of time to prepare. You can't go, like, an entire new power plant or something like that, but you could buy a sixty, sixty five dollar power and continue to buy it. And so we have clients who bought sixty dollar power and sold it for nine thousand dollars.

We had customers who made hundred million plus dollars, multiple customers that still remain nameless, who made over a hundred million dollars because we were able to tell them on February second, February third, this is coming, and we stayed on on the fourth, on the fifth, on the sixth. Winter storm Elliott, it's not good to shut everyone's power out. Don't do it on Christmas, please, because we're coming up on Christmas. It's just not nice.

Definitely puts you on the naughty list. We called winter storm Elliott eight days in advance, and so our customers were able to go there and prepare for that. If the utilities had been using us, people would have probably had a better and less memorable Christmas. And so those are the things that calling that eight days in advance is helpful, and we'll talk, I'll talk more a little bit about that midterm forecasting, but we just made a call that it's gonna be cold in Texas on December second.

We made that call on October thirtieth. While we were wearing shorts and getting ready to sweat out Halloween, we told everyone that it was gonna get freezing in Texas in thirty three days. Everyone thought it was nuts except for the people who went and bought forty dollar power and sold it at a much higher price.

And in order to call out those potential extreme weather events like storm URI, what different data streams do you use in your models? What is your competitive edge?

Competitive edge is I mean, a, data's messy. And so going in and having top flight data engineers who can go in and clean the data is extremely important. The next is weather. The way we look at weather, many of our competitors were looking at the nearest airport.

I don't live by an airport. I don't want to. Wife's a very light sleeper. It would not end well.

Then we had a bunch of engineers who just looked at data and came from data companies.

What they did is got open source Microsoft data of all the commercial buildings throughout the United States and then population density weighted our weather points.

And we ended up with about forty thousand weather points all across the United States. And then we took twenty three to twenty eight variables from those and then weighted those in our models, and then that gave us the edge. And that still gives us the edge today. And then running those models every hour, retraining those models every hour has now created where, as I talked earlier, we had a late mover advantage of being cloud native. Now we have an advantage from having about forty million meters of a hundred and eighty in the United States and and counting.

Quick break. If you listen to this show, then you probably work in energy. And at Modo Energy, we're not just talking about the energy transition. We help our users to actually make it happen.

All energy storage, solar, and wind assets on the global balance sheet need to be valued and benchmarked, and that's where Moto Energy comes in. Our benchmarks and forecasts are transparent, bankable, and trusted by the world's leading banks, asset managers, utilities, and developers. So if you want to learn more, go to modo energy dot com. And if you want more content like this, sign up to our weekly dispatch newsletter.

Enjoy the conversation.

You mentioned that you have midterm forecasting models. You also have short term and long term. Can you walk us through the three different types and the differences and what different customers would use each model for?

Yeah. So short term forecasting is what we're known for. And so the weather models, the ECMWF or the Euro, then the GFS, the Canadian, they run three hundred and sixty hours. Three hundred and sixteen hours is fifth three hundred and sixty hours is fifteen days.

Great. So we update those every hour. You get new weather data every hour or two or three hours depending on the vendor, and that's what pretty much everyone can use who has short term, any type of, like, operational risk. Battery companies are great.

Have a bunch of those as customers. That's where we're targeting. Everyone can use some variation of that model because electricity flows every minute, hopefully. And then the with that, we went to the midterm forecast recently, and it's just an expansion of the short term.

And it's honestly what it is is AI, I don't believe is going to take my job anytime soon, but I do believe it's gonna make my job a ton easier, especially when it comes to weather. We've seen great strides in weather here in the past year and a half. Our meteorologist has been doing this since working at NASA in the eighties, and he says he has literally seen more doctor Mark Shippum in the past year and a half than he has in his entire career.

That's unbelievable. And so now there's models that are going out forty five days that give six hour blocks, and then we cut that down into hourly. And then there's models that go out seven months that we then in six hour blocks that we then cut down to hourly. And then there's about to be a model that will go thirteen months, from the ECMWF euro, and that'll be out, in q one.

And then we're gonna as quick as we can get our hands on it, we're ready to go. So that those are the midterm forecast. And, again Yes. Everyone has exposure there.

So midterm forecast, short term forecast, long term forecast.

Now looking at the spectrum of different uses, you said load management. What are other type of uses for your models?

Yeah. Operational, if you own an asset. So you're sitting there and saying, is my wind turbine gonna put? Like, what's the output of my wind turbine?

What's the output of my utility scale solar installation? So they've obviously got a very clear use case because we have a better view on how sunny or how windy it is and curtailment and all of the different elements that go into our asset level forecast. We also public power and utility who are going in and their goal is to, as I gave them a hard time earlier, keep the lights on, and so they wanna understand what risk you're coming. And then we also have traders.

They're trying to make money. So that's something else that we really lean in on our sellers.

Know who you're selling to. Traders' goal is to make money, whereas municipalities, cooperatives, utilities, even really retail energy providers, their goal is to mitigate risk.

And you can use the same tool for that just in different formats or different use cases. And then we also work with some large commercial industrial. We've been doing this for a while, and we have, I think, almost ten data center customers on the platform, which is all anyone wants to talk about these days.

I see. Is there have you ever faced a moment where you had to confront a loss in focus because of this vertical, but also horizontal expansion into different lines?

Yeah. That's a great question. Absolutely. We have you only have so much capital. We've raised about thirty eight million to date, US.

And so with that, you can only do so many things. And we're eight years old, which seems like forever and also very young at the same time. And so we've definitely put more focus in to a certain product than another product and then you realize that the other product has lagged. So we've done something recently and dove into big rocks.

And so what I mean by big rocks is the leadership team sits down before every quarter and says these are the priorities that we're going to have. We still do other things, but these are the priorities that we're gonna have for the quarter, make sure we get these done. And so that has really helped us be able to have focus and also have a data driven solution. If one product's not selling a ton but isn't worth sunsetting and another product is taking off but still has a little bit of expansion work we can do, then we double down on that.

So as a data company, we try to make data driven decisions.

Okay. Thank you for sharing that as well. Is there with all the data that you have access to and all the people that you talk to in the industry, are there any key trends that you're seeing right now in the market from the demand side or generation from renewables that you could share with the audience?

I I was just having a conversation right before this with a with a journalist, and he brought up that Jimmy Kimmel is talking about energy. Wait. What? Why is mainstream media late night talk show host talking about energy?

This used to be something that just like us nerds would sit in the corner and talk about even though everyone was using electricity every single minute. This is now what people are talking about at Thanksgiving and at their Christmas dinner. It's literally just a daily topic. I joke that I've been doing the same thing for twenty one years and have been cool for about one, one and a half.

And so I think that's the big thing that is changing. And I also people realize that we didn't have demand growth for the last twenty years because we had demand growth and then we had the energy efficiency movement, which is great. We changed from really bad lighting to LED lighting and saved a ton of money that way. Tim Healy, who's we're fortunate enough to have served on the board, started a company called Internak, which really brought in demand response.

So we're backing power down.

We even upgraded Windows. We did so many building upgrades that demand was flat or even going down.

Like, why are you starting a demand forecasting company? Now I did not call this entire AI movement or I'd be well retired now, But now demand is on the way up. This is all anyone wants to talk about. The ISOs are coming out saying we're not sure if we have enough power.

Everybody's moving to Virginia. Everyone's moving to Texas. Everyone's finding just any spot they can possibly get their hands on for data centers, which is load growth. And so that's what's so exciting about right now is that it's energy security.

If we're gonna win this AI race against China and whoever else, we have to have more demand. And so energy is the top of everyone's mind. Electricity is the top of everyone's mind for the first time in my forty three years on Earth.

Yeah. And you mentioned the massive growth in data center and AI demands.

What are you seeing in your data that most people could be missing out or misunderstanding?

What everyone's saying is gonna happen is gonna happen, but at a slower scale. We're not gonna, like, click our heels together and have a hundred and fifty gigs in Texas. We're right now a really, really high peak day is, like, close to ninety. It's not gonna be a hundred and fifty in a year or two.

It's gonna take a minute. We're having huge problems in this country in permitting, getting things online fast enough. Also, we're turning into BYOG, bring your own generation. Wait.

But I can't get a gas turbine until twenty twenty nine or twenty thirty. Interesting. So as as everyone is fearful of what's happening, we've got a minute, but we've only got a couple minutes. And so we need to act fast, and we need to work as hard as we can to get more generation out there, get more batteries out there, be more efficient, use the resources we have better, and understand what's coming and mitigate these natural disasters that are gonna come because climate's gonna do what climate wants to do.

One more question that I'm curious about is, I can imagine that different regions have different availability and granularity. How do you confront that problem? And are there any regions where you offer more deeper and more complete products here in the US compared to other regions?

Oh, man. I just I just got a headache just listening to this question. The data is it can be rough, and I've thankfully don't personally have to deal with it. I've hired a really talented team led by Kalpenar.

Our first hire was Ito, and we're able to go in and and get through this data. And, yes, every market is different. So the most complete now I'm just gonna sound biased here, is ERCOT. It's Texas.

We get data from them that's quite clean, lots of smart meters on a three day lag.

The worst, hopefully, don't get booed out of here as I'm sitting in New York City, is New York ISO. They give things on a sixty to ninety day lag. It sometimes comes by CSV. It sometimes come by Carrier Pigeon, And so it's definitely our worst product, like, in worst forecast because that data lag is massive.

If I was wrong, please tell me in a couple of days, not in a couple of months. And so that's how we have our most complete. And then in between there, have different degrees of PJM on a five day lag. A lot of the other markets kind of average that, but those are kind of the the best and the worst.

And we've launched in Europe as well. And so we're active in twenty countries in Europe. And so we've built out sixteen on a forecast TSO basis. Transmission system operator, it's the same thing as ISO, independent system operator here in the US.

It's the market or the country. And so some of them have good data, and some of them have much more difficult data.

Have you seen any key differences differences in developing businesses in Europe compared to the US expanding there?

How long is this podcast? Is it twelve hours or eighteen hours? Yes.

I've seen a lot of differences.

I did a lot of research. I went over and I went over to Europe in twenty twenty two and started things the same way I started them here. I took as many meetings as I could. I asked for as much advice as possible and got a good idea.

What I would have done is hired a European product manager earlier. We've got an amazing one in Bilbao, and I would have hired him. If Alberto had come about a year earlier, it would have been lovely because the data differences each country is different. Each country speaks different languages or a lot of them speak different languages.

The the data access is different. The how the market set up is different. The Netherlands has the most difficult data because it's very, very advanced in terms of behind the meter, so it's our worst forecast by a lot. It's everyone's worst forecast by a lot.

And so there's so much that I didn't know. The advantage that we had here is we had so much US knowledge. I had a lot. Elliot had a lot as our first commercial hire.

As I said, we're turning eight years old in January. Elliot just had his six year anniversary, and so that's where we had that advantage of just knowing the market so in-depth. And if we didn't, there was someone in our phones who did, and we could call them and ask them for advice. Europe, we didn't have that advantage, and so there's a bot a lot that we've learned.

I will say that John Ecker and the team over there are doing a great job, and we're very happy that we've stood up so many customers. And the other thing too is we've built a truly global product.

Things that I did not think we would be doing in twenty twenty five. We have customers in Estonia, in Latvia, in Ukraine, in Jordan because they called and said, Amperon, we would love to have a wind or solar forecast. And we said, great.

We need your latitude and longitude, and we need any historical data you can give us and curtailment data and all of the data, and then we're able to stand those up, and then let them see a really good forecast that really changed things for them because no one else was focusing on those specific countries that we're able to come in and have a truly global product that's applicable in every single country in the world.

Now jumping to the end, is there anything that you would like to plug or promote to our audience? This is the moment.

Alright. I'll take it.

As I've talked about, the midterm forecast is what we're so excited about. And the reason why is the data's only been there for, like, months. And so we've had weather forecast going out fifteen days for decades. And this new data is months old, and there's stuff we're waiting on in February. So that's what's so exciting.

And we're kind enough that Naren Malik from Bloomberg covered it in the first of first of its kind exclusive in Bloomberg, and everyone always looked at those terms from simulations because there was no real weather data going out that far. So you'd run simulations. You'd look at similar years, but that only takes you so far. Real weather data is key.

So the next thing too is there's a cash trader, which focuses that's what I was. Focus on same day to, like, next month. Well, the markets are having a harder time making money because natural gas prices are low, renewable penetration tie. It's great.

There's sometimes when the sun goes down and the wind joins it that is volatile, but it's getting harder to make the money that you need for your seat cost. So you're trying to go out the curve a little bit more. You're trying to trade not just January and February, but see if there's some easy money in March and April. Then a term trader what a term trader normally trades years.

So they're looking at what's my twenty twenty six, my twenty twenty seven, twenty twenty eight position. As we know in the US and abroad, the macro is kind of a mess. And so they're like, I don't really wanna bet on what's gonna happen in twenty twenty nine. I would prefer to move in a little bit.

So that seven month window is exactly where everyone wants to be. And, again, very proud of the team for building the first of its kind, a real weather, hour granularity forecast that helps more people than we've served previously, the cash and the term trader, which makes up most of The Trade Desk.

Thank you for sharing that. Jumping to the final question, we always ask this question. Is there a contrarian view that you hold about the energy industry many people would not share with you?

So I've been saying this for a long time, and it's getting less contrarian. But do you want natural gas? Do you want renewables? And the whole entire time, I've said all of the above.

And we had the this administration had gone in all in on natural gas. The previous administration had gone in all in on renewables, and my thought is we need everything. We're gonna run out of power. We're having too many blackouts.

Weather's too crazy. We need generation. And so that's something that I've been saying for as long as I've had a stage, and now people are slowly coming around to it. I think this is becoming, hey.

This is a great idea. I've also been pushing nuclear since two thousand nine or when I was working at EDF Electrodes de France, helped with that integration of nine mile in Guinea. I've loved nuclear since then. It's reliable.

It's clean. It's the best of both worlds. So I've been pushing that since two thousand and nine and saying we already had the option Since the nineteen fifties, and now everyone's coming around. So I guess they're less contrarian now, but I've been saying them for a long time and I guess that maybe somebody listened to one of my podcasts.

I don't know.

I agree with that view.

Thank you very much for joining us today. It was a pleasure to have you here.

Thank you so much for having me.