The transcription of the video is below. 

Introduction and Amperon’s Founding Story 

Kerim: Hi, everyone. This is Kerim, Kerim Baran with SolarAcademy. I have today with me, Sean Kelly, of Amperon, joining us from Houston, Texas. Hi Sean. Good to have you.

Sean: Hey, thanks for having me.

Kerim: Great. And today, we’re going to talk about Amperon, which is a  new hot technology startup in the field of AI and machine learning in energy.

And Sean, thank you for being with us. And I know a little bit about your company, thanks to Nico in SunCast and the exposure that I had with you guys through that. 

Perhaps, you can tell us a little bit about your background, and your co-founder, Abe’s background and how it all came together for Amperon.

Sean: Yeah, absolutely. I mean, I think we have very different backgrounds, a little bit of the odd couple. A lot of founders have been like this, to work together. We’ve known each other a long time, etc. 

Abe and I met in the spring of 2017 at New York Energy Week, and so I’ve literally just moved to New York, full-time, like the week prior, for my, now wife. And then Abe was very interested in kind of where this energy transition was going, and so we both ended up there. 

And so my background, I spent 11 years prior to that, as an electricity trader and running power plants. And so I got asked the other day, like, “When did you actually come into cleantech?” And I was like, “Well, I’ve been here since the beginning.” 

Like literally, I was scheduling wind farms at Tenaska in 2005. That was my first job out of Texas. And so kind of I’ve been here the whole time. Like here we are, 18 plus years later, still playing with electricity. And so that was my background. Running power plants, just having a very deep understanding of what the grid looked like, where the grid was going.

And then Abe, just deep tech, one of the people, he was in New York. He’s still in New York. We started the company there. I’ve since moved to Houston back in 2019, but Abe, I mean, sold the Hacker company to Intel coming out of college was pretty IPOed at Etsy, earlier in Planet Labs, like spun up a data science, even at McKinsey, and was still early in his career when we met. 

And so we both volunteered for New York Energy Week. The real bonding moment, is they had this really cool rooftop event, and we volunteered, because it was like $2,000 a ticket to like go to the New York Energy Week. 

So we both volunteered. We both showed up there, and they didn’t let us in. And we’re like, “What the heck? We’ve been volunteering like going to all these training sessions.” And that was, we were both like, what the heck? And I think we went and like grabbed coffee or beer or something. And that’s kind of how it all started. I don’t think I’ve ever actually recorded that story before. 

And so we just kind of went back and forth and started chatting about all the things on the grid. And I think what was so interesting is you kind of have that like naivety of just asking a lot of questions. 

And whenever we have pure tech people, they ask a lot of like questions, and then I’ll say, “Well, this is the answer.” And he’ll be like, “Why is that the answer?” And I was like, “I don’t know. It just always kind of have been the answer.” 

And so I think that combination between the two of us of just having that early curiosity, plus, I mean, I’ve been networking pretty hard since I was 19 years old, and that was couple of years ago. 

And so with that, I’ve built up a really strong network. And I remember one of the first things we ever kind of like gotten to a really like interesting conversation about was called UFE. UFE is unaccounted for energy, and lo and behold, like that was one of the hidden things that was super scary during Winter Storm Uri, when the Texas grid melted down and UFE was one of those, and Abe wrote a really cool highlight on it that is still on our blog. 

And, but it was like that question was like, “It’s this number. It’s like 0.1%” and he was like, why? I was like, “This huge corporate Fortune 100 company, their chief of staff says this. That’s why.” 

And then they were wrong. And so I thought that’s really an interesting way. And that’s still us today of like, why does this work? Is there something we could do around this? Let’s be disruptive, but yet have that energy backbone, but also have that tech background. So I always joke that I view Amperon as a tech company that does energy, not an energy company that does tech. Because I’ve been in a lot of energy companies that did tech, and I sacrificed a lot of bonuses in their honor. Failed IT projects. So we definitely want to be a tech-first company, and Abe as CTO, really sets that backbone.

The problem Amperon is solving: Predicting the future by forecasting electricity demand and supply

Kerim: Got it. Good. And so tell us a little bit about the problem Amperon is solving.

Like, what exactly do you solve for the customers? What’s your approach? And you know, how does it relate to your customers? And who are your customers?  

Sean: Yeah, it’s a big question, but I’ll take it piece-by-piece. We predict the future. That’s what’s really hard. We’re the forecasting company for the energy transition. And so with that, we started back in 2018, with electricity demand forecasting. We knew that, like, EVs were coming, that rooftop solar was coming, that climate change was coming, that, I mean, we didn’t even know COVID was coming, to change everyone’s usage pattern. We didn’t even know work-from-home was coming to change everyone’s usage pattern. 

All these different things make it significantly harder to predict electricity demand. So that’s where we started. Then next, earlier this spring, we launched wind and solar forecasting. So we’re doing that right now on a utility scale basis. But we actually just kicked off site level solar forecasting which, to all our SolarAcademy friends out there, is our new fund product offering that we have all the weather data. We spend hundreds of thousands of dollars on weather data. And it’s a model. And we already have a really good solar model. So we’re taking it to site level, just like we do meter level. 

The fourth thing that we forecast is Scope 2. So Alex Robart is our Chief Revenue Officer. Before this he was Chief Commercial Officer of Sustainability at Microsoft. He is really, really on the weeds on carbon, just like I’m really, really weeds on electricity markets. 

And so the hardest portion of the whole Scope 2 is the Q, the quantity, according to Microsoft in the white paper that they put out, I guess, about 6 months ago now. 

So we’ve gone, and we’re still in proof-of-concept phase with multiple clients on that, but really excited about that, but also waiting for the government to give us a little bit more guardrails around what Scope 2 requirements are going to look like. 

And then the fifth is price. If you’ve got a really good wind forecast, you got a really good solar forecast, base load is pretty easy because it’s on when it’s on, and so you should theoretically be able to build the best price forecast. 

And so that’s something that our data science team literally just kicked off. They were at a retreat in Europe this last week, and it sounds like they made a ton of progress on that. 

So that’s the 5-pronged approach to forecasting that we’re tackling.

How does Amperon use short-term vs long-term weather data for forecast models?

Kerim: Right. And these forecasts, I assume, are consumed by utility companies, large Fortune 500 type companies, and maybe, part retailers.  

Sean: Yeah, so that it’s all over the place. So I guess, first there’s a short- term forecast. So the short-term forecast has real weather data. You’ve got, the HRRR runs every hour, runs out 36 hours. The NAM runs 72 hours, like every 3 hours. You got, the EURO and GFS both run 360 hours. 

So we’re not even running those raw models, and we’re not on solving those raw models. What we’re doing is we’re going and getting different weather vendors. So we work with some of the top weather vendors. We looked at almost everyone. But right now, with those 4 weather vendors, we use their inputs. 

And we choose between 23 and 28 different weather variables of what they’re giving us, and we look at things on a grid scale. So we have – I don’t know – I think per weather vendor is something like 20,000 points all across the United States.

And so we’re looking at weather on a really, really granular basis. So that’s the short-term model. That works for the demand, the wind, the solar and all input into the price. 

And then long-term model is more probabilistic. So we’re running a thousand Markov models, taking the P99, the P50, P1. We’re also giving people like the 15-year rolling average.

We’re giving them very climactic years such as like a 2011, 2021, 2023. Everybody is ready for it because this year’s wild. And then we’ll also go in and put in – one customer had us put in a climate change model. So that’s 15-year rolling average, plus 1, 2 and 3 degrees, minus 1, 2, and 3 degrees. 

So again, it’s probabilistic. It goes out five years, but we’re expanding it out right now and adding in kind of some macro economic variables on top of it. So really excited to continue to focus on that long-term forecast.  

What makes Amperon unique? Weather data, data-engineering, data science, customer focus.

Kerim: Got it. So when you deliver these services and your products offerings, I assume there’s a competitive landscape out there. What makes Amperon different and unique in the way you approach your offerings and your positioning in the market? 

It is part of the market that is probably leading edge, and that’s why I know very little about. Maybe you can tell us a little bit about that evolving part of the market and what Amperon special.  

Sean: Yeah. I would say, it’s really a 4-pronged approach, I guess. First off is just how deeply we dive into weather. Having our second hire being a PhD meteorologist is a little unorthodox. But Mark being there kind of helped in the early days, talked Abe through the different weather variables, the different weather vendors, was extremely helpful.

He consults with our clients. He also puts out a morning newsletter, so we look at weather deeper than anyone else. Some of our competitors actually ask the clients to give them their own weather. You can’t know what data you’re getting, and I mean, and models. It’s garbage in, garbage out. If the customer gives you bad data, you can’t be held accountable for that. But you also are going to produce pretty bad models. So weather is definitely one. 

The next piece is data engineering. Data is dirty. I mean, it’s got holes in it. It doesn’t always work out quite right. And so having people from that big tech who are dealing with the data point like in milliseconds, looking at data that often times, has a data point every 5 minutes or 15 minutes is not that big a data. So just having that deep engineering team on the technical side is, I mean, a huge advantage to what we’re doing.

The data science is the third prong. Our head data scientist was a former high-frequency trader. He’s been with us. He was our very first hire and he’s won all the load forecasting competitions that we’ve entered since 2018. So he holds the crown on the team that he has, a lot of them  have won at different load forecasting competitions, or have been very in the weeds at different utilities. So I mean, that team is unbelievable. 

And then, lastly, we have a trader-CEO. So I mean, I was one who sat in the seat. Elliott, who runs our product, he sat in the seat. I mean, there’s a number of people who actually like had to push the button and go out there and make the decision. And we’re the end-user. And so I think that 4-pronged approach has helped us. Really never been beat on accuracy. 

Who are Amperon’s key customers?

Because you are selling, too. We understand who we’re selling to. So we sell to just regular wholesale traders. So hedge fund banks, people like that. We sell to IPP (independent power producers) that used to mean like combined cycle. But now that means renewables. That means batteries. That means any power producer. So that segment’s picked up a ton – municipalities and co-ops. They are really, really focusing on what’s going on and really trying to better serve their constituents. 

Utilities, which you touched on, we had a little bit of an unorthodox approach in terms of, not focus on utilities and not doing every single like pilot in every accelerator for utilities. So we just actually signed our first utility, and I think it was client number like 90 and so we look at those. 

But then now, big C&I is super interesting. So I mean, you look and the thing to me that’s exciting and why, like our total addressable market. I always joke when you raise money, VCs care about 3 things TAM, TAM, and TAM. And so that’s all they talk about is how big is your total addressable market? 

So the thing that excites me is like Starbucks is now an electricity company. Google, Microsoft, Walmart, like what the heck? Like when I grew up in Houston, it was Exxon, Chevron, Shell – like those were the energy companies. 

And now these are energy companies, which to me is just super inspiring that you have, like the whole entire carbon push is being pushed by like Microsoft, Google, and Meta, like huh? 

So that’s what I think is really exciting to me. And that’s what makes me realize that our market is getting just bigger and bigger by the day.  

Kerim: Right. Thank you for all this insight. This is great to understand the vastness of the opportunity in the space.

And talking about that, as with climate change, we’re having more and more extreme events, too. 

How does Amperon predict extreme events?

I assume some of the biggest value you bring is around predicting extreme events, which is probably something that we should talk about for a few minutes.  

Sean: Oh, yeah. Is it a billion-dollar question or a trillion-dollar question? Probably it depends on the event. But yeah, looking at that, just a couple of examples is so, having a meteorologist on staff, he is always looking at everything. 

So when Winter Storm Uri, which happened February, Valentine’s day, weekend of 2021, and he made his first like push on that, and said, like, “This is going to be really bad”, and that was February 3rd. So February 15th, at 2 a.m.,  was when the actual lights went out and just offset like started what became a really, really nasty week for a lot of people. So I mean, he called out on February 3rd, so he called it about 10 days out.

We kind of knew everything was going down by the the 12th or 13th, but I had a number of clients who – I had one client who didn’t even end up in great shape, but they called me and thanked me, and they said, “You know, I bought $60 power, and I bought $90 power. I bought a $150 power, and I bought $300 power, and that week cleared like 7,000.” And so they’re like, “We still didn’t end up great, but we sure ended up better than we could have.” 

So that was really cool to see, and then another that I looked at is  recently. So we had, it was called Winter Storm Elliott. I don’t know when we started naming winter storms. It’s kind of weird, but this was over Christmas this last year, and so I had 3 text messages on Christmas. This was on, obviously, the 25th, but on Sunday. 

And so on the 18th, I had 3 people reach out and be like, “You’re like, 70 gigawatts of load on Christmas Day, or like this weekend. What the heck is going on? You’re like 8,000 megawatts higher than ERCOT.” And I was like, “That doesn’t seem right.” 

And so we went and looked at it. I pinged. I had a data scientist and head data scientist. And he either goes, “I’d stand by my prediction.” And I was like, “Cool. I stand by you.” And so I told them, “Hey, this is what it’s going to be.” And sure enough, like he was right. 

They had to come up massively, but all the people who saw that on Sunday were able to get out there on Monday had to appropriately to this new massive, high demand that we’d never seen before in winter, and I think it cleared about 72,000 megawatts in Texas. And so again, he gave them a week out notice by, on the model of what this demand was going to be. 

So I mean, those are just 2 very, very extreme examples when everything went great. But yeah, we’ve called, I mean, most of the events to, some degree, since inception, which is why we have such a wonderful NPS score.

Kerim: That’s great. Thank you for all this info. 

Amperon’s fundraising process and its recent $20M Series-B financing

So Sean, as we come to the end of this solar conversation, let me ask you, what other most frequently asked questions do you find yourself answering for your stakeholders? And what are some other questions that we can reserve for future solar conversations to go, maybe in deeper depth of Amperon’s offerings?

Sean: Yeah. Questions I get all the time.  

Kerim: Maybe about your team, about where you guys are at the evolution of your company, or you know, more customer questions or partner questions.

Sean: Yeah, I’ll go into the fundraising thing a little bit. See, I’d say, one of the questions I get a lot is, so we just announced our 20-million dollar Series B. 

Kerim: Congrats. 

Sean: Good. Thank you. It’s a rough year to raise money, but we were able to knock it out in a couple of months. Energized Capital led. It really, really encourage people to build relationships. 

If you’re a startup founder with VCs, we found out, I was in their LP meeting and my first email from John Tough over there, it’s February 1st in 2018, and our incorporation date is January of 2018. So literally, he met us a couple of weeks in. 

And so just building that relationship. So you don’t have to start the whole story from scratch. You can literally get to be like, “Hey, this is an update. This is where we are, et cetera.” And so I think that was really great. 

Also had D. E. Shaw, HSBC, who led our last round, came back in, or said again talking about the energy transition. They’re right at the cusp of it with a lot of the revolutionary stuff they’re doing from the wind standpoint. 

And then Veriten, which they actually announced their fund today in Houston. But I love seeing more venture into Houston. Anybody who knows me knows that I’m extremely pro-Houston. I grew up here, went to school about 90 miles from here. So it’s just awesome to see more venture come here. 

The energy transition needs energy professionals, and a lot of them live here. So yeah, I would say that that’s what I get asked a lot about. It’s just fundraising. 

And I think being capital-efficient is something that’s very important. But also, just don’t get distracted. We do one thing, and we do it better than anyone else in the world, and that’s what we’ve done, and that’s forecasting. 

And there’s been a lot of things that we’ve said no to people about, and so not getting distracted, when you’re an early stage company. We ran the company on 10 million dollars over 6 years, and now we have a fresh 20 on top of that. So really excited for where we’re going to be going from here.  

Kerim: This is great, Sean – words of wisdom, from a leader of a leading edge, AI/ML player in the energy space. Thank you. This is great, great information you shared with our audience. 

And hopefully, it also helps Amperon get new clients, new customers and prospects, and beyond. 

And we will continue these conversations with you, where we hopefully dive deeper into your tech offerings, your products, and customer case studies, and so on and so forth.  

Great. Thanks. 

Sean: That was great. Appreciate it.  

Kerim: Thank you. All right.