Battery finance: from backtest to bankable | Jacopo Buriollo

Okay, hello everyone, I'm Jacopo Buriollo, yeah, hard name and hard last name, so sorry about that, yeah, I'm the founder and CEO of Megawatt, Megawatt finances utility scales, solar and battery storage projects across Europe, and most of you work on the numbers behind the battery in terms of forecasting, trading, development, development and today I want to look at those numbers from the capital side where someone has to basically decide how much of a battery is worth financing and in particular I want to focus on a gap that there is between what the battery earns and what actually gets financed, what portion of it normally gets financed and we will look also at Britain to measure that gap because it has the longest record in Europe. So first of all let's start with a car which is a weird thing to do but let's start with a car and you know every car normally comes with a fuel consumption figure, right, that is normally taken in a lab so in a lab they have perfect condition, very strict temperature, a driver that never breaks too hard, perfect road and so on, right, so it's a real measurement but I guess nobody everybody here experiences that kind of experiences that kind of consumption on a motorway in January unfortunately and well a battery back test is kind of the same thing, it's that kind of lab figure, right, it's something that expects perfect foresight that basically thinks the battery is always available and that you know it doesn't take into account fees or degradation so it's great but normally it's not really true, right, so that number, the back test, normally then starts a journey where you remove the perfect foresight and you are left with what normally an optimizer has to work with and then you basically take out the hours where the battery isn't really available and then you take out the fees and the degradation, the wear and the cost of balancing and so on and normally you ask the question that every investor would ask which is what is what is what is what is what is what is what is what is what is left after a bad year, right, and you are left with that green bar which is basically the number where the debt is sized against, right, everything in between is normally where all the arguments happen and it's normally invisible from the outside. Now, we started with the back test, right, now we started with the back test, the other way to start from is starting from published revenue benchmarks, right, and normally these benchmarks can sit quite far from each other. For example, if we take the German market, we have same market, same month, August this year, right, and the strongest published revenue benchmark for two-hour batteries was like 62% higher than the lowest benchmark, right, even if we had the same prices, the same auctions, why is that? Yes, well, because there were five assumptions that actually decide this difference that are the foresight, so basically we ask did the model know prices in advance, right, the availability, how often was the battery assumed to be online, cycling, you know, how many cycles and was the wear counted, the usable capacity, like are we using a full charge range or are we like actually taking into account the margin that operators keep, you know, and then of course costs, like where fees taken off like where fees taken off or not, like where fees taken off or not, so don't get me wrong, a benchmark is normally right for the question it answers, the trouble normally starts when you take a benchmark and you lift it out of its context and you dropped it into a financial model as a revenue case, you know, and that's normally where the problem starts, and we can basically say that what we saw earlier, the backtest is basically the ceiling, the benchmark is an anchor, right, and the real work is basically moving from both down to what a real battery normally keeps. And out of those five assumptions that I mentioned earlier, I decided to dedicate one slide in particular to availability, because it's normally one of the assumptions that people think it's solved, and of course it's not. And so Accure, with the help of the US Bank, actually made a study among 50 operating projects right now, and it found out that three out of four of them were actually over-reporting their availability, right, and this normally comes down to definitions, because we normally define a battery as available because we normally define a battery as available because it's online, right, if it's online, it's available. In reality, a lot of batteries that are online, they do not answer when they are called, right, and so this is actually an issue from an investor perspective, from the financing side, because revenue is normally measured on the hours the battery is available, so if you are over-reporting it, what happens in reality is that you are taking into account revenue that in reality will never come, right, and so also from a lender perspective, right, Accure made some calculations and found out that if you reduce by 2% the revenue due to this, let's say, over-reported availability, you have a drop on the debt service cover from 1.30 to 1.26. Just to be clear, the debt service cover is the ratio between how much cash the battery makes divided by the loan that is outstanding, right, the loan that it has to pay back. So if a lender normally sized the loan against this debt service ratio and the minimum debt service ratio is 1.20, you can see that a very tiny change on the revenue basically eats 40% of the cushion that you have on that loan, right, so this is like quite a lot, so on top of that, normally availability is reported by the supplier-owned system by its own definitions, right, so it's clearly a biased measurement, and so normally you should ask as an investor who is measuring availability and from which data, and you should potentially ask also to see the raw numbers. Now, let's move to Britain. I was very happy to be here today for this talk because Britain has the longest operating revenue record in Europe, and in a lot of markets where I normally work in, we have one year, maybe two at most, so that's great. And here we see Montel's data, a breakdown of this year in particular, the first nine months, and for example, we see February, which was the weakest month, the numbers here are in thousands of pounds per megawatt per year. Most of them are two-hour batteries, not all of them, but quite a lot of them. And February came out at 26,000, I don't know, April 54, May 32, and then you have September, which was the best-performing month, and 91,000 pounds per megawatt per year, so the best-performing month thanks to expensive gas and windy nights. What you can see here is basically that September, the strongest month, is basically 3.5 times, the revenue is 3.5 times the lowest month, right? And it's very interesting from an investor perspective because normally debt has to survive the lowest month, and the upside comes on the highest month, right? And this disparity is normally due to how the system works. You know, as I said, gas prices, how windy and sunny it is. If we had an average month for the whole year and not this breakdown, we wouldn't have been able to spot the lowest and the highest number, right? And I think this is one of the things that I'm going to speak about soon because this kind of evidence is very important when you are investing in batteries. If we do a breakdown of that number for September, so same units, right, we noticed that wholesale contributed with 78,000 pounds, frequency response 14, imbalance 9, reserve 2, and then we have the balancing mechanism, which is minus 13, so basically the batteries paid more to buy energy than what they earn by selling it, right? There is one more line that is not included here, which are the capacity markets, basically because those are negotiated in advance, like normally 4 yen in advance, the T minus 4 auction for the winter to come, right? And it's the only basically line here that you know in advance before building a battery project. But most of the times, it normally pays the battery for less than his real size. It's real size, right? At the last T minus 4 auction, you had 22 gigawatts of batteries that were qualified to bid, but they were counted as if they were 6.9 gigawatts only, right? So basically a 3 to 10 factor. Why is that? Because if you imagine a long, cold steel spell, right? So with no wind, for example, most of these batteries are going to discharge, are going to be empty more or less at the same hour. As I said earlier, there are mostly two-hour batteries, right? So if you get this kind of long steel spell, there's no wind that is going to recharge them, right? And so all of them are going to be, let's say they are going to cover the peak, and then the shortage that is left, nobody can cover it basically among these batteries, right? So even if you bring new batteries in, they are not going to improve the situation, right? And so they are counted as less, which means that the bigger the fleet that you get, the thinner the contracted slice of revenue that you can get. And this is really important from an investor perspective because investors normally like to lend against contracted revenue or against the floor. And so the natural question here is, where is the floor coming from at this point, right? And while in Britain, the floor comes from contracts mainly. You know, we have two examples. One was July last year. The biggest listed storage fund in Great Britain, Gresham House, right, moved focus, moved this focus from, and moved 789 megawatts of tolls to floor agreements, right, together with StarCrafts and Merkle. And they basically said that this, the reason behind this was that this floor was key to better terms on new debt, right? Which is kind of the whole point of my talk, in the sense that the merchant upside is still there, but the debt is sized on the contracted portion of it. And if you look at long duration, I mean, the state is doing the same thing, right? Off-game cap and floor scheme, right? That includes 16 projects, about 7.6 gigawatts, with the duration which is between 8 and 22 hours. And what happens there, it's in the end the floor, right? If the revenue goes below the floor, the scheme tops it up. And if it goes above the cap, the excess goes back to consumers, right, in the end. So, the important part here is that even if we keep the merchant side of things, you need to be able for these assets to underwrite a contract or a floor, right? So, the expected value, normally the merchant side of things, isn't really a financeable value, right? And this is really important because if you have high expected revenue but no floor, you normally are able to raise less capital than if you have a lower expected revenue with a contract attached. That from a trader perspective, maybe it's a bit weird, but from an investor perspective is the whole work, right? And, I mean, you remember at the beginning I mentioned that there is a gap, right, between what a battery earns and what actually gets financed, right? And I said, well, one way to solve that gap we saw up until now is having a contract in place or a floor. There is another way to solve that gap, which is evidence, right? And, I mean, if you look at solar, for example, we have decades of irradiation history, right? With solar, you have P90s, which means you can basically find, like, know in advance, nine years out of ten, how much it will be produced, right? So, that's the P90, and normally the debt takes the P90 into account. With batteries, there's no P90s because, as we said, especially in, not in UK maybe, but also in the UK, you don't have such a long operating history like with solar, right? So, normally the revenue is set by prices, by market rules, and what the optimizers do, right? So, in order to, most of the countries where I work in, even outside the UK, where, you know, the record, the revenue record is a few years, as I was mentioning earlier, all investors would like to get the record that UK has, but you can't really buy history, right? You can build it, and you can find a way to build this history faster or slower. The faster way is that we have already a truth mechanism in the batteries themselves, right? It's the settlement meter at the grid connection, right? And with that settlement meter, we can basically find out what a battery is importing and exporting. It's published every settlement period, right? And with that, if we publish that record from day one, we are basically building that history, and we are turning that record into usable evidence that can get basically financed by investors, right? The important thing out of this slide is, I would say, the financing gap is bigger where the operating history is thinnest, right? And what I'm also trying to do with my own project, Megawatt, is exactly doing this, like getting this truth from the settlement meter, bringing it on a ledger, a public ledger, so every investor can see it, and it builds, like, that history and that evidence that makes investors make the move and put the money in. So, I'm going to leave you just with three questions, which are normally the questions that I ask every time I have to evaluate the revenue of a battery, and I think everybody should ask these questions on the capital side. One is, what does the model assume? You know, we have foresight, availability, cycling, everything that we saw earlier, right? If they aren't written down, the number is not complete, and that's really important from an investor perspective. The second question is, how bad it can get. Normally, investors don't finance a model. They finance, and they look at the worst year possible, right? And normally, that worst year is covered by a contract, hopefully. And the third question is, who can check it, right? I mean, if the only source in the system, as we saw earlier, is the same system that is being judged, of course, it's biased, and so you should expect that number to be discounted. So, I would say, to close it up, I think Great Britain gave Europe the longest operating record possible, right? And now, newer markets are going to learn the same lessons, but faster. And I think those markets that are able to turn that history and that operating history into evidence early are the ones that are going to attract cheaper capital first. Thank you. Excellent, Jacobo. Thank you very much indeed.