Catalyst with Shayle Kann - Climate tech startups need strong techno-economic analysis (TEA)

Episode Date: October 5, 2023

We have a flash sale for Transition-AI: New York through October 9th. Use the code FLASH30 to get 30% off your ticket price to our event on AI + energy. Spots are limited, so don't miss out! This might be our wonkiest topic yet: Techno-economic analysis, or TEA.  Before a startup has proven that its technology is commercially viable, it models how its technology would work. These TEAs include things like assumptions about inputs, prices, and market landscape. They help investors and entrepreneurs answer the question, will this technology compete? TEAs are important to the success of an early-stage climate-tech company. And a lot of startups get them wrong. As an investor at Energy Impact Partners (EIP), Shayle and his team see a lot of TEAs—and have some pet peeves. What can startups do to improve their TEAs? In this episode, Shayle talks to his colleagues Dr. Greg Thiel, EIP’s director of technology, and Dr. Melissa Ball, EIP’s associate director of technology. They cover topics like: Bad assumptions about things like levelized cost of production  Focusing on a component instead of a system Focusing on unhelpful metrics Using false precision—something Shayle calls “modeling theater” Recommended Resources: Activate: Techonomics: Establishing best practices in early stage technology modeling Department of Energy: Techno-economic, Energy, & Carbon Heuristic Tool for Early-Stage Technologies (TECHTEST) Tool National Renewable Energy Laboratory: Techno-Economic Analysis Sign up for Latitude Media’s Frontier Forum on January 29, featuring Crux CEO Alfred Johnson, who will break down the budding market for clean energy tax credits. We’ll dissect current transactions and pricing, compare buyer and seller expectations, and look at where the market is headed in 2024.

I have two words for you that you might like to hear. Flash sale. Transition AI New York, our one-day conference and workshop on AI and the energy system is coming up on October 19th in Manhattan, and from now until October 9th, you can get 30% off your ticket. Follow the link at the top of the show notes, and our discount code will be automatically applied to your ticket. And if you want to do it manually, it's Flash 30. And when you get to Transition AI, you're going to network with experts from Microsoft, GE Digital, AES, National Grid, Oracle, and a bunch of founders, executives, and academics who are building AI strategies right now.

The flash sale ends on October 9th. Don't miss out. And if you hear this after October 9th, you can still use the promo code PSPODs 10 at checkout to get your 10% discount. And come dive deep with us into AI and the energy transition at Transition AI, New York. See you, October 19th.

From the studios of PostScript Media, and Canary Media. I'm Shale Khan, and this is Catalyst. So you can spend a lot of time on individual parts of a TEA, which you may end up having to throw out later down the line because the system design changed because you learned something else

that was relevant in another part of the TVA. There's a right and a wrong way to do techno-economic analysis for novel climate technologies. Stay with me. We'll see what's why. When utilities need flexible capacity, they can count on, they turn to Energy Hub. Energy Hub works with more than 170 utilities, coordinating over 2.5 million devices

to manage 3.4 gigawatts of flexibility, built for the moments when utilities can't afford uncertainty. Energy Hub builds and operates virtual power plants that utilities actually stake their grid planning on, coordinating EVs, batteries, thermostats, and more through a single platform built for utility scale. Predictive, verifiable, and designed to perform when it counts. Learn more at Energy Hub,

I'm Shail Khan. I invest in revolutionary climate technologies at energy impact partners. Welcome. So I think most of you know this already, but I lead what we call our frontier fund at EIP. This is a $485 million fund that we launched a couple years ago, and it's dedicated to investing in what we call revolutionary climate technologies. So in this strategy, we're nearly always investing in some form of hard technology, and we're usually investing in. We're usually investing in. before that technology is fully commercial and mature and proven. So when we're evaluating a company for a potential investment, our diligence is more about the team and the market and the technology

than it is about the financial metrics, at least at that point. And when we launched this fund, we knew we needed to do things a little bit differently, given that our strategy was to invest in these frontier technologies, we knew that one of the core components of our approach was going to have to be to build our own internal technical chops and to develop some muscles that we could strengthen to be able to quickly and accurately evaluate the risk-reward trade for hard tech in climate.

So we built a tech team. And that tech team turned out to be far more valuable than I'd even imagined when we decided to build it. And there really are secret sauce in this fund. I'm excited to have this week's guests who comprise our tech team, Dr. Greg Thiel and Dr. Melissa Ball, both from EIP.

and the topic that we talk through is one that is near and dear to their and my heart as well. Basically, nearly every company we evaluate, one of the first jobs is to review that company's techno-economic analysis or TEA. So we've seen literally hundreds of them, and it is no exaggeration to say that TEAs have been the factor that have driven us to get conviction or to lose it many, many times. So it's super important, and we think it is often done poorly, even sometimes by very experienced entrepreneurs. So we have a lot of thoughts about it.

Greg actually has internally developed the nickname Dr. TeA. So this week we're going to talk about it, specifically how to do and how not to do TEA for novel climate technologies. What purpose does it serve? How much precision should we focus on? And what are the major pitfalls that we often see as companies are starting to develop their technologies and figure out where it might fit in the market? So I've been wanting to do this one for a very long time. I'm very excited about it. Here we go.

Greg and Mel, welcome. Hey, Shell. Thanks for having me on. Glad to be here. I can't tell you how excited I am for this conversation that the three of us have regularly anyway, but now we get to have more formally in front of microphones to talk about techno-economic analysis. Okay, so I'm going to start with you, Greg. Dr. TeA, as we call you internally. You get to answer this initial. question, which is, like, I think people probably understand what techno-economic analysis is. But, like, from your perspective, why is it important enough that we should dedicate an hour of conversation now in front of a lot of people to it? Like, what is the importance of it? And what purpose do you think of it as really serving beyond just, like, having a model that

climate tech startups can show investors in the data room? You know, I think it's something that is a useful tool at every stage of technology development. You know, from the get-go when you're kind of mulling around ideas, it's helpful for you to be able to, it's a way for you to be able to say, can this technology that I'm thinking about this idea that I'm mulling over, can it even compete in the marketplace today? And I think as, you know, you start with a back-of-the-envelope analysis and kind of refine it over time and figure out, you know, what numbers, where the sensitivities are, where the limits are and refine your estimates over time, it helps you develop a sort of roadmap to

techno-economic success. So it can help you define targets. If the thing that I'm working on isn't economic today, what does it have to do? What metrics does it have to meet in order to be competitive in the marketplace? And I think by exploring, you know, maybe a level further in terms of sensitivities and limits in the model, it helps you figure out what matters, what design decisions matter to affordability and hitting the customer value prop and which don't. And when you're a small company and you've got limited engineering and scientific resources, it helps you prioritize. Yeah, I think of it in some ways, particularly the early days. Like if you're trying to build some novel technology, if it's the type of thing that we would get excited about,

then almost inherently it requires some degree of magical thinking. like in the early days, you have to believe something can be built that has never been built before by definition, basically. But you have to understand what degree of magical thinking and in what specific way and, you know, what is it going to take to get there. And like, all those things are born out of techno-economic modeling even in the early days. I want to add one thing there. I think we said that we think everyone might know what it is. And I know this is something that comes up here internally, but I think, you know, I certainly knew before EIP what a technical economic model was. Mostly also, as part of the case study, we have to do one to be hired.

But I think it really depends, like a lot of the founders coming from academia. Some of our founders are engineers, and I think they're going to know probably a lot more about what this means and how to do it. My background is chemistry, and I think the chemist in the room and maybe some of the physicists or those disciplines, it might not be honest. obvious. So I think it's one of the reasons, like, this episode in particular, I'm super excited to do it because I think it really highlights to like all of our founders, whether they're engineers or their chemist or whatever their discipline, what it is, why it's important and how they can use it. That is a good point. Okay, so here's what I think we should do. We obviously could spend a long time just like talking about what is TIA and how to do it and all that. But I think the more interesting way to do it is basically for each of us to lay out our pet peeves about things that we've seen. from having looked at hundreds, literally hundreds of TEA models and analyses, on the things that are commonly done wrong. And we should do it as much as possible with the frame of actual examples, right, in climate tech,

and figure out sort of through that vein, like, okay, what is the right way to do it. So, Mel, I'm going to start with you. Name a pet peeve in TEA models. I'm so excited for this. So number one pet peeve for me would be unreasonable assumptions. So I think we probably have a few examples in all of our brains. My number one here to the spirit of an example is this tension between capacity factor and electricity price. And so let's kind of unpack it a bit. So capacity factor, I think most people might know what that is. But in the highest level, it's your actual output divided by your theoretical output.

So if you could have continuous operation. So in power generation, it's your actual megawatt hours divided by your nameplate capacity times by the number of hours in a year. And so we know in some power generation like nuclear, that's going to be really high. And then in some power generations like solar or wind, we're thinking more of a capacity factor, 30% of a really good solar resource, 50% really good wind resource. And so if we unpack energy cost, often what we see in these TEAs, or what I would say, are a levelized cost of energy.

And so, you know, ignoring capacity factor, why I think that's independently not the right energy cost to put in your TEA is that essentially what the end customer is paying is a generation plus a transmission or distribution. Or basically you need to generate that energy and then you need to get it to where you want it to be. And so, you know, I think when you put the two together, this is where my pet peeve is. If you're going to be in climate tech, you want to be green. And so you're almost certainly tethering yourself to a renewable source, which means your capacity factor isn't going to be 100%.

So it's kind of like having your cake and eating it too. Do you want two-cent electricity at 100% capacity factor, I don't know where that exists. Or if it does exist, I think everyone's going to want that gigawatt. And there's going to be extreme competition for that gigawatt. And so that would be my number one pet peeve here. Yes, totally agreed. Just to unpack this one a little bit more, right? So this is a problem for companies for who their technology is using electricity as a primary input, right?

And, you know, what you see often, I think, are people who are at the macro level, you know, trying to draw upon this future trend of declining cost of renewables and saying, because of the declining cost, cost, of renewables, it's going to be economic for me to electrify X, whatever. I'm going to produce chemicals. I'm going to produce fuel. I'm going to make steel. I'm going to, whatever it might be. But in order to do so, what they often do, take the cost of renewables, and you already made this point. Cost versus delivered price of electricity, two different things. But they take the cost of renewables, and then they also assume that they will have that cost 24-7. And those two things are pretty incompatible outside of like hydropower in Quebec. Right. It's, it's a lot. It, it, At best, it really limits your geographic applicability.

At worst, it's totally impossible to achieve. So I totally agree with you on, like, if you're using an electrified process, what you actually should do is figure out real delivered cost of electricity to customers who look like what you will end up being, which is very different if you're like a big industrial facility versus a residential customer or something like that. And then assume those costs. And if you want to take a bet that they're going to decline somewhat over time, like you could take that bet, though that is not the historical trend. But that's what you should be looking at.

As you said, two cent per kilowatt hour electricity at 100% capacity factor is like not a thing that you should be betting on, or at least not a thing you should be relying on for your economics to pencil. I think that's right. I think also in the spirit of what Greg mentioned on the point of what TIA is, like understanding your sensitivities and limits. So what I would say is like the levelized cost of energy doesn't equal what you're going to pay, which doesn't equal what you should necessarily. put in your TEA. And, you know, I think if you want to put that two-cent in, and for that rosier world that we all, you know, want to believe in, I think, you know, on the flip side, what you said is right, have that range where you can see how economic,

if you don't get the rosy assumptions, you know, are you still in the money? Yeah, and it's not just electricity that we see as one of these, like, unreasonable inputs, I think. We also see this oftentimes, even if the inputs are on the, on the molecule side, right? No, totally. I think I'm very passionate about this one being an organic chemist and the idea that organics, i.e., those molecules that are made from hydrogen and carbon, so hydrocarbons, are cheap. And it's a relic of being an organic chemist, which usually these people are coming from that discipline, and we write it in all our papers, and that's like the promise of using organics. But organic molecules, i.e., those made from carbon, and hydrogen are not always cheap. And the reason is, is that there's purification. So again, even that system boundary really matters because your yield matters a lot and your purification matters a lot. And so a good example is Redox Flow batteries. You know, one of the compactive species is an organic molecule. And, you know, everyone pencils in something

that's really, really cheap on a dollar per kilogram basis. And I think that the way that I like to think about it is I, you know, I bound it like ethylene, one of the most ubiquitous organic molecules that there is, is a dollar per kilogram. But you're, and I don't think you're probably going to come close to that. So in that example, it's how cheap does that organic need to be to be competitive? And you have to get really close to ethylene to try and beat LFP or vanadium redox flow batteries in order to be competitive. So it's also one of those system versus system boundary ones as well. Right. Okay, so we've talked about two categories of unreasonable inputs so far, but Greg, I feel like there's some others that I've heard you rail against. Is there anything else that springs to mind for you for unreasonable inputs?

Yeah, there's one that really does spring to mind as a pet peeve as much as I hate to air that out loud. But that one is free waste heat. I thought that's what you were going to say. I was taking a guess in my head. I saw it on your face. Look, I'm a thermal engineer at my core, and am and will always be. And so if I can find a way to use waste heat and make my process more efficient, I'm going to do it. And I think everybody should. When you look at sort of the availability of waste heat out there, there's a lot of it, and it can be tempting to look at that and say, hey, that's all just being wasted.

Why can't we use that and improve our energy efficiency or improve our heat recovery or do something? with it that's useful. And that's really tempting, and I get that. But I think when you start to do more detailed TEAs, what you can see is that the cost of integrating that waste heat may be outweigh its benefits, maybe too big, right? And so, you know, if waste heat is in the form of slow-flowing flu gas that's at kind of moderate temperature,

that can add up to a lot on a big sort of energy system model, When you start looking at a process and you've got to put a big heat exchanger around a long, long pipe, it just might be too expensive to be worth it. That's a good one. What do you guys think about? I mean, another category that I think is an interesting one within this, like the input assumptions that drive your costs, are input assumptions around things that are currently very expensive, but you want to take a bet on them getting cheaper. So maybe the classic example of this would be e-fuels, right, where we're talking about like synthetic jet fuel and that kind of thing.

Your primary input costs into that are hydrogen and CO2. And if you were to produce synthetic jet fuel today at today's hydrogen prices, particularly clean hydrogen prices, which is the point. And if you were to use atmospheric CO2 or biogenic CO2, which you certainly need to do from an emissions perspective, at the end of the day, no matter how good your SAF technology is, that's going to be incredibly expensive. jet fuel. So every single one of these TEAs in that space has a combination of like assumptions around their technology specifically getting higher yield or lower cost or KAPX or whatever it might be, but also assuming some measure of decline in the delivered cost of CO2 and hydrogen. How do you think about that portion of it? And like what what, what's reasonable for those input assumptions and what's not? That's a hard one. And I think, you know,

It's one that varies depending on, as you say, the time frame that you're looking at, the geography that you're looking at, and so forth and so on. So, you know, I think if you're going to needle me here to put a number on it, you know, I would say I think about it more not in terms of what's achievable today, but what you have to do in order to hit competitiveness, right, to go back to the kind of spirit of the TEA and defining targets. and you know that for a fuel, if you want to get anywhere close to economically competitive, you know, subsidies decide you have to have hydrogen that's going to be on the order of a dollar per kilogram and you have to have CO2 that's in that $100 to $200 per ton range and that CO2 has to be CO2 that's coming, as you say, from the atmosphere or from a biogenic source. Otherwise, the fuel won't be truly carbon neutral. I think also on that one going a level deeper in TEA, so what you're

you have to believe to believe, like the hydrogen price will go down. I think that's something that we also try and do. So it's not just we want to believe in that assumption. It's what's driving the price, CAPEX, energy. What do we have to believe in those two components? And then, I think, bound it there and then get comfortable with that in number. All right. So if I can encapsulate this first category then on unreasonable inputs,

it's basically, again, you're going to have to assume something. in terms of your core technology that is going to be challenging in the first place, if you add on the additional layer of that, of whatever input, is it waste heat, is it hydrogen, is it CO2, whatever, is it electricity, whatever it is, if that is an additional layer of magical thinking, it makes it all the more difficult to sort of believe the overall picture.

So try to isolate your magical thinking to the technology leap that you need to take in the first place, because that is within your control much more so than the inputs that you're going to get. Okay, Greg, your turn. Give us a pet peeve. I don't know if I can call this as much a pet peeve is just, I think, a common pitfall that I've probably been guilty of in my own techno-economic analysis journey. And that one is thinking about a component instead of a system. So I think this applies whether you're building a widget or you're making a new process

to make power or a fuel or a chemical or what have you, it's really tempting when you're thinking about what does it cost to make any of those things, to think about the core component itself, the widget bill of materials cost or the core equipment in your chemical process. But in reality, the cost of production of any of those things is more than just the core componentry.

It's more than just the bill of materials cost. And thinking about sort of a chemical or a fuel synthesis type process, the core equipment might only, or the total installed cost of a facility might be two, three, four even more times the cost of the core componentry itself. And so if you end up focusing on just the core componentry, you might miss the actual cost of what it's going to take to do what you want to do. Can you give like a good representative example? Yeah. I mean, I think maybe an easy one there is battery systems, right?

There's a lot of focus, and there should be a lot of technical R&D focus on the cell because that's where all the magic happens. But when you start thinking about deploying those systems, it's more than just a cell that you're going to put on the grid to provide some set of services and grid storage. You're going to be deploying a system that has all those other things beyond cell that make up balance of system and a total system.

Yeah, I mean, and it becomes even more of a problem to think about things that way as these markets get even more mature. Both batteries and solar are good examples of this where like the prices that get reported these days in the case of batteries or sell prices often. And people, in fact, just recently have been like, there's been much noise about battery cell prices getting below $100 per kilowatt hour. That is not representative of the system cost, particularly if we're talking about stationary storage. of the delivered cost, turnkey cost of a battery system. And in solar, because it's a more mature market, it's even more extreme, right? Where you can talk about the cost of the solar module even, not the cell. And in utility scale solar, you know, the module is a minority share of the overall project cost.

And so most of the costs now fall into a combination of the balance of systems and then increasingly not the hardware, right? It's labor and interconnection and permitting and all these other things. And so as a overall share of the total system, that core component, which admittedly is, as you said, where all the magic happens, like becomes less and less important over time. I think as what we see it when we're looking at like early stage companies in new markets or new technologies, to me the version that we see a lot that is challenging is the like, you know, there's a lot of focus on this core component, which is where the special sauce is for whatever the company is trying to build. but they don't have a full appreciation for how big a portion of the overall system cost their thing is. So maybe they're 50% better than state of the art on their core thing, but maybe their core thing is 20% of the overall system costs.

And so in total, the savings at the system level are pretty low. Mel, I know you've, in fact, we've had a few recent examples where you've pointed this out. No, I love that example. And I actually love this category in general. I think it really speaks to me because it seems, you know, if you're in an academic lab, you're trying to often solve a fundamental science problem. But to me, the difference between that and a startup is like there is this where you draw your system boundary. And so your system boundary when you're, you know, a PhD student is really different. It's one small thing that you work on for five plus years.

And then you start a company. And so upstream and downstream of that secret sauce really matters. And to speak to one of your examples, I think we saw this a lot in the modular ammonia space, where the secret sauce is on this low pressure, low temperature, relatively reactor. And that's a really nice to have. But what we found doing our own internal work is that upstream of your ammonia reactor, there's two things you need. You need a hydrogen source and you need a nitrogen source.

And those scaled down well that they can be prohibitively expensive. And so when you think, again, like where you draw that system boundary, if you draw it just around the ammonia CAPEX, you're going to miss it. You're going to miss that nitrogen source and hydrogen source. That is really the bulk of where the level ice costs come from. Yeah. And in that case, it's another example, too, of like, maybe you can build a much better reactor,

but if ultimately the costs are dominated by producing hydrogen and nitrogen, and you don't have any special sauce in that component you plan to use off-the-shelf stuff, then your overall unit economics are driven more by those things than they are by whatever you're building. Exactly. I think there's a few other examples in this category

that are a little bit different to. Sometimes you can end up focusing only on one component, and if you don't look at it in a system's context, you can miss the trades, the decisions you're making about how a component is to be designed or is to be operated can affect the system performance cost, etc. And so, you know,

one of my sort of favorite examples there is thinking about EV drive trains. You know, as I'm sure everyone here knows that the biggest single line item cost in an electric vehicle is the batteries. And a lot of the drive train, you know, power conversion equipment, motors and so forth, are really, really efficient.

And so it could be tempting to kind of, if you're looking at some piece of that drive train, say, hey, this is already 90% efficient plus. What does efficiency matter here? But a few points in efficiency might actually matter a lot at the system level because every kilowatt hour that you waste in the drive train is a kilowatt hour that you have to store in your battery, and that's a really, really expensive kilowatt hour to store.

So again, that's sort of thinking about, puts and takes from a system perspective instead of a component perspective can lead you to a better place. Virtual power plants are becoming a reliable way for utilities to manage capacity, but enrolling devices is just the start. What really matters is confidence, knowing those resources will perform when dispatched and being able to prove it, from the control room to the living room. Energy Hub's platform handles the full picture, from near-real-time forecasting, locational dispatch, and the kind of rigorous verification that holds up when regulators, grid operators, or leadership ask, did it deliver? Easy enrollment creates momentum, proven performance builds trust.

That's why more than 170 utilities rely on Energy Hub to manage over 2.5 million devices delivering 3.4 gigawatts of flexible capacity. See what that looks like at energy hub.com. Okay, so first two that we've covered, unreasonable inputs into the model. The second is thinking only at the component level or the core reactor level or whatever it is rather than the full system level. All right. Let's do another one. Mel, back to you. All right. Another really, I think something we see often is we're going to call apples to oranges comparison. And so what I mean by that is comparing your levelized cost to something like a market selling price.

And so I kind of want to be clear, like the apples to oranges is just apples aren't better than oranges and vice versa. It's just being super clear on what you're comparing to and why that matters is the ultimate goal is to make sure your technology is competitive and understanding the puts and takes there. And so if you're comparing your levelized cost of production to a market selling price, I usually take a step back and like, do you want to make a profit? Because it's really not the same thing. On top of that production, you've got to get the thing you made to where you want it to be. And then on top of that, presumably there's a profit to be had. And so I think that's one that I really always squint at when I see and have to adjust myself. Yeah, and I think that's particularly important because a lot of things in climate tech are ultimately commodity markets, think chemicals, think energy, think, you know, fertilizer, whatever it might be.

and so they're commodity markets commodity prices and those are notoriously difficult to to build startups in because they're volatile prices and all that but also because price is not cost and so if you're saying okay i can produce my thing at factory gate at the same cost that or at 10% lower even than the market price that i've got from some market report then if you're successful and you bring your thing into the market, and let's just say you're selling a 10% cheaper, well, the real salient question is, what is the floor price, which is basically the ongoing cost, the operating cost of the alternative, because otherwise everybody else is going to just drop their price closer to their cost and you get undercut anyway. So it's really, and I think you also made the important point of like, what ultimately matters is the delivered price to a customer.

And they're going to compare two things. Now, maybe you think there's going to be a green premium, and you can make that bet, but you should be clear on that if that is the case. At the end of the day, you're going to have to deliver a thing to a customer, and it's going to have to be better, for some reason, cheaper or otherwise,

than the thing that they otherwise would have been buying. Any good examples? Spring to mind on this one? Yeah, I was going to say, I think also implicit in this is that the distribution cost, or are low. And that's in from some of like whether it's hydrogen or it's ammonia or it's energy, that's certainly not true. And I know, you know, Greg's also been looking at some of this with hydrogen. But with, you know, ammonia, we did a deep dive on what those distribution costs could be.

And, you know, you, depending on where you are in the U.S. or I'll stay U.S. centered, but this is even more so outside the U.S., those distribution and transport costs can be even 2x your your production cost. And so it really matters which, you know, target your, you're comparing yourself to because it's not just 10% off at times. It can be, you know, it can be way off. And then that really impacts your TEA. The other way I think we think about this one. So there's the problem of comparing your cost to the market price. I think that's like fairly straightforward. The other problem is a time horizon one. And this is the one where, or not even just a time horizon one. It's like

truly understanding how cheap the competition could be. And so the classic historical example of this in climate tech is all the thin film solar companies that emerged in the late 2000s, think Cilindra and Mia Salae and all these companies. The value proposition for that suite of technologies was basically we are going to be cheaper than today at that time, today's price of silicon-based solar panels, right? And it turns out that what happened is that the cost, floor of silicon-based solar panels was much, much lower, and it moved much, much faster than anybody expected. So by the time all these thin film companies, with the exception of first solar,

basically, came to market, they were way out of the money because crystal and silicon had fallen much, much cheaper. Greg, I know you've thought a lot about this in today's context as it pertains to batteries, because there's all these new battery chemistries that are being introduced to the market or hope to be introduced to the markets to compete with lithium ion, how do you think about, like, how cheap do they need to be to be tomorrow's lithium ion prices, not today? Yeah, it's a great point because, you know, energy storage, as we all know, is critical to decarbonizing the power grid,

and grid storage systems still aren't as cheap as we would like them to be. But the question that comes up almost every time we see a new grid energy storage technology, be it a new battery chemistry or pumped heat or some sort of compressed gas or variation on compressed air, any of those kind of things is from a total installed cost perspective, can you beat something like lithium iron phosphate batteries, not just today, but in 2035, given that you're probably going to have a substantial development horizon in front of you. And like you say, the thing to beat won't be LFP 10 years ago at that point. It'll be LFP then, right? So I think the numbers that we've landed on in our work are sort of in the $100 to $150 per kilowatt hour total installed system cost.

If you can see a path in a pretty clear path to those numbers with your system, then you probably have a pretty good shot of being competitive with future battery chemistries in the 2030s. As compared to, can you compare that to today's LFP system costs? Yeah, that might be anywhere from half to a third. third of where things are today. So, you know, $200 to $300 per kilowatt hour installed. I mean, the other thing, Greg, that I know I've seen you point out a few times is like when somebody is producing something and they're comparing their levelized cost to what they believe is the sort of right comparison, but it's not thinking about the other technologies

that are coming down the pike. And so it's looking at like a stagnant view of the future that is just today's technologies maybe improving, maybe not, but the reality is that this is a dynamic world. So curious how you think about that. Yeah, totally. Any good benchmarking exercise involves thinking really hard about what actually is state of the art.

And that is, as you say, it's a moving target, and especially in climate tech, where we're seeing so much innovation happening all the time. It can be hard, even in your own field, to keep up with what's going on. And so maybe this is, this is a tried example, but one example that we see a lot comes from the world of hydrogen transportation. For companies that are looking at novel media for storing hydrogen that they might

put on a truck, oftentimes we see benchmarks like steel tube trailers as the mark for cost in moving that hydrogen on a truck. But in fact, there's been a ton of work in recent years and beyond on making really high-strength, lightweight tubes out of composite materials that can store a lot more hydrogen per load than a steel tube trailer look could store, and therefore driving down drastically the cost of transporting hydrogen because you can get more hydrogen per truckload on the trailer. And so since that's such a big lever over the steel tube trailers, if you've got something that's a little bit better than a steel tube trailer, or even

a lot better than the stable tube trail and you're not looking at that composite tube benchmark, you might be given yourself a false sense of how much better you are than where the industry is today. Right. Greg, I feel like one more that we've talked about a lot is when people are building a TEA, like what are the metrics that they're focused on versus what are the metrics that really matter? How do you think about that? Yeah, so the last one, in some ways, kind of relates back to the system versus componentary story, but it's focusing on the wrong metric or maybe solving the wrong problem. I think there's a translational issue specifically that that arises when companies are coming out of R&D heavy environments and they're trying to make a venture-backable

startup. And it was something that Mel, I think, alluded to nicely earlier. I think in R&D, there can be a tendency to focus on core performance metrics, be it deficiency or power density or conversion in a chemical catalysis or chemical reaction sense, something like that. And I'm not here to knock on a focus on any of those. I think they're great goals, and they can move the needle. But from a venture perspective, we're always looking for things that can move the needle in a big way to justify an investment for us. And so if you end up, you know, there are certain systems that you might look at where,

you know, there's been a relentless focus on something like efficiency, but if you go back to the techno-economic model and you think about the sensitivities in terms of energy cost as it contributes to total system production cost or what have you, you might see that it may not move the needle a whole lot. It may not move the needle a whole lot compared to other costs in the system. And so a focus on efficiency just might not be the right prioritization for making your system better and cheaper. Greg, in that example, is that have a function of capacity factor as well?

So in that energy example, so energy consumption versus CAPEX, CapEx could matter more to our earlier point about if you're operating at a reduced capacity factor. Is that kind of what you're touching on? Absolutely. Can you give like a real world example of this one in action? Yeah, I mean, I think one that comes to mind is green methanol synthesis. And this isn't an efficiency story, but it's a performance metric story. There's a ton of work in the literature, and I know Mel, you've been digging into this too,

on making better catalysts for converting CO2 and hydrogen into ethanol. And again, I don't think I'm not trying to knock on that. I think there's room for improvement, and that's good. But from a venture perspective, if you think about that from a process level, if you get a better conversion of CO2 and hydrogen to methanol in a single pass, it just really doesn't move the overall economics in a huge way, because, again, back to a previous point, CO2 and hydrogen are the big, cost drivers in that system. And so, you know, do it great, but it's a hard venture story.

I think this one also is one thing that we see sometimes, too, and we've been talking a lot about chemical synthesis and batteries and stuff like that. And a lot of that has to do with at the end of the day, what you care about is cost for the most part. But in some cases, it's also about what the customer actually cares about and making sure that you're optimizing for that. as opposed to some other metric that isn't as important. So as an example there, maybe, let's just say you're building like robots to do weeding for agriculture or something like that, right? And you could really, really optimize your KAPX on the robot, but that might actually

not matter that much relative to how quickly the robot can move through the field because that's what the farmer ultimately cares about in how it fits in with their operations. We see this in like mining where there's lots of new technologies to extract minerals in new ways. And sometimes I think we see companies that like focus a ton on, I don't know, one metric like maximum extraction. Can you get 99% of the mineral liberated? And that's great, but it's only great if that system also fits in with everything else that matters to the mining operation. So for example, if you have really bad kinetics and it takes, you know, doing leaching or something like that, it takes years to get that mineral out, then you have an

existing mining operation that can't operate its downstream capacity fast enough, and it's never going to work for them anyway. So, like, to me, it's sometimes this one focusing on the wrong problem or solving the wrong problem, it's about like cost, ultimately, but other times it's about delivering what matters to your customer. I love that. It seems like I might even like put it a little bit different words, it seems that it's not like solving the wrong metric. It's in order to be successful, you need to, there's a combination of a couple of metrics that will really matter. So the optimization of those metrics is the value that you're trying to deliver to that in customer as well. I think this also speaks just to the power of techno economic analysis and the

power of, as much as it sounds like a cliche, of, you know, doing this kind of analysis in, in teams that have really strong commercial and technical components, because techno-economic analysis is about technology and economics and the interplay between those two things. And if you don't have a sense of the customer value proposition, you can end up, as we say, optimizing for the wrong thing. So doing good work to understand exactly what's driving customer interest, what's driving customer value is a key part of doing good techno-economic analysis. It's not just cost in engineering modeling. I guess I'll add one more myself, just to wrap up, which is, in some ways, it kind of runs

counter to the rest of the examples that we've described, because basically everything else that we've described is like, okay, here's how to be, here's how to not put enough thought into components of the model or the inputs or what you actually are optimizing for. So the implication, I think collectively of all the other things that we've talked about is like you should dedicate a lot of time and effort to your TEA in order to really understand what business you're building. And I think that is true. But I guess the last thing that is occasionally a pet peeve of mine is seeing TEEA models

with false precision, right? Like sometimes you'll see a seed stage company with one of these models that's like got four decimal places at the end of every number in it. And realistically, there's a lot. you can't know, particularly at the early stages. And so there's this push-pull dynamic that I'm curious to get both of your perspectives on in terms of, yes, there's a lot you can, a lot of value to be gained from doing this work. But there's also only so much of it that you really can do at a certain stage. And so how do you find that line between what actually adds value and what

is just like modeling theater, basically. I don't know, Mel, do you have a view there? Yeah, I think that especially some of the TAs we've looked at are the founders, they're paying for a TEEA. So some people are using consultants, which I would assume is going to be very expensive. So back to the top level, like, it's supposed to be a tool to help drive your technological progress. And so if you're paying good money for this, I'm actually curious y'all's thoughts on that too. I think there's like over specification. My worry with that, and I think we saw, we've seen this recently, is that you can essentially miss the forest from the trees. So if you're so busy counting the number, you know, the power in your pumps and the number of little widgets and valves and etc.,

you might miss something that's really crucial that actually is a driver of your economics because you were focusing on so much that you didn't hit like the really, the couple of things at the stage that you're at, that's going to allow you to get to the next milestone. And so, I mean, I'm sensitive to it. I think, you know, ultimately when we receive a TEA, I think Greg and I probably always look at it, of course, and I think we independently are making our own, so we can teach ourselves what is the drivers in that in technology

and what should be important at the stage that company is at. Yeah. You know, I think that's a great point, and I would maybe add a couple of things. One, I think TEA for me is something that should be very much thought of as a living document in the sense that, you know, it's never too early to start, and you're always going to refine it as you go. So I think it's also really challenging because sometimes the design at an early stage is still very much in flux. And so you can spend a lot of time on individual parts of a TEA or individual parts of a system design, which you may end up having to. throw out later down the line because the system design changed because you learned something else that was relevant in another part of the TVA.

So I think it's better or it can be helpful just to put error bars and understand the sensitivities on things earlier on and go down the deeper, more detailed design rabbit holes when the higher level stuff is fixed. Yeah, I guess for me, if I could boil down, like what is the, what is really the purpose of the DEA for early stage deep tech companies in climate. I think it's three things. One is like how hard do I have to squint? How much magical thinking does it require for me to reach the promised land?

Whatever my version of the promised land is. I'm trying to be 10x better at something than everybody else. Like how hard is it to believe that? Two, what, as you said, what are the major levers? What are the sensitivities, right? What swings my success or failure the most so that I know what I do need to focus on and can spend less time on the things that I don't? And then third is what is the critical path, right?

Like from where I am today to where I need to be, what are the things that I would need to prove or disprove to reach the next stage in that journey? And that sets you on a path that is valuable. That also is, you know, having real. One of the things that I've observed, I'm curious if this is, I'm curious if this is true for you guys. as well. The best companies that I've invested in have a really clear view of critical path. This is the next thing that is in front of us, that we have to, this is the hurdle we need to jump over to prove the next thing in our progression of our technology. And it's not only a

TEA thing, but the TEA can really help you figure that out because you can figure out where those sensitivities are, where you are today, where the biggest delta is, and that'll tell where your critical path needs to be. So if you use DEA to say, how hard do I need to squint? What are the big sensitivities and what's my critical path? I feel like you've done your job.

If you're using it to get to a ridiculous level of precision on the cost structure that you expect to achieve in five years, you've probably wasted some time. 100%. Precious time that can be on doing the technical work. All right. Well, this was a lot of,

fun for me as a fellow TEA enthusiast along with the two of you, we obviously have a lot of thoughts on this topic. But also, I mean, I do think that this is it's underappreciated how valuable this exercise can be for early stage companies who are building

something, particularly something physical in the types of spaces that tend to dominate climate tech. So this is, to me, it's a mechanical thing, but it's an important one. So, Mel, Greg, thank you so much for talking through it with me. Great to be with you.

Thanks for having us. Melissa Ball is the Associate Director of Technology with me at EIP. Greg Thiel is our director of technology. This show is a co-production of PostScript Media and Canary Media. You can head over at CanaryMedia.com for links to today's topics. PostScript is supported by Prelude Ventures, a venture capital firm that partners with entrepreneurs to address climate change across a range of sectors,

including advanced energy, food and ag, transportation, and logistics, advanced materials and manufacturing, and advanced computing. This episode was produced by Daniel Waldorf, mixing by Roy Campanella and Sean Marquan, theme song by Sean Marquan. I'm Shail Khan, and this is Catalyst.