Emerging tech

2024 trends, part 2: ESG, carbon certifications, curtailment, and AI

Our second installment with Nat Bullard about the biggest trends in climate tech.

February 15, 2024
Listen to the episode on:

There was so much to talk about in Nat Bullard’s 200-page slide deck on 2024’s biggest decarbonization trends that we broke the conversation into two parts. For the first half of our conversation with Nat, listen here

Nat has worked as an analyst and writer in climate tech for two decades and was BloombergNEF’s chief content officer until 2022.

In this second part of the conversation, Shayle and Nat cover topics like:

  • How ESG has become the new third rail of finance, falling out of the spotlight of corporate reports and the annual Larry letter 
  • The vexing problem of what to do with curtailed power and why we need to design around the intermittency
  • Whether you can have too many carbon certification standards
  • How biodiesel is eating up Europe’s biofuel supply
  • First Solar’s underappreciated success in surviving the decline of U.S. solar manufacturing
  • Plus: Declining hydropower, slowing coal growth, and the rising hype around AI

Recommended resources

  • Nathaniel Bullard: Decarbonization: Stocks and flows, abundance and scarcity, net zero
  • Washington Post: ‘Greenhushing’: Why some companies quietly hide their climate pledges
  • World Economic Forum: Hydropower: How droughts are affecting the world's biggest renewable energy source

Catalyst is supported by Antenna Group. For 25 years, Antenna has partnered with leading clean-economy innovators to build their brands and accelerate business growth. If you’re a startup, investor, enterprise or innovation ecosystem that’s creating positive change, Antenna is ready to power your impact. Visit antennagroup.com to learn more.

Catalyst is brought to you by Atmos Financial. Atmos is revolutionizing finance by leveraging your deposits to exclusively fund decarbonization solutions, like residential solar and electrification. FDIC-insured with market-leading savings rates, cash-back checking, and zero fees. Get an account in minutes at joinatmos.com.

Listen to the episode on:


Nat Bullard: Now you can see the IRA in this slide, you can see the gap up that starts happening in 2022.

Shayle Kann: Last week we made it through half of Nat Bullard's mega deck on trends in decarbonization. This week, the other half.

I'm Shayle Kann, I invest in revolutionary climate technologies at Energy Impact Partners. Welcome. All right, here's part two. If you don't know what I mean by that, you should go back and listen to last week's episode, but if you do, this is the second half of my conversation with Nat Bullard.

All right, we're back. Let's start with category number one for today, category number three overall, which I decided to call the new third rail, which is ESG. And you have a bunch of interesting data on ESG in here and we also haven't talked about it a whole lot on this podcast. It's clearly become this hot button political issue in the United States. And so the interesting question is, yes, it's clearly a political issue, but what is actually happening in terms of company ESG reporting and ESG fund flows and all these things that had grown up as ESG the concept had gained prominence. So let's start with ESG reporting from corporates, how has that changed?

Nat Bullard: So I want to start by saying that when I set out to do this, I did not anticipate having one slide out of every eight be about sustainability, ESG, carbon markets reporting, things like that. This is a case of letting the data and letting the market lead you in a kind of constructive way was really interesting that this is where I devoted a lot of my time and attention to pulling information because it is so hot button, it's such a third rail as you rightly say.

But let's dig in a little bit. So US company ESG Reports, this is some great research that the Harvard Law School Forum on Corporate Governance had put together. They did something that I find very useful and very kind of empathetic on behalf of practitioners, which is, let's just look at reports in a very blunt way. How big are they? How much information is included, in what ways? When do they get published?

So this was a fun slide to put together, the average number of pages for a sustainability report in 2021 was about 70 pages. It's gone up to 80 pages, 82 pages in 2023. But those report lengths range from 11 to 262 pages. I footnoted that, but kind of worth noting. That's not great. What's also not great is that the resolution, the data quality or the readability of this stuff has declined and this is frankly a terrible outcome. So the number of reports that had an ESG data table, to my mind, the minimum good grace you could give an analyst reading has gone from 76% to 54%. The number of reports that had a press release has gone down from about 76% to 50% and now only 20% of these reports have even a summary or highlights version.

It does raise the question for me, what purpose are these reports serving if they are not conveying this very basic information? If this was the equivalent of your annual report and you didn't have a summary data table in it would be ridiculous. The market would be absolutely up in arms. Another thing, more of a neutral statement, because I don't really have a comp to the rest of things, but the data publication is really interesting, that it clusters much later in the year than say an annual report does from any kind of listed public company. So like 28% of these things are published in April and more like 75% of them are published between April and May and June as opposed to being published in January or February.

Shayle Kann: I mean the story you could imagine telling about what this data is suggesting is that companies feel obligated to continue to produce the ESG reports, however they have less interest in crowing about them, hence the press releases declining from 76% to 50%. They're not really that interested in making them more robust, hence the data table inclusions going long. And they'd kind of like to just slip them in there for people who really need them rather than making them prominent. That's the story the data is suggesting here and that probably makes sense given the politicization of the issue.

Nat Bullard: I could file this away again, under markets respond to incentives, but it's also I think in the long run maybe a Pyrrhic victory to sort of obfuscate your data by not putting it into a table, because for one thing people will get it and they may be less sympathetic to your position if they had to do a lot of extra work to get there, but also you leave more to the discretion of those who are forced to do the work. You leave more to the discretion of those who have to go through the shoe leather cost of getting information out of your reports.

Shayle Kann: All right, speaking of instead of now just ESG reports, but sustainable fund flows, because this is the other thing that, ESG and sustainable funds, not the same thing but related to each other, but you have a good slide on net fund flows into and out of "sustainable" funds. How has that trended?

Nat Bullard: So this is data from Morningstar, which helpfully publishes this all quarterly, and this is just for the US, but yeah, we had sustainable fund flows in late 2021 in the range of more than $20 billion of inflow a quarter and it's been negative for the last five quarters, which de facto means that for the entire calendar year of 2023, we had negative flows into US sustainable funds. Not massively, I mean it was down at the end of 2022, it was negative 6 billion, it was negative 5 billion at the end of last year on a quarterly basis.

So it doesn't mean that the total assets under management have declined in an absent moot sense, but the flows have. And it's important, if you're managing money, you would like to have positive inflows just as a sign of a growing addressable market for you. So I sort of don't want to overplay it, but it's important to know that this means money has been withdrawn. It's not just that AUM has changed, Assets Under Management has changed because of the varying value of the equities with that, but because people are actively removing money from these instruments.

Shayle Kann: Okay. And then a final data point that I just thought was interesting that's all along the same lines but is very specific, has to do with the "Larry Letter", which is the annual letter written by Larry Fink, the CEO of BlackRock, which a lot of people paid a lot of attention to in ESG and sustainability world because of its focus on those topics for a number of years, but of course you're able to track how that focus has changed.

Nat Bullard: Speaking of shoe leather, there's really no way to do this analysis beyond opening them up and looking through and finding all these word searches and looking at documents. But it is really interesting because in 2020 he mentioned sustainability and climate almost 50 times in the course of what is not a huge long document, this annual letter. He mentioned ESG a lot in 2016 relatively and then again in 2021, a little bit in 2022, but by 2023, ESG never shows up at all.

And by 2022 the conversation is tilting more towards climate than it is towards sustainability, which I find interesting. Climate is probably harder to, I think it's harder to dispute than any particular lesson might be about sustainability. The nature of the letter has changed as well, it's no longer... Last year there was no longer a letter to CEOs, there was only a letter published to investors, which is a much bigger publication. But it's another important little marker. People respond to the– people respond to market sentiment and if you're going to be hassled by this at a political level, then you have every reason to not devote a lot of talking point time in your annual letter to the notion of sustainability or climate or of ESG.

Shayle Kann: Now of course the politicization bit about ESG, this is one thing I haven't quite been able to pin down, it's clearly a United States phenomenon much more so than it is a global phenomenon and there seems to be a pretty big difference in terms of whether ESG really is the new third rail in the US versus Europe, for example. And so we don't have great data on the politicization piece, but I did think some data that you had on fund flows comparing region to region with sort of interesting and you can contrast the United States to Europe, if you look at slide 91 where the story is a little bit different.

Nat Bullard: That's right. So fund flows in Europe were positive in the fourth quarter-

Shayle Kann: Sustainable fund flows.

Nat Bullard: Sorry, sustainable fund flows were positive in Europe in the fourth quarter, not much, like $3.3 billion. There are other markets like Japan and Canada, but it's also important to note that sustainable fund flows in general are pretty rank fenced to a small number of total markets.

But on an asset basis, look, Europe is 84% of the almost $3 trillion of assets under management, sustainable assets under management at the end of last year, and the US is 11%. The US is 9% of the almost 7,500 individual sustainable funds in the world and Europe is 73% of them. So these arguments are very loud and they're very present for those of us in the US capital markets and in particular in US political markets. But globally, Europe by far dominates and follows its own path that's very different from what happens in the United States. So just, not a word of warning, but a bit of context I think is important to note for anybody else in the world that what we're talking about here is mostly in the US and also the US is a much smaller share of the global market than Europe is.

Shayle Kann: All right, let's do away with the new third rail category and move on to talking a little bit about power, and in particular talking about one thing that I've been learning more about recently, but I have not heard a lot of folks talking about, which is what appears to be a secular trend downward in the capacity factor of hydropower, which is very important to certain markets in the world. So slide 102 has data across multiple countries on this. Talk about what's been happening with hydropower and why it's important.

Nat Bullard: So hydropower, it's the backbone of zero carbon power in many economies and/it is the backbone of power in general in a number of different economies. In particular, I'm thinking here of Brazil. And it's not great when Brazil's hydro capacity factor went from almost 60% in the early 2010s to below 40% in 2021. It shows that there's this high degree of exposure to climate in general as a long run trend, but also to weather more specifically, intra-year patterns on precipitation and on drought that really impact the availability of this very important resource from a decarbonization and just power in general perspective. And this is not just in Brazil, Canada's gone from about a 60% capacity factor to a little over 50. The EU has gone from mid-30s to mid-20s capacity factor for hydropower. And worth noting that when you're down to mid-20s capacity factor, that's-

Shayle Kann: You’re solar, baby.

Nat Bullard: Oh yeah, your solar in Chile, or your wind in Europe, you're not any different. Now you may be easier to time and dispatch, but in terms of total delivery over the course of the year, you are not delivering more output per nameplate capacity than wind power is. It's firmer, but it's not volumetrically that much different, if different at all.

Shayle Kann: It's also variable year to year in a way that I think people don't always appreciate. We're investors in form energy for example, which is a hundred-hour battery and it's a capacity resource and why does that have value, for example in a market like the Pacific Northwest in the United States, which has a lot of hydropower? Because you can't always count on that hydropower year to year and it may be declining over the longer arc of history. So if you're trying to decarbonize those power sectors, the one resource that you've got that has been driving your cheap abundant clean power for a very long time, it's not so certain as you might think it is.

Nat Bullard: And it becomes the sort of flection point for the dispatch of everything else, even on a global level. If hydro's down, if global hydro capacity factors are down 3 or 5% in a year, something has to make that up. I did not put it in this deck because it's from two years ago, but the droughts in China two summers ago were very bad, impacted hydro, drove temperatures up, there were also very, very high temperatures as well, meaning cooling load went way up, and the instantaneous year-on-year gap up in peak demand in China from the thermal fleet was the size of Germany's peak capacity.

Shayle Kann: Well that's a good segue to-

Nat Bullard: Pretty extraordinary.

Shayle Kann: Well that's a good segue to, let's talk about China. Or really, let's talk about coal, which is mostly talking about China, at this point. How much new coal are we adding to the grid at this point globally?

Nat Bullard: Well, okay, so in an absolute sense, a lot. There was about, the first half of the year, about 25 gigawatts of coal-fired power generation capacity added around the world, which is down from more than a hundred gigawatts added in 2015. So it's definitely come down quite a bit, but 25 gigawatts of coal-fired power is a lot and it definitely delivers more energy and has more infinitely more emissions than 25 gigawatts of say solar that gets incorporated into the grid. You highlight that this is a China thing, it is. It's also an India and Southeast Asia thing. And important to note that these are very policy dependent decisions. There are market forces that play in there, but these assets are so big and the investment capital needed for them is so significant and the planning is so long that we can think about these as having different decisions behind them than say residential solar.

Shayle Kann: All right, one more thing in the power category, let's talk about curtailment. This is one of these things that I hope we look back on at some point in the future and think, "Huh, isn't it crazy that at some point back in the early 2020s when we were building a lot of new renewables and we hadn't quite figured out how to manage the grid with increasing penetration of renewables, we were literally just throwing a lot of power into the dumpster, effectively or into the ground and doing nothing with it?" I hope that that ends up being the case because if not the trajectory of curtailment in places that are adding a lot of renewables, so California being our perfect example, a solar specific issue, but let's just talk about California. The amount of curtailment that we are starting to do in California is quite high.

Nat Bullard: It's very high. So it's always in the spring. It's at a time when solar resource in particular is starting to ramp up and when grid demand itself is relatively subdued and the classic cases where you get the California duck curve of net generation going down sometimes negative is on say a very sunny, windy, cool afternoon in the spring on the weekend in California. But look, it's yeah, not great that we curtailed more than 600 gigawatt hours of solar in, I think it was in April of 2023, up from in 2020 half that much. This is not a great use of things.

It is a little bit more subtle though in the sense that if the economics don't support doing anything to non curtail, and [if they're not doing anything] to use that energy, then it won't be used. If transmission doesn't get built, obviously it won't be transmitted anywhere. If nothing emerges to soak that up, then this is the economic decision. And batteries, as you know in California are incentivized very specifically around essentially a four-hour economic window. So it makes it hard to have the case right now to do anything with stored energy to take advantage of that right now. Maybe you could, we sort of joke about this, I think we probably didn't ask you that you could do cryptocurrency mining with all of this. And we have in the past in a non-snarky way thought a lot about applications that might be available for this, that might absorb this as available electricity as it comes on. But there's many ways to solve it and I hope that we do.

Shayle Kann: Yeah, I mean the white whale for me, which I continue to look for, and if any listeners have an answered to this, I look forward to them telling me about it. So everybody sort of knows and talks a lot about we have super cheap power some of the time, and in this case we have zero or negative price power some of the time, it's not a whole lot of the time, but it is some of the time. And so if you really want to be a friendly resource to the grid and you want to take advantage of that, you want to do something clean, you should do something that is worthwhile, so I'm cutting out the cryptocurrency mining from this description, giving my own bias away, but set that aside, you do something worthwhile where you can afford a very low capacity factor, which means that you need to be extremely low CapEx relative to OpEx and OpEx should mostly be electricity.

In other words, most of the cost of the thing that you want to produce should be born in the cost of electricity. And if it is a good enough ratio between those two and you can ramp, then you should be able to afford to operate only a few hours of the day. Now there's lots of folks who want to do this, and there's some interesting applications. You can throw all your excess energy into a thermal battery and then use it for industrial heat. You can ideally get a really cheap electrolyzer and operate it at low capacity factor, turn it into hydrogen. But I just do not see enough of this happening yet. And it's such a big trend that you would think there would be more innovation there. Can you redesign an aluminum smelter to make the economics work for this, or something?

Nat Bullard: And this is what's really tricky is that I think people under clock how much redesigning you need to do. It's-

Shayle Kann: Yeah, it's hard, which is why-

Nat Bullard: It's very hard. It's not just like, "Well, we'll redesign our smelter to run 8% of the time." No, that's-

Shayle Kann: It fundamentally won't work with the existing design.

Nat Bullard: It will not work. It would probably be just righteously inefficient to do that. And so it requires I think not just re-imagining existing processes, but somebody with a real blank sheet of paper being like, "I have to work backwards, or we have to work upwards rather from this precondition. I'm going to get all the power I want for nothing 8% of the time, but I'm also going to get it 150% of the time that I wanted it in April."

Shayle Kann: Right. Yes, that's right.

Nat Bullard: So how am I going to build, or rather, not how am I going to build something I know today around that, what am I going to build with that as my operating condition? What am I going to build knowing that and not being able to change it? How do I take that and use that as the beginning to build something? And it's hard to do, but I think one of the reasons that I put this in there every year is to sort of emphasize that this is something that exists. And while you can view this in the classical engineering grid operating perspective as kind of a disaster or at least a very suboptimal outcome, I hope, and on behalf of you as an investor, I hope too that there are people who view this as the signal, the substrate on which to build something very new and I think that that's going to be pretty cool.

Shayle Kann: I agree with that. And there are, obviously. A lot of, I think what we've seen so far, like, what do you do about this super cheap energy, is the right kind of battery. I mentioned a thermal battery, there are electrochemical batteries, let's store it rather than dump it and we'll see a lot of that, I think. But I remain interested in the, can you make something with it directly, as well.

Let's dispense with power and move on to my favorite category, which I call random additions, just things that I think were interesting but couldn't quite fit into another category. Starting with talking about carbon certifications, this is a soapbox that I've been on. Before on this podcast, not that long ago, so people know that I find the state of the voluntary carbon market annoying and frustrating and I want it to be fixed, and you have some really good data in there that is indicative to me of why this is a broken market. Can you talk about what's happened with different certification standards for carbon credits over the years?

Nat Bullard: I call this, the everything happens so much series of slides. There are certain areas in which there is a great deal of abundance of activity with the implicit question as to how much value we derive from it. In 1996 we had one carbon certification standard, in 2023, we have 37 of them, and the rate of increase in those has been increasing. In 2011, when the number was much smaller than it is today, we had six standards added and there were 16 in total at the end of the year. We also added six of them last year. And I have to wonder, a value neutral statement, how constructive it is in the long run to have this many different standards for what should fundamentally be something that coalesces I think around a smaller number of more robust standards. This is a question as old as the markets themselves, which is, how much additional value do we derive from greater resolution in one vector versus from greater applicability across a number of them?

But I don't think that in the long run markets are asking for more standards, they're asking for better application of a number of standards. This makes work very, very complex if I'm in this business. How do I pick between these 37 different standards? And the challenge is that they, and it's a healthy one, is that no one is designing them to just further complicate people's certification efforts. They're adding them because they do one thing better. They serve this market better, that technology better, they serve this particular end user in a better way, but there has to be some kind of limit to how many of these things we can crank out. I don't think the market will be better served if this keeps going and we have 50 of them by 2025.

Shayle Kann: I will offer no additional comment, but the numbers speak for themselves. 37 different nonprofit carbon certification standards, enough said. Okay, the next one is a slide that I like to call First Solar is Killing It. Now I want to talk about this one for a minute. So we're investors in Electric Hydrogen, which is an electrolyzer company which is founded by two former successive CTOs of First Solar, and so as I've described the investment thesis, particularly in the early days when it was more of a bet on a team than it was really a bet on what they're actually doing now, and I've had to tell this story a lot, but I do not think people appreciate how impressive First Solar is.

Which is, okay, so you and I both spent a lot of time in solar back in the late 2000s, and what happened during that time was literally over a hundred new companies were founded or were scaling to do thin-film solar, different versions of thin-film solar. There was CIGS, C-I-G-S, there was amorphous silicon and there was all these different technologies including Cadmium Telluride, which is what First Solar was producing, but there were hundreds of them.

Of all of those hundreds, one of them survived. In fact, there is only one western thin-film solar manufacturer still today and very few western based solar manufacturers, period. And so the only one that has made it through that gauntlet for this long is First Solar. And it is not like First Solar is on its last legs. In fact, if anything First Solar has been going from strength to strength in part thanks to the IRA. So First Solar is a story people I think do not sufficiently appreciate. Within the solar universe, it is a differentiated technology. As you said, we sort of think that one technology dominates solar and that's crystal and silicon. It does generally, except First Solar, which is still making Cadmium Telluride panels. And now we'll get to the actual data in your slide, is absolutely killing it.

Nat Bullard: In 2020 First Solar's quarterly order book was about 10 gigawatts and its most recently quarterly order book was 80 gigawatts. Now you can see the IRA in this slide, you can see the gap up that starts happening in 2022 and then how crazily that goes from in late 2022, less than 25 gigawatts to 80 gigawatts a year later.

It's a fascinating look, not just that how this technology has survived and thrived, but also it's a great lens on the US market because it's a US manufacturer, it is probably uniquely suited to taking advantage of all of the mechanisms within the IRA and therefore it's a great kind of forward look on what people think about the US market. And for reference, yeah, North America, read mostly the United States, is almost 80% of the new potential booking opportunities that First Solar sees, not just of its book, but of what it sees happening in the future. Yeah, it is really fascinating. First Solar has done it, immensely to its credit, and its own product, which I haven't looked at in deep detail, has improved immensely over time too compared to the products that we used to analyze as solar hands 15 years ago.

Shayle Kann: Well, yeah. So the things that I find interesting about the First Solar story, in addition to what I already said, they have a different technology from the rest of the industry. And so you look at what everybody else... When people talk about why solar module cost reductions have come so rapidly, how have solar modules gotten to be so cheap? So much of it is industry scale. The industry scaled up and the supply chain scaled up and this is why Asia has been dominant historically and all of that. First Solar benefits from none of that. They have a completely different process with completely different inputs and they're alone on an island with their own technology and yet they've been able to compete. And second, in addition to that, nobody else has replicated it. You think about are there technology moats, there's presumably a technology moat in cadmium Telluride. There at one point were a number of cad-tell producers, but now not really. It is really still just First Solar making those panels. So it's this exception to every rule we think of in solar.

Nat Bullard: That's right, that's right. I agree with that completely.

Shayle Kann: All right, next slide that I just found interesting, biodiesel, in Europe in particular. We haven't really talked about biodiesel in Europe. It's a bigger market than I think many people in the United States appreciate and recognize. And it's interesting where it is coming from and what's going on basically with biofuels in Europe. So give me the high level stats there.

Nat Bullard: There's good data on Europe's animal fat supply, which yuck, when you just think about it in the abstract, it doesn't sound particularly appetizing or particularly amusing, but it's extremely important when we think about what this animal fat supply is as a feedstock and where does it go. Historically, this went to pet feed, it went into oleochemistry, some of it went into heating oil, if you go all the way back to the mid-2000s rather. But in the most recent print of data here from transport and environment in the UK, almost half of Europe's animal fat supply is not going towards its biodiesel. And that's really significant because it's going to start crowding out these other uses for it. And if biodiesel appetite is going to increase, then you'll have to start finding other sources of it. And that's a big open question. Are you going to start importing animal fats from somewhere else? Are you going to find other feedstocks that you use for your biodiesel?

And then another way of thinking about it is, what might cause that demand to grow? Let's say that sustainable aviation fuel becomes a place where you need to start having a lot of these feedstocks, well, this is biodiesel that's going into road vehicles for the most part. What happens when you've got sustainable aviation fuel demand coming into this market? Are we going to have a run? Are we going to have bid ups in the prices? Are we going to have new supply come online? I think it's useful to look at this because it's a microcosm, it's a relatively small market, but it's an important one when we think about how resources are going to compete in the future.

This is a case where, for me, I'm always trying to interrogate my priors over years and years of doing this. Generally, we had a food versus fuel debate years and years ago, are we going to use the land to grow corn to make gasoline, or are we going to grow that corn and make either high fructose corn syrup with it or some other kind of food? Or are we going to, now you can add in, are we going to use that land for energy production of some other kind? Are we going to use that for making hydrogen, are we going to use that for making solar and then making hydrogen, whatever it might be.

But I think it's worth us starting to think about, where are we going to source material? Where are we going to source feedstocks in a world where people are rightly trying to do all these new things, they're trying to push into new markets. So yeah, it's a little bit of an edge case, but a nice illustration of just how quickly some of these things, these little areas that we look at in the sort of decarbonization realm start gapping up into other larger, wider fields that have their own established supply and demand balances.

Shayle Kann: All right, a very final topic in my random additions catalog and a final topic for us to talk about this year, it would be ridiculous for us to have this long of a conversation, talk about this many technology trends and not talk about AI, so we got to talk about AI. Let's start with just AI overall. I mean, we know it's hyped, but there's some fun data that you've got in here on slide 180 on how much executives are mentioning AI in earnings statements and in company presentations. So how has that gone?

Nat Bullard: Yeah, I mean it's like tens of thousands of mentions per quarter now up from a few thousand mentions maybe five, six, seven years ago. It feels like it's something where every company needs to have a strategy or at least has to be thinking about it. Now, you could be a bit dismissive of that and say that it's the same way that every company needed to have a metaverse strategy two years ago. But more importantly, I think you start to see a lot of discussions wherein there are questions of A, how is this a general purpose business technology? It's going to enter into the way that general business gets done, whether you are a power generator or a retailer or an agribusiness.

And then there's I think the deeper, more profound set of questions, which is, how is this going to not just inform, but transform businesses over the long run? How are you going to build new things that only exist because of AI as opposed to just optimizing processes and things like that? But also, what I think is very relevant for you and me and for most of the people listening to this is that this is going to have impacts in the built environment and in energy and in water and things like that because it is a very electricity intensive process, AI, in a way that, actually, I think we've had a bit of a holiday from thinking about this in even the hyperscale data centers in the last couple of years.

Shayle Kann: Yeah. And so that's the final bit of data we should talk through, which is, it has been true historically that we have grown our compute capacity globally a lot, our data center capacity effectively, without growing energy consumption from them in aggregate a whole lot because they've been becoming more and more efficient. And the term that is used is Power Use Efficiency, PUE. So like you said, we've been on a holiday there. I think all indications at the moment are that that holiday is over both because of just the amount of new capacity that is expected, but also the fact that those efficiency improvements appear to have plateaued, right?

Nat Bullard: Yeah. So PUE is now pretty much flat when you adjusted for seasons since 2021. This is data from Google's data centers and they're about as transparent as you can get with this figure. And to be honest, look, there has to be some diminutives here, you cannot will away all of the energy use in your facilities. There are things besides compute that you do in buildings. Even if you want to run a building at 40 degrees centigrade inside, you still have to do air handling, you still have to pump water, you have to use energy for things other than compute itself.

So that just can't... There has to be a floor on it. I don't know, maybe if you were running things at Lagrangian 0.1 in deep space cooled at absolute zero and fueled by solar, you would have a one PUE, but you can't really do that here on earth, and so you're going to reach this floor. And so, funny, in a way, we're almost all the way back to where we started with the per capita or specific data versus growing basis. We've reached it probably, or very close to this particular way of measuring efficiency in data centers, but they're just going to grow so much that demand growth is going to come up. You can see it localized into places where particular grids are like, "Okay, I need to now incorporate a gigawatt worth of capacity." Or, "I need to allocate that much-"

Shayle Kann: Gigawatts?

Nat Bullard: Yeah.

Shayle Kann: Oh my God, a gigawatt is nothing. If you're in Northern Virginia right now, a gigawatt would be-

Nat Bullard: Oh, I was saying on a unit basis. Yeah.

Shayle Kann: On a unit basis?

Nat Bullard: Yeah. Yeah, yeah.

Shayle Kann: Well, sure, even so, gigawatts.

Nat Bullard: Exactly, plural. And so yeah, I'm in Northern Virginia and I'm like, "I have to plan..." I have to plan for load growth and a specific kind of load growth. I'm not planning for subdivision cul-de-sac load growth here, I'm planning for industrial load growth of the sort that grids haven't known since maybe they were building greenfield smelting capacities or things like that. And it's a really big challenge. Fortunately, it's happening with a good deal of sunlight on it. I think a lot of players involved they're being clear about it and you can't hide it that this is happening. And so I'm very curious to see one, where this trend evolves, two, what comes along to change it. The answer to AI data consumption growth or energy consumption growth is probably AI, finding other applications for it, and we see these from some of the providers.

I would also add that making sure that there's a lot of utility in decarbonization for your AI applications. You can see that with some of the work that's being done, not just around data centers themselves, but some of the work that the big AI companies have done in looking at say, airline emissions or the forcing effects for global warming from airplane contrails. What happens if you can redirect flights based on AI and computation and analysis to reduce the forcing that comes from contrails? Fascinating stuff like this. I also think about all the applications you could get from new understandings of materials, of crystals, of things like that. New applications in proteins that come from protein folding problems that have been solved with AI. And so hopefully we find all of these things being used to a decarbonization effect to some degree because they are going to be using a lot of energy and more energy in the future.

Shayle Kann: All right, Nat, we've done it. We've gone through 30-ish of your 200 slides. Thank you for letting me pick and choose and for spending some time again.

Nat Bullard: Always a pleasure, Shayle, and I look forward to the next one.

Shayle Kann: Nat Bullard is a longtime climate tech analyst and writer. He is formerly of Bloomberg now doing his own thing, including writing this deck. This show is a production of Latitude Media. You can head over to latitudemedia.com for links to today's topics. Latitude is supported by Prelude Ventures, Prelude backs visionaries accelerating climate innovation that will reshape the global economy for the betterment of people and planet. Learn more at preludeventures.com. This episode was produced by Daniel Waldorf, mixing by Roy Campanella and Sean Marquand. Theme song by Sean Marquand. I'm Shayle Kann, and this is Catalyst.

No items found.
No items found.
No items found.
No items found.
No items found.