What sustainable investing looks like in practice with Shanu Mathew (Lazard Asset Management)

Hello, everybody. Today on Transmission, I'm joined by Shanu Matthew, senior VP and portfolio manager at Lazard Asset Management.

Shanu looks after more than a billion dollars of client capital, and he does it through a sustainable investing lens. He picks companies that aren't just profitable, but are actively shaping a greener, fairer and safer future. We get into what sustainable investing actually means in practice, and it's not just wind turbines and solar stocks. Shanu explains how companies like HVAC manufacturers, waste firms, and even chipmakers are benefiting from the energy transition.

And we also dive deep into what public markets are saying about AI and data centers, the staggering sums that are being spent and the bottlenecks in turbines and transformers, and why the race is all about speed to power now. You'll hear us talk about that quite a bit. If you're in energy finance or just curious about how Wall Street is digesting the energy transition, you'll find this conversation fascinating. And for me personally, I really liked how Shanu brought abstract concepts and numbers and made them real stories.

So thinking about how, decommissioned nuclear sites are gonna be brought back online just to serve data centers, things like that, It is incredible.

So let's get into it.

Hello, Shanu. Welcome to the podcast.

Hey. How's it going?

Well, I mean, not as good as you. I can't believe that you've come straight to so Shanu's with us. He's come straight from ringing the bell at the opening of the stock exchange in New York. Tell us about that.

Yeah. I was lucky enough to witness it. I wasn't ringing it myself, but, I was at a conference this pat these past few days with the energy council, and they were the team was ringing the bell today. And it's a really cool moment to be on the floor and see that and kind of the celebration of markets and, you know, just the it was it was an eye opening and really cool experience, I think.

So they ring it every day. Was there was there a listing today?

Was it or is it just I don't think so.

I I think there was a listing today as as part of the reason the energy council team was doing it. I think if sometimes if there's a brand new listing or a big IPO, like, that team will ring it. Yeah.

I can refer to you.

Yeah. But they still had Kramer and CNBC right there, and, you know, the environment was still pretty captivating regardless.

You know what? I I hope one day we get to ring that bell. I'm I I that will be the day Apparently. Although, apparently, the work starts. As soon as we become a public company, that's when the real work starts. Right?

Yeah. Well, then you have people like me calling on you every quarter.

So Yeah.

Yeah. Okay. Let's get straight into that then. So, what do you do in the public markets, Shanu?

Yeah. Absolutely. So my name is Shanu, and I'm SVP and portfolio manager at Lazard Asset Management. We are a global asset manager that predominantly focuses on listed public equities.

So, basically, on behalf of institutional clients, pensions, endowments, you know, people with large amounts of assets, they will invest with active managers such as ourselves. And, basically, we're picking portfolios that, in theory, will beat the market, and generate a healthy return for our clients. And at the same time, clients have different preferences or styles that they really, like, want to be exposed to. And that's things that, like, my team, or our firm will do in generally whatever flavors you want, whether it's, you know, global, US focus, or geography splits, or style splits, value, growth, quality, and even just the the type of of investing.

So there's fundamental bottom up, which is folks like ourselves that are, you know, we have analysts and we're looking at stocks individually or we have, you know, quantitative strategies or systematic strategies that, you know, apply a little bit more software or machine learning or, you know, things of that nature. So it's basically whatever flavor the clients want. We're manage money on behalf of them, and we're picking stocks at the end of the day.

So you you look after other people's money in funds and then buy stocks in the stock market to get a return for them.

Yes. Correct.

Simple simple terms. And, yeah, what whose money is it?

Yeah. Predominantly, we at least for my portfolio, I'm a portfolio manager on our US sustainable product. And so, basically, we have a specific focus, and our thesis is based that companies that contribute to or are assisting or accelerating the move to a more sustainable society by four pillars, greener, fairer, healthier, safer, will outperform over the long term. And so as you can imagine, you know, a lot of the folks focus on sustainability or those type of strategies are prominently from the European region.

So our clients, most of the institutions we work with are European today. And at a global scale, eighty percent of the capital that goes into, like, ESG slash sustainable investing type products are from Europe. And then increasingly, you're seeing pockets from the Americas and APAC and things of that nature. But largely today, whether it's active or passive, that a lot of that's coming from the European region today.

Sounds striking. So, you said that as if it was, you know, common knowledge and obvious, but I've never heard that before, but reflects the fact that I do a very different job to you. So so so Europeans in general put more money into ESG and sustainability linked assets versus Americans?

Correct. Yeah. Correct. There's definitely, like, regional disparities. I I mean, if you rewind, like, the last four years, there was, like, the growth of ESG globally and things like that.

But in terms of, like, global equity markets, still a relatively small percentage. I mean, like, there's, you know, there's some fluffy numbers there that, like, markets always say, like, a third of the assets are now ESG linked. And if you actually parse through the numbers, it's a lot smaller than that. But at least today, what we've seen is that the European region is generally a little bit more focused on these issues and the frameworks, and we're, you know, pioneering in a lot of ways.

And so I think Americans are still figuring out, you know, the different types of strategies they'd like to do with that and whether they wanna be, like, alpha oriented or, in integrate some of these sustainability issues. I think strategies like ours will appeal to investors all over the world, but just the pattern of client interest so far, it seems to be, you know, largely driven from the European region.

And so you look after a pool of capital, so five, six hundred million dollars, something like that. Right?

It's actually it'll be one point five billion.

One point five billion. Okay. So I'm out I'm out a little bit there. So you look after that money.

You put it into the stock market and get a return for your clients. And so then you have to choose which companies, and you have to do that through the lens of, you you have four very useful pillars there, but, ESG and, generally, climate related and sustainability related activities. So so then how do you choose those companies? Because you still gotta make money.

Right? Your job is to make money, but in in that world so so so start from the beginning. What's the what's the process?

Absolutely.

So the our process is fundamentally built on what we call two tests for any name that goes in the portfolio, and one's very fundamental driven and one's sustainability driven. And just to be clear, like, both these things happen simultaneously, and it's driven by the analysts that cover the same stock. So we don't have a model where there's a separate ESG team or or things of that nature. The analysts that cover the stocks that have had these relationships for twenty years are the same ones doing the sustainability work because we find that they're the most in tune with the actual issues and material opportunities or risks at a sector level for individual stocks.

So, basically, our financial and fundamental test is based on cash flow return on investment. We're basically looking at high quality companies, the top three deciles generally that generate a really strong cash flow return on investment. And then from a sustainability standpoint, we have a proprietary scorecard that we use that measures companies across products products and services as well as their operations, you know, how they treat their employees, supply chain, resource intensity, resource management, things of that nature, and that ultimately leads to a weighted average score, and that ranges from negative five to plus five.

Basically, every company in our portfolio has a positive score. We don't own anything that scores negative on that framework, and then we have certain percentages of our allocation that will go towards the highest end of that scoring framework. So that's generally how we think about the world where, you know, we're looking at things very much from a fundamental standpoint, but we're ultimately trying to also capture that sustainability score and, you know, where we do best or where we find the best ideas is often when we can tie the sustainability opportunity or risk to the fundamental business.

So is it a new product opportunity for them? Is it a new, growth of a TAM, so a total addressable market will allow them to accelerate their margin growth? Will it allow for better returns? And ultimately, if we capture those inflection points before the market recognizes them, that allows us to generate healthy returns for our investors.

And so the ideal situation would be that, sustainability linked stocks would return the best returns. Right?

Is that true?

It's it's a loaded question. Kind of as it depends on how you define that, but I'm I'm glad that you brought that up because sometimes you say ESG or sustainable fund, and everyone assumes that you only invest in solar, wind, and batteries, and and nothing else. But the way that we look at it, right, is we're probably diversified. We're looking at, incumbent businesses.

We're looking at new businesses all the time. And at least in the recent period and, again, what, you know, all the normal caveats around, if you measure, you know, one year, three year returns, that can look a lot different than five, ten, twenty year returns, etcetera. But in recent years, like, the, you know, quote, unquote, pure plays or the most exposed transition related stocks have actually done not that well. I mean, you can look at it from, like, the general solar index or the global clean tech index compared to the S and P, which has been tremendous over the last few years.

There's been a huge divergence in performance, and that's due to a multitude of things, supply demand mismatch after COVID, high interest rates, you know, political switches in the administration, things of that nature. And so those stocks have actually struggled. And, you know, we've been able to sidestep a lot of that because often what we're where we find the best opportunity is incumbent businesses with underappreciated tailwinds from sustainability. So that might be an HVAC company that, you know, regulation changes and it drives it to a more efficient use of refrigerants.

And this company has a, you know, really dominant position in commercial HVAC that might be in a waste company that is increasingly, you know, capping their liabilities with flared methane and actually capturing and selling it back to premium RNG markets. It might be a an advanced chip company that is selling increasingly to automotive and industrial, which is both going through electrification. Right? So you have the better chips, with a higher content per unit in autos or industrial with robots, that leads to, you know, faster acceleration and top line and or margins.

And so, you know, companies that we typically own, or have owned in the past, you know, like like, for example, Trane, Waste Management, Rockwell, all those three examples I just gave to you, are seeing opportunities as a result of the transition. They have really strong incumbent businesses that have been around for decades, and they're also capitalizing on these new opportunities that the secular momentum is allowing.

And the reason why I wanted to start here in this conversation is I think the public perception is that ESG investing is basically buying wind turbine companies and, you know, stuff in green energy and that's about it. But the the ESG surface area is much, much wider, isn't it? And, I I just wanted to touch on that. And then before we go any further, so for our listeners, how long do you hold these stocks for generally? When you take a position in one of those companies you talked about, how long do you hold that for? And I know that, obviously, it's how long is a piece of string, but roughly speaking.

No. It's it's a great question, and I'm really glad you harped on that because, our approach is what we call in the industry long only, Meaning that we hold stocks or we buy and hold stocks, and we wanna own them for the long term. So, typically, when we're underwriting a stock for the portfolio, we generally wanna underwrite it for the next three to five years. We're not in a quarter to quarter business that, like, hedge funds or other asset managers might be, But we're looking to be a long term partner in sort of capital, with long term ownership.

So I mean, some of the names in our portfolio, or even across the firm have been in the portfolios for five, ten, fifteen years. Right? If it's if the thesis remains intact and we have a really favorable view of the stock, we'll continue to own it. We wanna be long term partners.

There's other there's also other stocks, right, that have been a shorter time window and that could be as a result of a few things, right, market appreciation. If a stock doubles, all of a sudden, the risk reward might be skewed unfavorably and we're ultimately always looking for the best risk reward management for our clients. And so others you know? And then also, you you have situations where your thesis breaks.

You might think that a market's a lot bigger or a company's able to fix, you know, their financial productivity a lot sooner than is anticipated. And then when in situations like that, we're ultimately rotating out of names we have less conviction in and then moving toward the ones we have highest conviction in with a good risk reward. So generally, like, multiyear holding period, though.

And I just wanted to ground the conversation because when you hear the word fund, many people think about, you know, screens with lots of, graphs on them and algorithmic trading and Flash Boys and whatever Michael Lewis is gonna write next. But you're closer to the kind of Warren Buffett approach holding and really thinking longer term about these these businesses.

I think it's a fair way to put it.

In my simple simplistic to obviously, it's more complicated than that.

In no way, I'm comparing myself to Warren Buffett. But Yeah.

Yeah. Yeah. Yeah.

Yeah. Yeah.

Go on.

That philosophy works a lot better than the the Yeah. Yeah.

Yeah. Yeah. Yeah. Quick break. If you listen to this show, then you probably work in energy.

And at Modo Energy, we're not just talking about the energy transition. We help our users to actually make it happen. All energy storage, solar, and wind assets on the global balance sheet need to be valued and benchmarked. And that's where Modo Energy comes in.

Our benchmarks and forecasts are transparent, bankable, and trusted by the world's leading banks, asset managers, utilities, and developers. So if you want to learn more, go to modoenergy dot com. And if you want more content like this, sign up to our weekly dispatch newsletter.

Enjoy the conversation.

So now I wanna talk about, you listen in on a lot of earnings calls and all around the public markets, everyone's talking about AI and data centers and infrastructure investments. And so I know you've been spending a lot of time tracking in data center growth and what you're hearing on these calls. Could you just, could you set the scene with how AI and specifically data centers, because we think a lot about energy, how how data centers and data center investment is now turning up everywhere? Bit of a leading question, but I think it's a fair one.

No. Absolutely.

It's all anyone can talk about, to be honest. So, tracking it to an excruciating level of detail, but, ultimately, that's what the market, you know, that that's the common dominant narrative right now. And so if you look over the last few years, pretty much since the release of Chatt GBT, when you look at the top, you know, fifty stocks, in terms of performance, a vast majority of them are levered to AI in some way, whether it's the actual, like, chips, or the hyperscalers that are deploying this in the cloud or all the way down, like, infrastructure owners, like, that own the land or the intellectual equipment and things of that nature. So we actually look at things, relative, like, on on the on the index basis like we've mentioned.

Up to forty percent of the S and P right now is levered to this AI trade. So by virtue of, like, if you're in the public markets and you pay attention to the index, which many of us do, you have to have a view on what's going on with AI. And so, you know, we look up and down the supply chain. Our portfolio has different manifestations of the relative, you know, bets that we would wanna make in terms of favorable risk reward.

But, you know, it goes anywhere from upstream. The hyper scalers are the ones spending the CapEx, and this is like the Googles, Amazon's, Metas, of the world. And it goes down to the chip level, which I'm, you know, sure Nvidia, everyone's well aware of now. It is eight percent weighted in the S and P five hundred.

And you have the different other GPU providers or ASICs, it would be called, which are like the Broadcoms, Marvells, AMDs of the world. You move upstream to the people that provide the cables. There's, like, opticals. There's copper, etcetera.

You go down even further. There's, you know, EPCs that actually build these data centers and do the mechanical, electrical plumbing in there or the outside the wall power, or transmission associated with that facility. And then you have various forms of electrical or cooling equipment. You know, that's, like, the Eaton's, the Vertiv's, the trains, etcetera of the world.

And so there's quite a bit to go through. I'm happy to kind of talk about the main themes, but I'd say, in general, the last earning season, so this most recent two q, the last month, all the signals are pretty bullish, to be perfectly honest, in terms of just the the feedback you're hearing, a lot of pipelines that are being elongated or record levels, really strong order growth, strong visibility. I mean, there's a tech conference going on right now and Coreweave, which offers, you know, is more of like a NeoCloud type of service model to various providers that, wanna use AI. And, you know, they're talking about the words insatiable or relentless in terms of demand.

And so that kind of gives you an idea of the relative strength of this market right now. That doesn't mean they'll go on forever. I wanna be very clear about that. But, ultimately, right now, in terms of the signals we saw up and down the supply chain this most recent quarter, it was pretty strong.

So happy to dive into a little more detail there.

But if we go to first principles, where is this money coming from? As you said, the first thing you said when you answered that question, which I thought was personal, was about CapEx.

It's it's who's spending the CapEx, and they trickles through, right, whether that's optical cables or transformers or cooling systems.

So let's start with the CapEx and that these hyperscalers. Well, actually, what's a hyperscaler?

Yeah. So it's ultimately the the large tech companies that we said that have the ability to, you know, stand up a lot of compute very quickly. And it's ultimately their money that's allowing the development of AI more broadly. So, I mean, it kinda depends on the estimates you're using and the numbers, but, I mean, just call it round figures. We're probably gonna spend somewhere around half a trillion dollars, and the vast majority of that is coming from, you know, those four to five companies.

And so Half a trillion dollars when?

This year. This year in terms of CapEx. Yeah. And if you look at the second quarter earnings period, a lot of them actually increased the amount of CapEx they plan to spend in coming quarters and the next fiscal year.

Half a trillion dollars. This is like twenty nuclear power stations in CapEx.

Yeah. Maybe even more. More. More.

Unless you're in China where you can build them cheaply. Yeah. So you're talking yeah. You're, tens of nuclear power stations in a year in CapEx spent on, data center build out.

Yeah. And that and that figure is, like, fifty percent up year over year. And looking into next year, I mean, you know, there's a lot of large numbers that eventually kicks in, but, again, it's still staggering in terms of I mean, look at the different bank estimates. You know, some of them have that number going to a trillion dollars in coming years. Others have it slowing down, but that just gives you like, it's eye watering. Right?

We're talking about the entire GDPs of nations at this point in terms of the level of of nations being spent in a year on new data centers.

Yeah. And that's split across chips and physical infrastructure. I'd say, like, at least the last few years, it might have been skewed, you know, fifty fifty or maybe slightly towards the actual physical infrastructure because that's longer lead time. But increasingly, it'll be spent on, you know, what we call I IT IT equipment or the the actual chips and the servers and stuff going forward.

And I imagine, similar to what we spend a lot of time looking at in the battery world, that a lot of the infrastructure when you build one of these sites, you can replace the cells, but a lot of the infrastructure might last, you know, many, many decades. And so I assume that data centers are thinking this way. I've heard that the, you know, some of the estimated replacement rate is, like, two and a half years, something like that in some of these, chips.

Yeah. The chips' useful lives, there's a there's a huge debate on it, and I don't think anyone really knows the answer to it unless you have a huge fleet of GPUs running and you can kinda look at it day to day. But, ultimately, yeah, I think and from accounting purposes standpoint, a lot of companies point to four to six years. If you talk to people that actually are deploying these chips, they say it might be closer to two to three years for running them all out. But those will be replaced very, very often. The longer term infrastructure could be in place for ten, twenty years.

Your cables and your transformers, they're gonna be they're gonna be there for a while. And so one thing we're trying to get a head around, like, you know, the whole power industry, which is obviously what we we focus our time, is trying to get ahead around, is what happens with all this load growth and and what that means for the for the power sector. And, so so how do you think about that problem when you're thinking about the broad based effect on the economy?

Yeah. It's a it's a great question. I mean, increasingly, just given the the eye watering sums of money being invested, AI is increasingly tied to the overall, economic picture globally, especially so in the US. I think one of the things that you really introduced that I would love to stress for listeners is that duration mismatch.

Right? In power, we're typically talking about things on twenty, thirty, forty year timelines. And this data center trend, again, ChattyPT came out three years ago. Right?

And so and we're talking about two years on chip chip timelines and things like that. So there is this really big duration mishmash where utilities often plan for twenty, thirty, forty years. And in terms of visibility, I mean, sure, it looks really strong right now, and I just talked about all the optimism around data centers and AI. But who's not to say that that could slow down?

And all of a sudden, if you build a power plant that needs to be in service and then the rate payers are gonna pay for over twenty, thirty years, all of a sudden, if there's no more demand and no more load growth in five years, like, that creates a really big fundamental mismatch and problems. So that's what's happening across the the US and the globe, quite honestly. It's just like, how real is this demand cycle, and then how long will it last? So I don't think anyone's debating the the level and velocity of dollars today, but it's just like, will that continue and how long can it continue?

Because that is, you know, increasingly a very, very high number. Like, even the most profitable companies in the world can only spend so much money. Right? And so it ultimately starts to focus on, like, what is the incremental source of, you know, CapEx dollars?

Like, is are the models gonna get better? Are there different scaling vectors that they can continue to do? Because if so, then we will ultimately see all these numbers flow into load growth. But there's a really wide range.

So, I mean, just to not spend too much time on the numbers, but in terms of the pipeline of activity, like, typically with power, if anyone's been a student of history, the power industry is really poor at forecasting. There's a really good chart chart from S and P that we maybe could link in the, show notes. For the last twenty years, if you look at each individual year of the annual power forecast, it always shows up into the right, and the actual line has been flat. So they're really, really bad forecasting.

And, you know, this time could be different and all these things, and I think there's a fundamental drivers. But historically, we overestimate. And then the more challenging part too is that we don't know really how long duration the cycle could be. So, in terms like wide variance of, you know, estimates, for example, we'll use the Lawrence Berkeley National Lab study.

Data centers today are called three to four percent of overall US electricity, and they forecast by twenty thirty, it can be anywhere from seven to twelve percent.

That's a massive variance in electricity markets. You know, the difference between seven and eight percent is huge and nominal, and we're talking about a range that's literally double, six to twelve on the high end. So that gives you a range of, why it gives, like, utility operators or owners heartburn in terms of, you know, if they get this wrong, that's a lot of infrastructure that needs to be paid for. These are, I mean, plants or you go up plus plants, a couple billion dollars depending on the gen type.

This isn't you know, you can fall asleep and forget about a type of things. Like, this will be, really profound. And then I think it's also coming at a time politically where electricity bill inflation is becoming a, you know, kitchen table issue. And I I wouldn't be surprised if the next few elections have that core to them.

And, you know, if we have a lot of load growth, that means electricity bills are going up too. And so it becomes a question of, like, who's responsible for it? How much can these tech companies or other users pay for this? And, you know, are we gonna feel the impact of it?

Like, you and I obtained our local utility bill.

Yeah. It's a hot potato in in the PGM market at the moment, isn't it?

Because the the democratization of the cost of all these connections is, as you say, I mean, it's just such an easy, easy thing to point to politically as to get angry about Yeah.

Even if, you know, the broader picture is a bit more complicated. I've got so many questions on this, but coming back to the data center build out for a second, could you just you had a really good rule of thumb there. So a couple of billion dollars for a gigawatt data center is roughly what you said. Right?

For the for a power plant, and that is so a gigawatt data center is actually a lot more than that.

So traditional data centers, so, like, CPU based compute, you can call it, for the data center, like, anywhere from eight to twelve million dollars per square foot.

So this is none this is normal cloud stuff, Amazon Web Services. This is what we're all used to. Right? This is not AI stuff, and it's not Bitcoin mining. We'll we'll keep that out of this for now.

Correct.

So so normal cloud, you're saying Eight to twelve million dollars per square foot.

So call it, you know, eight billion to twelve billion on a gigawatt data center. K. Again, data centers typically aren't size that big, but they have to be for AI. Now we're moving into our arena where it's typically people will quote you in the twenty to thirty million dollars per square foot range.

So a massive leap. Right? A gigawatt plant could cost twenty to thirty billion dollars. And, actually, if you kinda do the math on the recent announcements, it's suggesting even higher numbers than that.

And so that's split across a bunch of things. So that number is fully loaded. So, for example, let's just use the twenty just for sake of ease. Twenty billion dollars, ten billions of that is chips and servers, you know, maybe even sixty percent.

And that leaves over, like, thirty percent for cooling, electrical, EPC, labor, land, cables, things like that. But that gives you the idea of the scale. Like, these are massive projects. And so it introduces a question of because it's a whole issue about power and all of a sudden you're hearing about people building behind the meter or their own power stations and you're you're wondering, why would someone wanna spend a few billion dollars on a power plant?

It's a lot of money. But if you're bill if you're spending twenty, thirty billion and get the data center built and all these chips, and, again, they only have a useful life of a few years before they, you know, get deprecated or become the older version, you can understand what's at stake in terms of generating a good return. And that's why it's motivating some of this, you know, speed to power or time to power type of thinking.

It's like another Manhattan project, isn't it? Yeah. It's just enormous and, all private privately done.

Yeah. I mean, there's a there's different pockets of capital now that are contributing, but it's very yeah. But it's private driven. I mean and we're sitting here in New York right now.

Fate Meta, if you look at their project for the the new data center campuses they have going up, they're gonna be bigger than all of Manhattan. I think it was one of the the images they showed just to give you a scale. Like, these are, like, football field sized, huge facilities. They go for miles and miles.

Let's talk about so can you pull out a couple of exam that's a great example. So you're you're on the earnings call. You're studying these hyperscaler companies. So you mentioned one meta there with a project bigger than Manhattan.

You know, what other crazy things are going on in the market?

Yeah. It's it's a good question. I mean, I think it it really is just around this whole theme of how are people bringing this infrastructure faster than the typical timeline. So for the listeners, generally, when you try to request power from utility, that can take anywhere from three to seven years because they need to get the appropriate permitting and siting and and things of that nature.

So, typically, you go through the avenue because you're not in the business of building your own power. Now what we're starting to see in the market is, you know, so you're seeing increasingly seeing companies start to procure power on their own terms. So the big hyperscalers now have teams of tens of hundreds sometimes people procuring energy on their behalf. So we've seen different deals in the market.

I mean, just yesterday, you know, a water utility, had to put a filing out, that, you know, there's a nine hundred forty four behind the meet megawatt behind the meter project going up, which is, you know, a massive project that's gonna go entirely behind the meter with the CCGT as the power source. You're seeing you saw Meta.

Not not a private wire.

So So that that theoretically is going to be private wire because you're not plugged into the u it's gonna be behind the meter.

Yeah. Wow. Yeah. So behind the meter, meaning you're not plugged into a meter that the utility ultimately bills you for.

And so you're starting to see more and more of these announcements. The thing about these is, like, there's a lot of bottlenecks associated with doing so. As you can imagine, like, it's already complex enough to build a massive data center and and or a power plant. And when you're combining all these things together, like, it adds on layers of complexity.

So you still need to get you know, you need to have a turbine slot to ensure that you can actually get your equipment. You need to have the appropriate land. You need to have the ability to, you know, lay lay down your wire, ensure that you have enough space for diesel tanks or or backup generation, things of that nature. And so we'll see how many of these actually get built.

But in terms of announcements, we're seeing that. You're also seeing, other transactions directly with utilities. So Meta, for example, just did a deal with Energy. They're putting they're gonna put up a two gigawatt data center campus, and they signed a fifteen year bilateral deal, meaning that, you know, they're guaranteeing the utility fifteen years of a revenue source.

And that what that does is that helps mitigate the amount of cost inflation that their rate payers will pay. Right? Because Meta is trying to, you know, pay for their commensurate share of what they're gonna need in terms of infrastructure. That created a little bit of a stir because, the local substation and some of the, the wires there needed upgrading, and that will be on the rate payers, not Meta.

And, you know, that created a holster. But part of the issue is if you look in the deal details is because the amount of capacity Meta will need is so much higher. They need to upgrade the the lines, but that actually benefits everybody. Like, we have a higher voltage and things like that nature.

So there's always weird nuances.

They're damned if they do. They're damned if they don't.

Exactly. Exactly. And then you're seeing other deals too where, like, for example, Microsoft, with Three Mile Island is bringing back a nuclear plant that was previously decommissioned, and that's all that's additional power. Right?

It's going to a local grid. You saw you're seeing twenty of your PPAs on other assets. For example, Meta did that with the Clinton nuclear plant. So there's quite a bit of deals going on.

So nuclear plants coming out of retirement just to power data centers?

In this case, yes. Just to be clear, though, there's only a handful of those that that can happen. So I I think sometimes people see that and they're like, oh, we can do this everywhere. It's there's only a handful of plants in the US that are capable of doing so because there's not far enough along in decommissioning. So that is a very limited source. We only have a few of those, but, I mean, you're talking you have people talking about new nuclear now and small modular reactor nuclear and stuff like that.

But amazing.

I mean, I say big, I'll say.

I say just to power data centers. I don't I'm not I'm not in a disparaging way like that. That's not not a good use of power. But, I mean, as a as a as a power professional, it just it's just mind boggling.

And so speed speed to power is a new thing. Right? Rather than speed to market, everything's about time to get your connection or time to power. And you talked earlier about if you're gonna build a data center, you've gotta be thinking about supply chain, and you mentioned specifically, I wanna just double double down on this bit, on turbine slots. So lots and lots of talk and chatter about how turbine manufacturers are, totally at capacity.

So when you're this must come up a lot in earnings calls. How are people thinking about getting hold of gas, gas generation turbines?

This has been a really topical headline as you allude to, and I think people sometimes confuse the fact that there are multiple types of turbines. Oftentimes, when you hear the large turbine providers, so it's like GEV, Siemens, Mitsubishi, there's a global oligopoly. You know, they control, I don't know, call it seventy to eighty percent of the global market. They talk about long timelines leading into twenty twenty eight, twenty twenty nine, twenty thirty onward.

That oftentimes is for a certain type of turbine. Right? So that that could be like your workhorse CCGT, which is a combined cycle gas plant. But that's different than, for example, like, aeroderivative engine or sorry.

A rice engine aeroderivative turbine or a simple cycle industrial turbine. And those different turbine types across the different providers have different lead times. But generally, you're looking at a few years is the moral of the story. But sometimes that headline goes crazy where it's like you can't get any type of turbine for five years and that's generally not true or we're seeing that.

But you are seeing folks look at access to equipment because if you can't get something in four years, then that time to power question obviously comes into question. So I think what we're starting to see though with some of these behind the meter setups is people getting flexible about the types of turbines they'll use or plugging in a bunch of reciprocating engines together. You hear a long time about wacky things from data center developers in terms of people that want to just get power soon. And then this idea that they'll do that for only two to three years until they can actually get plugged in with the utility.

Because building a massive CCGT again is is a huge project. That takes a few years on its own.

Absolutely. Yeah.

If you're doing that, that's a whole different avenue, but a lot of people are trying to focus on this bridge element to where you're taking these smaller turbines that might be cheaper, easier to get, stringing together a bunch, and then hopefully you can get plugged in a little bit sooner with the utility.

It's funny. I've got this this this vision of we have this big problem. Actually, I noticed from here in New York as well, catalytic catalytic converters being stolen from cars. Right?

Because you can go get a hundred dollars or something. I've just got this this vision of, you know, these, these Airbus planes. You wake up one morning and someone's taking a turbine off it. There's just a big shortage.

There must be all sorts of, weird and wacky combinations that are being considered now. As you say, air driven turbines and and whatnot.

Yeah. Yeah. Absolutely. I'll give you a funny anecdote, something that helps bring to life. I haven't seen any validation of it, so take it for a grain of salt.

But, apparently, XAI is rumored to be importing a power plant from elsewhere in the world because it's that in terms of backlogs and access to capital, you know, you're not able to get one sooner that they're thinking about, like, deconstructing a plant that's that must be modular in nature from elsewhere in the world and shipping it into the US. And I'm talking data cell data center developers that work in different markets. And if you have a underutilized turbine somewhere, you might be able to move it and get a higher value elsewhere in a different market. So you're hearing about crazy things like that.

Like, this is so foreign sounding to me. I've been in I haven't been in power as long as decades, but been in it for ten years enough to see, you know, different cycles and behavior and this type of stuff I haven't seen before. So it just goes to show that the lengths people are willing to go to.

And so so one more on the data center craziness because I think that's the only way to describe it. Also totally rational as well. So what about these other supply chain companies? I know, that you are you know a lot about companies like Eaton, for example, and and other infrastructure companies or but they provide, you know, power infrastructure or whatnot. How are they responding to this? You know, transformer manufacturers or and then you've got this the sideline is you got the the big made in America movement as well. So all of this coming together, how are they responding to to this?

Absolutely. I mean, you're seeing a lot of growth, quite frankly. Right? I mean, you mentioned, Eaton, which is a big electrical equipment provider at in the recent quarter.

Off the top of my heads are not memorizing, but I think it was, you know, that fifty percent revenue growth and fifty percent order growth as well, I think, in that data center business. You have phone companies like Vertiv, which is, you know, the massive electrical and cooling provider to the data center business, and their trailing twelve month backlog is up twenty percent and, you know, record new bookings in the quarter. You have the EPC companies, like, you know, the Qantas, Comfort Systems, and etcetera, subcontractors. But, generally, you know, you're seeing really strong healthy growth in their backlogs.

I think for Comfort Systems, for example, like data centers are a third of their overall, RPO, which is, you know, remaining performance obligations. So kind of a backlog type of figure. So you're seeing a lot of folks, you know, ultimately see that they're booking new business. The other big thing, though, is labor cons is increasingly constrained, and some of this equipment is, you know, a long lead time.

So on the electrical and cooling side, you've seen a lot of capacity come online in the last few years. So people investing in bigger factories or, you know, opening new lines to manufacture more units. On the EPC side, you can't just create more skilled labor out of it takes time. Right?

It takes a few months. So you're increasingly seeing, issues with that where, you know, folks that actually have a huge workforce or the ability to train up workforce quickly might win more projects. And certain companies that, you know, point to that in terms of their competitive advantage. You're also seeing the the, you know, the introduction or, like, of or at least not introduction, but increased demand for prefab, which is basically, like, trying to build, you know, certain equipment or parts of the facility in a modular sense.

You can just ship it to the site. And that way, you don't need to, like, ship materials and have people assemble it on-site. You can kinda do that, through a modular or prefab type basis, and you're starting to see that uptake just because there's, you know, so little labor. And, again, it's all about the speed to speed to market.

Speed to market, speed to power. Yeah. So when you're doing your research on these these, these hyperscalers and you're you're listening to these these, investor calls, it all feels very America focused. Have I got that right, or, or is this spreading around the world? I mean, European data centers are are we gonna have European data centers at the scale that, we're seeing in Virginia, for example?

It's a good question. I I definitely say it's a it's a global trend, but I'd say it's very driven by certain pockets of, you know, regions right now, the US being one of them. Also, there's, like, the n of one issue where I focus on US equity. So a lot of the companies and commentary I'm pointing to is predominantly US focused.

But, you know, these some of these companies do sell in other markets and talk about strength. I think, ultimately, just to your point that it ultimately comes down to the constraint, like, what's the key bottleneck? And it is the access to power. And so, you know, the US becomes a main area because a lot of our labs are here.

And then, you know, we theoretically are trying to answer the call to bring on power sooner. You saw that with the executive order and the, you know, AI acceleration plan. But you're also seeing other regions. Like, you know, for example, China has considerably stronger reserve margins, which is basically a source of how much excess power you have in your markets.

And you'll start to see, you know, demand go elsewhere if they can't get power here sooner. You know, European market seems to be strong. A lot of the private capital providers, the big PE shops, the world really are bullish on growth there. But at least just from a a land perspective and cost perspective to get it done.

Right? It's probably a little bit easier in the US or China than it may be in Europe, just not knowing that market is close. So definitely a global trend, but I'd say, you know, the bulk of activity is probably heating up in the US and China regions. I forget on a global capacity basis, but I wanna say the US is somewhere like fifty to sixty percent of some of these, like, high end facilities.

And obviously, all the leading labs and models are coming out of the US, but increasingly so China, especially on open source. So that was definitely start to raise, gear up. But one thing I would also call for listeners, though, is, like, the issue of sovereignty is gonna really start to play a a you know, a huge issue soon. You could think about how much data people are putting into these AI machines.

And all of a sudden, you know, the US versus China and their geo geopolitical battle will get increasingly even more important here where all of a sudden you might start to see a government say, like, you need to have your data or the data you can hold on our consumers needs to be built in the US and not go through a Chinese data center or or elsewhere. So it's an important thing to pay attention to.

I've said it before on this on this podcast. So, anybody who knows me knows I'm a bit of a Bitcoin bug and, which might be a controversial thing to say. But one one of the things I I I've just got this this this idea that even if there is total overbuild out of the date data center, so I'm I'm a believer that I think sovereignty of Bitcoin mining will become an important thing over time. I don't know what your opinion is on on on Bitcoin in general, but that's my belief And, that mining Bitcoin on on on home turf will become a a big political issue, at some point in the future.

But the the equipment that you have for these AI data centers is is the same equipment that you need to mine Bitcoin, right, give or take a little bit. And so even if there's overbuild, you know, the the the optimist in me says, well, even if there's overbuild in this all of this, AI data center infrastructure, if it turns out the bubble bursts, then just we'll just use it all for Bitcoin mining, and that's that's a good thing. That's that that that is a good thing if you if you are one of the very few people who believes that that, that that's that's a good solid future for the world.

Okay. Let's let's talk about I wanna come back to you investing in public equities and back to your fund for a moment before we get to your contrarian view. So I wanna come on something more forward looking now than so so, some themes to watch or or signals.

So what are some of the things you've heard in the public markets recently that you think we should just, keep an eye on, some examples? I know you've mentioned some, but, what should we keep a look keep a look at?

Yeah. I mean, I think a lot of what I've pointed to is probably what I would continue to point to in terms of just, like, the, you know, the backlogs have been pipelines for these companies. Like, are the pipelines and the book to bill ratio for these construction companies continue to going up because they're probably the immediate beneficiary in the nearest term indication of things are getting built or not. And then upstream on the equipment, which is a little bit longer lead time, are those backlogs still growing?

Are they getting elongated? Are they coming in, which is a sign of maybe supply and demand being the same? Nvidia's earnings calls yesterday are right. And I think if I got the number right, Jensen was calling out for three to four trillion dollars of compute infrastructure that's gonna be spent on, you know, in the next over the course of the next few years.

And so if those numbers are true, right, like, I'm we're talking about half a trillion dollars earlier being a huge number. I mean, three to four is bigger. So, I think paying attention to some of those things where, like, you know, these are grand proclamations and seeing if the numbers actually meet the narratives. Because the big thing we're trying to do as investors is, you know, everyone's always selling something.

Right? And everyone's always talking about, you know, how their charge is up in the right and will continue to go up in the right. And I think, ultimately, it's putting numbers to pen and paper. It's like, does this make sense?

And can we actually grow that that big? And it's ultimately just doing that exercise of tying that big macro picture to is that actually happening at a at an independent company level or on a bottom up basis?

Okay. Now now I wanna get to well, my last two questions. Firstly, is there anything you wanna plug for our listeners? We'll put some links to your website and some of the other media stuff you've done in the show notes. But anything you wanna plug, and then we're gonna get to your Contrarian view.

Not necessarily. I appreciate that. I mean, I'm definitely always trying to share, useful information. And when I do kind of some of these speaking things, it's it's generally trying to educate folks and, and learn and honestly debate in public.

So I think if anything, the plug engage me in some type of capacity, if you have different views or you think I'm misunderstanding or misinterpreting something, I'm always looking to, you know, learn more and, you know, update my assumptions on the world.

And so now onto your contrarian view. So so what is it that you believe that not a lot of other people believe?

Yeah. I'd say I'm not sure that not a lot of people necessarily don't believe it, but I think it's getting underappreciated relative to the opportunity in front of us is that, with these data centers, a lot of times, you know, we've obviously talked about the heating needs and the cooling needs and the electricity use at the chip level. What a lot of people don't realize is a lot of times these chips oscillate in energy use. So you have a data center that can sit idle and be, like, a forty percent utilization of their raw capacity, and that will spike up to a hundred percent in a matter of milliseconds.

And that's largely because of how the chips work today. And I think over time, you know, one of the ways to smooth that out that's that's usually not great for a power plant, by the way. Power plants wanna run, especially like a gas plant. It's not meant to flex up and flex down.

It wants to just run a stable Stable. Stable load factor. And so one of the things that you'll see and what you're already seeing in some sites, for example, the XAI one, is batteries are really smoothing that out over. And so I think, you know, there's this this almost, quote, like, cultish or religious battle going on between the solar and storage folks versus gas folks.

And, you know, there's a lot of, like, highlights about gas. But I think in either scenario, you're probably still gonna need a lot of storage, just due to some of this transience workload issues as, they're referred to in the industry. And so you can see that with that project I mentioned earlier in the podcast, the nine hundred forty four megawatt behind the meter gas generation, that is being paired with storage. So gas and storage there, whereas typically you'll think of solar and storage going together.

I think storage is gonna boom in kind of either respect just given that transience of workload issues and as well as as you increasingly get, you know, more and more of this load growth. I mean, data centers are strong and pronounced right now, but we still have the electrification of vehicles coming up. We have electrification of industry coming up. And I think, you know, the growing importance of storage will, continue to be there.

And so I think regardless of it even if we build more gas than previously anticipated or as anticipated a few years ago, storage still is gonna play a role, and you're seeing that in the growth rates and development plans.

Yeah. It's one one of the one of the overlooked things is data sensors assumes to be baseload, and they, they can behave a little bit differently.

Yeah. Yeah. I'd say they generally are, like, a higher load factor, higher base load, but there are these weird issues where, again, you know, if you're spending billions of dollars, you don't want your equipment to wear out sooner than it should.

And so your take on that is, even with the load of this data center base load in inverted commas, we're still gonna need a lot more storage.

Yes. Yeah. Even even if gas is meeting the call because I think that's what a lot of folks are talking about today, and I think some of the legacy thermal asset, you know, favorites or people that support that, often underplay the role that storage will play regardless.

Alright. Well, thank you very much for coming to join us on the podcast. It really has been a lot of fun. If you wanna find out more about Shanu and the work that Lazard are doing, we're gonna put links in the show notes. And, I would recommend that you sign up for Shanu's monthly blog, where he shares all things, climate investing and public equities and lots of data center power stuff in there, which is great. And thank you for coming to see us in in New York. It's been a real pleasure.

Thank you for having me and a really productive conversation. I really enjoyed it. Thank you.