The Bridge Ep. 29: Who is Funding the AI Buildout?

The Bridge EP 29, Winfield Sickles/ Blackstone – Transcript

Winfield Sickles (00:00:00 -> 00:00:04)
Electricity was invented, but it was another 50, 60 years before he had night baseball.

Sonali Basak (00:00:04 -> 00:00:12)
So, Blackstone, of course, is one of the largest alternative asset managers in the world and is very closely tied to everything happening in artificial intelligence,

Winfield Sickles (00:00:13 -> 00:00:15)
Covering everything that we’re doing in AI infrastructure,

Sonali Basak (00:00:16 -> 00:00:20)
From data centers to compute and the cost of compute and the Frontier labs.

Winfield Sickles: 00:00:20 -> 00:00:23)
What we are doing is very diversified and different.

Sonali Basak (00:00:23 -> 00:00:27)
And so, we’re going to talk to Winfield Sickles who’s going to explain how it’s impacting everything.

Winfield Sickle: (00:00:27 -> 00:00:33)
We’re big believers in the productivity that sits on the end of this AI infrastructure build out.

Sonali Basak (00:00:51 -> 00:01:33)
Welcome to the latest episode of the Bridge by iCapital. I’m Sonali Basak, the Chief investment strategist at iCapital. And today I am joined by Winfield Sickles. He is a managing director and co-leads the global private wealth investment strategy at Blackstone. And Blackstone has been very bullish on artificial intelligence at a really interesting juncture. Winfield, you know, you think about just how far, how fast we’ve come and recently John Gray has said that it’s probably the single most important investment theme for Blackstone’s growth for your clients. How do you expect this all to play out? Now that we’ve hit a crossroads in the trade, we’ve seen a little bit of nervousness build. Um, where’s the optimism right now and where’s your concern?

Winfield Sickles (00:01:34 -> 00:03:09)
Yeah, well it’s great to be with you Sonali, and I think maybe crossroads in the public markets, but in the private markets at least how we’re attacking it, we would submit that it’s earlier in the cycle. And I think the reason there is because we’ve invested holistically across multiple verticals in this stack of opportunities. Really our bread and butter to start has been everything around land, power, shell, all of those bottlenecks. So, you think about land, you need permitting. Power, the lead times to energize a data center can be five to seven years now. And so, if you think about electricity demand in the US had been flatlining for 30 years. Now it’s going to grow 40% to 50% over 10 years. So, anything from generation, transmission, distribution we are investing in. And then the shell, the data center itself where we do these 15 plus year leases with escalators with high grade, uh, hyperscalers. And so, starting with that bread and butter, but now compute is an asset class, which I know is something we were talking about earlier. Yeah. That early talk we, we, we wanted, we wanted to talk about. But just really quickly because you know, I used to be a commodity trader and just this morning I was looking at the Ford curves in compute. If you look at B 200 chips, they’re basically flat out three years. Even the A 100 chip, which is the GPU chip from 2020 is up year to date. So, we’ve been investing in compute and then as you know, and we’re also discussing earlier, some of the frontier LLMs, the application layer. So, for us it’s really been more of an all of the above approach.
Yeah, well it’s great to be with you Sonali, and I think maybe crossroads in the public markets, but in the private markets at least how we’re attacking it, we would submit that it’s earlier in the cycle. And I think the reason there is because we’ve invested holistically across multiple verticals in this stack of opportunities. Really our bread and butter to start has been everything around land, power, shell, all of those bottlenecks. So, you think about land, you need permitting. Power, the lead times to energize a data center can be five to seven years now. And so, if you think about electricity demand in the US had been flatlining for 30 years. Now it’s going to grow 40% to 50% over 10 years. So, anything from generation, transmission, distribution we are investing in. And then the shell, the data center itself where we do these 15 plus year leases with escalators with high grade, uh, hyperscalers. And so, starting with that bread and butter, but now compute is an asset class,

Sonali:
Yeah, we’ll definitely talk

Winfield:
Which I know is something we were talking about earlier. We wanted, we wanted to talk about but just really quickly because you know, I used to be a commodity trader and just this morning I was looking at the forward curves in compute. If you look at B200 chips, they’re basically flat out three years. Even the A100 chip, which is the GPU chip from 2020 is up year to date. So, we’ve been investing in compute and then as you know, and we’re also discussing earlier, some of the frontier LLMs, the application layer. So, for us it’s really been more of an all of the above approach.

Sonali Basak (00:03:09 -> 00:03:33)
So, we’ll, we’ll break down that food chain. We have to, right. Because it’s really important, I think for investors to understand that opportunity set in different parts of the food chain. You mentioned power. Um, how much of the power trade is tied to AI and how much is decoupled? Because one of the most interesting things about power is that even without AI it’s interesting, important and demand is expected to rise.

Winfield Sickles (00:03:33 -> 00:04:01)
Yeah, I think that’s a great point because as much as people attach it to the data center dynamic, if you actually look at the call on power today in the US only about 3% of it is data centers. It is going to double and triple over the next 10 years. So, it’s going to be a key driver. But bigger than that is digitizing the economy. I’m patting my phone in my pocket, right? The electrification of the economy, re industrializing the economy. So that’s one of the things that’s driving power as much as anything.

Sonali Basak (00:04:01 -> 00:04:31)
Yeah, people forget that AI is not the only technology out there.

Winfield:
Exactly.

Sonali:
Yeah, and so, you know, you also mentioned the frontier labs, you mentioned the data centers. As we sit here today, there’s started to be a lot of nervousness building around neoclouds. This idea that you have a subset of players out there who tend to be, you know, shorter duration in the debt that they take on. Um, maybe a little more highly levered, thinner operating margins as a whole. This is general generalization.

Winfield:
Yep.

Sonali:
But I believe this is a place you believe in too.

Winfield Sickles (00:04:31 -> 00:05:05)
Yeah, no, we’ve, again, building on that bread and butter, land, power, shell, compute is something that we’re incredibly excited about because when you think about those dynamics, I started to unpack in terms of the value of compute. It’s a huge bottleneck and so, if we can do these three-to-five-year financing deals with the Broadcoms, the Nvidia of the world, frankly there’s a big enough bottleneck that we created a JV that we’re doing with Google to get their TPUs out into the market. So, of course there’ll be some misallocations of capital, but in terms of how we are doing it, we feel really good about it.

Sonali Basak (00:05:05 -> 00:05:23)
Yeah, it’s interesting because all the names that you mentioned are all kind of very large companies, um, tried and tested if you will. So, you know, are there places that look frothy in the industry? Broadly speaking is how important is it for you guys to be right next to a blue-chip player side by side?

Winfield Sickles (00:05:23 -> 00:05:44)
I think it’s critical, because at the end of the day, we’re focused on discipline, selectivity, quality cash flows. I think up until, to your earlier point about the crossroads, I think up until a few weeks ago, maybe some of that discipline had been thrown to the wind for some actors. But we’ve stayed true to the way we invest. And again, we feel good about the go forward.

Sonali Basak (00:05:44 -> 00:06:10)
Right, because the reason, we’re here too is because the cost of capital is rising, right? Uh, and that has really interesting implications, particularly for somebody like you because if you’re on the other side and you’re lending, then you have better rates of return, right? But you also know that there’s probably going to be some players there that are struggling under the weight of that capital also. How are you thinking about the AI build out in light of a higher interest rate environment?

Winfield Sickles (00:06:10 -> 00:07:44)
Well, I think first of all, and you’ll have to interrupt me because again, talking about my background earlier as a macro trader, I could go on and on about higher rates. But I think some perspective is helpful in terms of how we got here. And you know, on one hand that higher cost of capital, there’s sort of a negative piece to it, right? It’s everything that’s happening in terms of geopolitics and wars, higher energy prices, food prices, it’s shown up in inflation. And then this massive pile of debt, not just in the US but in developed market economies in general. But then on the other side you have this resilient global economy, strong corporate earnings, solid consumer and then sort of undergirding all of that as this huge AI infrastructure buildout, which is frankly broadening to include reshoring, supply chains, reindustrializing power, which we were talking about. So, when you put it all together, I feel like that’s what’s taken the cost of capital higher. And it almost feels like Sonali, that we’re undoing the first 20 years of the century where you had this excess savings glut, right? You had negative interest rate policy, you interviewed people for years about that, right? You had suppressed volatility. And so, in many respects it’s like we’re just normalizing the cost of capital to reflect this amazing excess investment opportunity that we see today. I’m actually about to put out a piece, it might actually go out today that zooms out on the 10-year treasury yield at 5%. If you go back to 1990, we’re literally just in the middle of the pre-GFC range. So, I think that context is important.

Sonali Basak (00:07:44 -> 00:08:17)
Yeah, it’s interesting, we were also looking at bond market volatility and you’re kind of just pre GFC is, is what we go back to.

Winfield:
Totally.

Sonali:
And uh, you were very steeped in the hedge fund world at that time. Uh, you say you’re a self-identified macro nerd. Uh, me too. Uh, you know, with that said, you’ve also said, you know, AI is everything everywhere, all at once. And the way that looks as it pertains to GDP is it’s been estimated different ways, but as much of a third of GDP contributions of late, right? And so, if you see the AI build out, start to slow at all, what does that mean for the broader macro?

Winfield Sickles (00:08:18 -> 00:09:53)
So, let’s take a step back again, and this is a good transition in terms of what we’re seeing in our portfolio companies because we just had our latest drop of third quarter data. And on the hard data side, if you look at our portfolio companies, you see double digit revenue and earnings growth. Now, if you sort of pull out sectorally, what’s driving that? Who are the outperformers? It is that stuff that’s downstream of AI infrastructure. Think energy, think industrials. I take great value as well in the soft data that we get from our CEOs. We survey them quarterly and ask them all kinds of questions about where’s your CapEx going, where are your revenues and earnings going. On all three of those pieces, it’s quite positive. In fact, over 95% of our CEOs expect their revenues and earnings to grow through 2027. And I sort of go there with your question because as much as there’s outperformance in those industrial and energy sectors, there’s not really any key underperformance if you look across all of the sectors that we’re in. And then if you look at the labor market, and I want to connect this to AI too. You see a lot of balance. We’ve seen wages come down pretty significantly across our portfolio portfolio over the last three years. But we’re seeing this big blue collar job renaissance because to build out the infrastructure, we have this huge shortage of electrical engineers and welders and heavy machinery operators. So QTS, our data center play primarily in the US had something like 13,000 people on site beginning of 2025, they’ll have 40,000 by the end of the year. So, to land the,

Sonali Basak (00:09:53 -> 00:09:54)
That’s not a small jump.

Winfield Sickles (00:09:55 -> 00:10:26)
So, to land the plane on this, it’s not small at all. And it could go to 70,000 next year. What we think is underappreciated, now, it’s not the whole economy, totally understand that. But if you were downstream of AI infrastructure, again, broadly speaking, industrialized, digitize, electrify, power, et cetera, strong corporate profits, strong CapEx, blue collar jobs that are high paying supports the consumer. That’s a flywheel that we think is underappreciated in the economy.

Sonali Basak (00:10:27 -> 00:10:36)
Fair enough. I mean, the thing is, when you look at the road ahead, um, you said, let me get this right, you said 15,000 or was it, what was it, early 2025 for QTF?

Winfield Sickles (00:10:36 -> 00:10:36)
Uh, 13,000.

Sonali Basak (00:10:37 -> 00:10:40)
13,000,

Winfield:
yep.

Sonali:
up to 40,000?

Winfield Sickles (00:10:40 -> 00:10:40)
40 by the end of the year.

Sonali Basak (00:10:40 -> 00:10:56)
Then 70,000. The question that people have about these jobs is how sustainable are they? How often do you get asked that? And how do you think about kind of the sustainability of this employment potential, right? And we, I don’t think we can call it a boom yet, can we?

Winfield Sickles (00:10:56 -> 00:11:57)
I think, I think it’s, again, I, I think we would say it’s underappreciated now, you’re right, we get asked this a lot. At the end of the day, it is contract labor. But that term contract, I think, you know, it kind of connotes like, oh, a few months or a year, these data centers, you know, a one-gigawatt data center now over probably $50 billion to build three, four, or five even longer years. So, in that sense, these are medium to long-term jobs. And again, I think the context of the starting point of where the labor’s going to come from, yes, we have bottlenecks and land and power and compute and capital, but we also have it in people. There are estimates that we are hundreds of thousands of jobs short to do all of that skilled labor because most of the people that are trained to do it are my parents’ age and are demographically aging out. And we have not trained a new generation to do these jobs. In fact, Blackstone’s made some commitments, some donations in some states to help bring that skills gap up.

Sonali Basak (00:11:57 -> 00:12:30)
So, we’ll talk more about the macro because I want to get your thoughts, but I do have another question about AI first before we move on because the part of the food chain we didn’t really tackle was the frontier labs. And the frontier labs, when you look at them today, they’ve grown so tremendously. But the token costs that we see in the world are starting to drop. Um, there are a lot of questions about whether that stays that way or not. At the same time, there’s this potential for margins to be compressed because of the cost of the build out, right? Everything everywhere is also more expensive. So how do you think about the future of the frontier labs from where you sit?

Winfield Sickles (00:12:31 -> 00:13:18)
I think it’s important to, if you are investing in the frontier labs, you want to be with the, the leading players. I think that’s important, number one. Number two, if you look at the breakdown between training and inference, I think there will always be demand for the leading-edge training piece of the puzzle. I think about, we were talking about my hedge fund background earlier, some of my buddies in the hedge fund space who are working at scale with these models and are using a lot of orchestration tools, of course, where you can route some of that task load to a cheaper cost, you’ll do it, but a significant amount is still going to that higher end because of how sophisticated I think you want to be depending on your profession. And so, I think the LLMs are, are, are really important.

Sonali Basak (00:13:19 -> 00:13:41)
You know, it’s interesting about, um, the compute dynamic you were saying earlier. I think one thing that I just want to unpack for a moment here, you started talking about the chips and the value of the chips too. Um, when people think about the neoclouds, I don’t know, I would argue actually that they, they’re maybe underestimating the fact that these chips can stay valuable for more, for more time than initially expected.

Winfield Sickles (00:13:41 -> 00:14:12)
Yeh, they were, I mean initially some of the depreciation forecasts were pretty intense, you know, three, four years. Uh, that was, you know, I brought up the A100 from 2020 earlier. The fact that this 6-year-old chip because of the bottleneck in compute is now trading higher today than it was in the beginning of the year. The fact that if you look at these forward curves, as I was bringing up earlier, the B200 is essentially, um, flat all the way out to three years. Just shows you that the market is pricing in a real shortage of compute going forward.

Sonali Basak (00:14:12 -> 00:14:40)
Yeah, it’s interesting. So, there’s another angle here on AI that kind of catapults you into a whole new territory. So, a a year ago people would’ve said you could’ve built data centers anywhere in the world for use anywhere in the world, but the world has arguably gotten a lot less safe. And that notion has changed and it’s made defense actually a huge theme in national security. I mean, how do you think about this buildout when you think about how much people need to actually bring their compute home?

Winfield Sickles (00:14:42 -> 00:16:18)
Interesting. I mean, there are two different directions I can go there. I mean, one direction is terrestrial, are we going to outer space with data centers and Gavin Baker has been interesting on his podcast talking about that. I’ll just say he’s a technologist, he’s an expert and he believes in all, in all of the above strategy, even if we get there, uh, in terms of Elon Musk’s vision. But on the national security side, I mean, I think that’s something that’s really worth double clicking on. And you would undoubtedly have to interrupt me here because I could go on forever on this one, but I think for perspective, and I’m going to broaden it out beyond data centers if that’s okay…

Sonali:
Yeah.

Winfield:
with you, Sonali, is I want you to think about the world, right up until we had kinetic conflict in Russia and Ukraine. 20 years leading up to that moment in 2022, you had seen national security spending from Europe, US, and Japan drop by about 70 to a hundred percent. Okay, that was the world that Russia, Ukraine landed on. Ergo, NATO wanting to increase defense spending from two percentage GDP to five percentage of GDP, this is trillions of dollars of more spending. So, the spending is getting bigger, but what has us so excited about it is how different the spending is going to be. And that’s what gets to your point. The modern battlefield is unmanned, it’s autonomous, it’s drones, it’s protecting critical cybersecurity; infrastructure, water, telecom data as you said, power and so, we think national security because of that, has become a whole of economy concept really for the first time since World War II.

Sonali Basak (00:16:18 -> 00:16:54)
So how are you tackling it? Because, I don’t know, it just feels to me that when you look to 2027, it’s going to continue to be one of the biggest areas that investors are going to want to keep flocking to, right? Because you have this dynamic where, um, as we said, the world is not as safe, people are looking to near shore. Um, you mentioned water, right? Somebody told me 40% of American water is wasted through broken piping, right? So just to protect, um, you know, our livelihoods, there’s actually a really large investment going on and it’s AI, infrastructure, and defense all kind of float rope together.

Winfield:
Totally.

Sonali:
So, what’s the strategy?

Winfield Sickles (00:16:54 -> 00:18:07)
Yeah, that’s why I put, I sort of pull them together in the beginning when we talk about AI infrastructure, I like to have this idea of strategic autonomy running right beside it because that is re-arm, reshore, resupply, everything that you just said. I think in terms of national security specifically for us, I’ll sort of break it into two barbells. On one end, you have to think about the unit economics today. Let’s pretend like you, you and I are in the Pentagon, right? We’ve got a trillion-dollar defense budget. We’re trying to figure out how to spend money. Do we spend $50 million on one fighter jet, or do we take that $50 million and get a thousand, $50,000 drones, right? And so, on one end of the spectrum, we’re really excited to invest in kind of that high tech development capital starve space on the other end, which in some ways rhymes with the land, power, shell, the critical infrastructure of national security. I want people to think about where is all of this new hardware going to go. There’s a stat out there that Europe needs something like almost 40 million square meters of just new logistics to store all of those drones and whatnot that they’re going to invest in. So, I think both on that infrastructure tail and then on the high-tech tail, we’re quite excited.

Sonali Basak (00:18:07 -> 00:18:33)
Yeah, I, it’s interesting because when we think about the concept of defense these days, we were talking about, you know, national security as it pertains to investing in our own infrastructure, making America a competitive place, right? Um, but then you think about areas that get closer to the center of war like Europe as you’ve, as you were saying. And these are kind of long-term structural shifts. I think that’s maybe what a lot of people fail to understand when they’re thinking about, you know, the meaning and of defense in this environment.

Winfield Sickles (00:18:34 -> 00:18:56)
Absolutely. I think you nailed it and I think, I think the way you combine it together with those things that are more approximate to the battlefield with just basic, what we take for granted, I think. But when you travel to some other countries and you see where they’ve been ahead of the curve on infrastructure, you know, in the US that we need to step up our game and so, it’s good to see all of those arrows moving in the right direction.

Sonali Basak (00:18:56 -> 00:19:20)
Yeah, I’m going to somehow that you’re going to be amazed at the jump I’m about to make here. You know, when I think about defense spending across the world, I also think about not just short-term interest rates, but longer-term ones and the fiscal term premiums we’re seeing across the world, right?

Winfield:
Yes.

Sonali:
Because it just feels like that defense spend, um, alongside interest spend is going to keep driving up the long end. Um, do you feel that that’s a possibility?

Winfield Sickles (00:19:21 -> 00:21:06)
So, Sonali, if you look at spending in the US, let’s just look at the US today, right? We spend about $7 trillion a year as a country. $4 trillion of that is mandatory entitlements, social security, Medicare, Medicaid. We haven’t really tried to tackle that since the Simpson Bowles Commission, which I love to bring up because I’m from North Carolina and Erskine Bowles hails from that state. That was the last time that we tried to tackle entitlements, and since then, it’s become kind of political kryptonite to go there. Um, if you look at the rest of the spending, it’s about a trillion for national security, as you said, a trillion for discretionary spending, which was the DOGE effort that we saw, um, in the last couple of years, which, which didn’t move the needle. That last trillion is the cost to service the debt. And that was something like $2 billion a day 10 years ago, and it’s projected to go to 3, $4 billion a day. So, I mean, mathematically the answer is yes to your question. It matters. I think what we are more optimistic about, and the Treasury Secretary’s talked about this is, is growing our way out of it. We’re big believers in the productivity that sits on the end of this AI infrastructure build out with the help of Ode, our partnership with Anthropic to help our portfolio companies get their arms around this technology. We’re really starting to see that productivity come through in our portfolio companies. And if the late nineties is any guide, we went from the early nineties to the late nineties, productivity, labor productivity tripled to 3%. We went from a fiscal deficit to a fiscal surplus. Labor productivity 2013 to 2023, also one, 1.5%. We’re already starting to trend above 2%, so, the green shoots are there for that kind of optimistic scenario coming to play.

Sonali Basak (00:21:06 -> 00:21:14)
Right, I think that that, this is kind of the, the battleground question in macroeconomics. How much is AI filtering through the economy?

Winfield Sickles (00:21:15 -> 00:21:40)
And and the problem is, as you know, is that there’s some people that would take the other side of my comments and say, oh, you can’t see it yet. But if you again, look at the nineties because of the measurement issues and productivity, just from a pure measurement perspective, you couldn’t necessarily see it until a few years later. So, from a mechanical standpoint, you might not be able to measure it until five years later. So, we’re going to stick with what we can see in our actual portfolio companies.

Sonali Basak (00:21:40 -> 00:22:12)
So, uh, let’s, let’s just sit on that for just a second because yeah, you’re right. There are people out there who will say, well, wait a minute, you’re not seeing AI really impact anything else but the technology space. They who shall not be named , right? Uh, so what do you say to that? Where are you seeing real ability of AI to pay off within your portfolio companies, right? Because you know, some of the stats that you have under your portfolio, you have what, 700, more than 700,000 employees, more than 5,000 issuers you’re working with here?

Winfield:
Yes.

Sonali:
So, where’s the proof in the pudding?

Winfield Sickles (00:22:12 -> 00:22:21)
Almost 300 portfolio companies, I mean, and, and listen, if we look at those PE portfolio companies today, spend LLM spend is up 20 times year over year.

Sonali Basak (00:22:21 -> 00:22:28)
So, you’re seeing it in your portfolio companies. Do you think you’re really seeing it in the broader macro yet?

Winfield Sickles (00:22:29 -> 00:23:18)

In terms of labor? I think you can see it. I think you can see it in the public markets if you look at measurements like revenue per employee, we can see that in our portfolio companies too. I think what’s exciting for me and for us is that, you know, from a labor market perspective, we talked about this a little bit earlier, but we’re still growing jobs across, you know, you said 700,000 plus employees, we’re still seeing net jobs growth. Um, again, driven mostly by that blue collar jobs renaissance. But based on what happened in the first quarter of this year, who would’ve thought that, and this is now public market stat, that software engineer hiring would be up 10% year over year. Like there are being new jobs created by this AI boom, just as there are some being, you know, changed.

Sonali Basak (00:23:18 -> 00:23:28)
But do you feel that, you know, there’s kind of a big responsibility for retraining at this point too, given that even the software jobs that we’re seeing, they’re very different than the software jobs of the past?

Winfield Sickles (00:23:28 -> 00:24:12)
Absolutely. Retraining is critical. I think, you know, this is an interesting, again, sort of public ecosystem stat. IKEA had, I think it was somewhere like 500 customer service agents where AI made 50% of their workflow obsolete. And what IKEA did instead of, you know, laying some of those people off is they retrain them into remote design agents. Um, great for those workers. I think it also generated 500 million to a billion dollars of incremental revenue for IKEA. So, um, just using that as an example, we’re big proponents of retraining, and again, to my earlier point, it’s one of the reasons why we’ve made some donations in states like Iowa to help that retraining take place.

Sonali Basak (00:24:12 -> 00:24:22)
Yeah, it just feels like the private sector is going to have to kind of reorient in, in this pretty meaningful way to get people into the jobs that are, um, you know, consistent with the economy of the future.

Winfield Sickles (00:24:22 -> 00:24:50)
Totally agree. And my mom’s teacher, I grew up in education, all this stuff is important. I think another thing worth flagging is in the United States, just on an average year, we create and destroy both about 30 million jobs a year. And we, we usually net create. Again, using the nineties as an analog for when we last had a big technology boom. What’s interesting, Sonali, is that the destruction went up, but the creation went up even more. And I think that’s exactly what we see playing out there.

Sonali Basak (00:24:50 -> 00:25:04)
But is it all in a straight line you think? I mean, you know, between now and kind of that point of change where we see that productivity like flow through companies, the economy, margins, jobs, you know, what happens in between, it can’t be a straight line, right?

Winfield Sickles (00:25:05 -> 00:25:40)
I don’t think, I don’t think anything’s ever a straight line, but you know, if you think about just the disruption that’s coming to companies, you know, and, and we are very, we try to be very prescriptive and disciplined about how we think about it. You know, software was under the gun a lot this year. And if you’re in the software space and you’re a rules-based business that does not have a proprietary mode and you’re in the business of scraping public data and creating an aesthetic and charging a premium for it, the technology doesn’t mean that you’re going out of business, it just means maybe that premium multiple isn’t worth it anymore and we’ve tried to stay on the other side of that.

Sonali Basak (00:25:40 -> 00:26:25)
You know, uh, the, the crazy thing about all of this is that, you know, we sat here in a year where we saw companies raise pretty tremendous amounts of capital in public markets, in equities, in debt markets, public debt markets and private debt markets. The private debt markets have had these entirely, you know, never before seen stickers, right? Uh, $40 billion, $80 billion, $90 billion. So, but the crazy thing to me is that we actually even haven’t seen the tip of the iceberg, yet when it comes to the private capital that’s available for this space.

Winfield:
Yes.

Sonali:
It feels like the AI build out still actually, let’s say at least through 2030 if not much longer, has actually a lot of potential investments still ahead of it.

Winfield:
Yes.

Sonali:
How do you think about that?

Winfield Sickles (00:26:25 -> 00:26:36)
I think there are estimates out there, I won’t even go all the way out to 2030, maybe just two or three years out that we need something like $3 trillion of spend and not including power, by the way,

Sonali Basak (00:26:37 -> 00:26:40)
And estimated $1.2 trillion alone next year.

Winfield Sickles (00:26:40 -> 00:26:41)
Exactly, exactly.

Sonali Basak (00:26:41 -> 00:26:43)
You can see that getting a $1.5 easy if not higher

Winfield Sickles (00:26:43 -> 00:27:10)
From, from the hyperscale’s. And so, you know, if you, if you look at that $3 trillion as I laid it out, if if half of it is going to come continue to come from operating cash flow off balance sheet, those numbers are shrinking, right? Then the other half needs to come from the debt markets, and if that’s one and a half trillion, the estimates are that about half of that needs to come from private capital. So, I agree with you. I think we’re just scratching the, the, the, the tip of the iceberg, so to speak.

Sonali Basak (00:27:10 -> 00:27:28)
Yeah, it, it feels like everything that happened this year is actually just something that’s just beginning.

Winfield:
Yes.

Sonali:
And so, um, you know, last note on AI, you know, I know everything’s not AI, um, and we got to talk about some of the other areas in the macro, but what do you think is the most misunderstood part of the story?

Winfield Sickles (00:27:28 -> 00:28:32)
I think probably the most misunderstood part of the story is just where we are in the cycle. You know, John Gray likes to use this analogy that electricity was invented, but it was another 50, 60 years before we had night baseball. You know, diffusion takes time. Adoption isn’t diffusion. Diffusion is when it is truly seeped into every aspect of your daily life or your workflows at your business. And again, that’s why for as incredible as as it is, I mentioned the CEO survey earlier and we asked those CEOs, you know, where are you in your adoption right, versus tests, cycle, and the fact that we’re at 60% today for our portfolio companies and we’re very AI forward, is both a testament to the success of getting them to use it, but it also shows you that there’s still a lot of room to grow. And so again, forming that partnership with Anthropic Ode to help companies first our own portfolio companies later others, really understand how to use it and infuse it into their systems. Like we were really just at the beginning of that journey.

Sonali Basak (00:28:32 -> 00:28:38)
Very cool. Winfield, thank you so much for joining us here in the studio. That is Winfield Sickles of Blackstone. And you’ve been watching The Bridge by iCapital.