Where’s the Beef? Blackstone's Jon Gray on the Payoff From the Enormous AI CapEx Spend | Sept '26

0:00 You can build with Blackstone. I love that song. 2024 Blackstone Holiday Video. The 2026 edition is coming soon. I love this meeting. I love the fact that all of you have come from around the world to be here. Huge thanks. And of course, we would not have a business without the people in this room, and we are forever indebted to all of you. So thank you. Now what we want to do is talk about the big questions that are on your mind as you think about investing enormous amounts of capital.

0:37 I thought what would be best is to start with the most important question that I know you're thinking about, which is this: how to be a global finance fitness influencer? This, of course, is very important to me. I've got five pointers for you here. First, you need abundant sweat. This fortunately comes naturally to me, but it gives you real authenticity. You need to elevate the camera angle. My daughters constantly remind me.

1:08 Otherwise you get that dreaded double chin. Third thing shameless branding. You want to feel like a NASCAR driver. You've got a logo. You're important. Fourth, dorky dad vibes. All I can say is, check, and five: endless emojis. It speaks to tons of enthusiasm. Makes it fun. So now that we've got that done, let's go to the day job. Four questions we're actually going to try to answer this morning.

1:42 First up, what happened since most of us were in this room together last year? Secondly, what can we learn from the past, particularly the distant past? Third, where are we as we sit today? And finally, where are we going? What's going to happen in the future? Now we're going to do this through the lens of two letters - of course, AI. It doesn't mean we don't recognize the importance of what's happening geopolitically in Ukraine, in the Middle East, driving up commodity pricing, leading to inflation and higher near-term rates, as we saw yesterday, short-term rates.

2:22 Obviously, there are concerns about deficits, longer-term rates. All those things do matter to us as investors. But when we think about what ultimately will drive value in our portfolios, what's happening in the economy, what's happening in the future, we've got to think about those two letters. And so I want to go back in time exactly one year. We were in this room September 17th, 2025. And talk about what were we thinking back then. Here we go.

2:53 What is the main reason I'm enthusiastic - we're enthusiastic - what is the main thing? It is AI. Okay, we are so original. We are very consistent. One year later, and I just noticed my choice of dress has not evolved much in a year. Okay, when we think about AI over the last year or the last couple of years in this case, what is happening using Google token use, you could see off of a base of almost nothing in 24, moved up to 480 trillion monthly tokens.

3:26 But look: as we went to May of ‘26, that's 3.2 quadrillion - a new word I learned for this presentation - tokens. And if we went to September, obviously those numbers would be higher. What's happening at the big model companies that Anthropic and OpenAI? These companies had almost no revenue a couple of years ago. They went up 100-fold by July of 2025. And then look at these numbers, $105 billion of run rate as of July.

3:59 This is really unprecedented in human history. If Now you may ask, is this really happening? Well, we have an amazing lens at Blackstone through our portfolio companies. This data comes from our portfolio companies, our GP stakes portfolio companies, and our borrowers. We had, amongst these 1,400 companies, $25 million of run- rate revenue in September of ‘25. That number is up 21-fold, annualized run-rate revenue, to $525 million of spend with Anthropic.

4:34 That is what's happening out there because these companies are getting extraordinary returns on the investment. We’ll talk about that. Now, obviously it impacts valuations. We're fortunate to have invested in both of these companies. Their value last September, $683 billion. What is it today? Estimated combined between $2 to $3 trillion as they start to move towards going public. Obviously the strength of their revenue model is driving valuations.

5:04 Now, one other thing we talked about really a lot last year was not just the models and the technology, but the infrastructure to get there, and we had some thoughts about that. CapEx spent in this world is up sixfold. These numbers are going to be even bigger next year. Well, I put that up there because I got that prediction right. So obviously I want to look good. Here we go. So hyperscaler capex, five companies, $415 billion last year.

5:39 That number has doubled this year to $820 billion. That's equal to 2.5% of the GDP of the United States. These are extraordinary numbers we're seeing. And it's not just the hyperscalers, of course. We have been actively involved with data center leasing. We have the biggest platform out there across the globe. ‘24 we leased a gigawatt; ‘25, two gigawatts. This year, we think we're going to do at least six gigawatts.

6:10 That's $100 billion, nearly $100 billion, of CapEx. The tenants will put in another couple hundred billion dollars of chips. The scale of this, of course, is unprecedented. Now, one area we unfortunately did not invest was in the memory chip companies. Micron, SK Hynix, two leaders in this space, their stocks were up a modest 600% and 500%. Pretty impressive. Now when you see a chart like this, your immediate reaction as an investor is, wow, this feels like a bubble.

6:43 SK Hynix today trades at a four P/E (price earnings) multiple. It is not 2000 when Cisco was trading at 150 times earnings. People are obviously worried these companies are over-earning. They're going to need a lot of CapEx. But it does speak to the fact that there is still a lot of skepticism about what's happening. Now, we in this last year, given our enthusiasm, went out there and deployed a ton of capital. We did it in financing GPUs and TPUs.

7:14 We committed with Nvidia and Broadcom. We were big in neoclouds. We created a joint venture with Google with this company, Crux. We invested with Firmus in Australia, Neysa in India, a bunch of data center companies around the world. We made a huge commitment here. We think this is going to continue to pay big dividends because of that need for compute. And one more going to the past. We’re obviously focused on data centers.

7:45 We talked about chips, but there's more. It's not just data centers, of course, it's power. Last year we talked about The Graduate: not plastics, but power. That is so important in everything that's going to happen. Why is that the case? This is utility CapEx expenditures, $800 billion over the previous five years, expected to nearly double over the coming five years. I was in Canada this week. They were talking about spending hundreds of billions of dollars to expand and upgrade their grid.

8:20 It's going to happen around the world. The need for electricity is profound. And of course, the stocks here have reacted—not as much as the memory chip companies—but this used to be a very boring business. And now companies here who make turbines and cooling and electrical equipment go up 44% to 81% over the course of a year. We, fortunately, again, invested a ton of capital in this space. Sempra Infrastructure; we made a huge credit offering to these guys in order to finance a major LNG project.

8:58 Enverus, a data analytics business in energy we invested in. Williams to build utilities, to build power plants along with data centers they’re building. MacLean electrical equipment. In Europe, Eurowind and Sunotec, renewable companies. There is an enormous need for energy, and in our energy transition and infrastructure businesses, we have made enormous investments. And what's happening from all these AI related investments we've made?

9:31 We looked at the ten investments we've made across our entire firm that appreciated the most in the second quarter this year. And magically, nine of the ten are obviously in this AI space, including some of the model companies. Just one, Aster Care hospitals in India, is the exception. I would guess when you look in Q3 it will look very similar again. So what can we learn from the past, particularly the distant past?

10:02 Well, I want to take the clock back to September 17th, 1870. I was not around, of course, but I think a bit of history is helpful to think about what's happening today. Here is me, AI-generated me, doing my best American Gothic with the pitchfork and the straw hat. I would likely be living on a farm. This is where most Americans were at that point. This is post-Civil War America. Virtually every structure would have been made of wood.

10:37 I love that sound. If I was going to see my family, of course, in the neighboring town, I would take a horse. And if I was reading my investment committee memos at night, I would have a candle. So move the clock just 30 years and we are living in a radically different world in the United States for many. This is me looking like a well-dressed accountant in 1900 with the pocket watch. I'm much more likely to be living here now in an urban environment.

11:11 The structures, many of them, are made of steel. If I was going to see the family, I'm likely to take a rail car there. Starting to be a few cars on the road, and when nighttime comes, thanks to Thomas Edison, we get to read by light. I would contend that where we are today is much more akin to 1870, and that a major industrial-revolution- type change is coming for us. And what happened in this period of time, from 1870 to 1900, economically?

11:46 Well, annual labor productivity doubled, GDP went up fourfold, real manufacturing went up six times. And the stock market, since we're investors, went up seven times. I'm not arguing that's what's going to happen today, but it does show you the power of technological change, innovation, what it ends up meaning for economies and investments. And this is, I think, a powerful case study when you get this kind of change.

12:22 But even in a boom, there is plenty of bust. Back in the day, the key infrastructure for what was happening, of course, was railroads. There were 200 railroads. This is a list of them that went into bankruptcy or insolvency during that period of time. And this is the cautionary tale. Now, what's different today is back then they were building supply on a highly leveraged basis ahead of the demand, very similar to what we saw with telecom infrastructure.

12:57 They were anticipating what's to come. And of course, it didn't turn out well. Today we're seeing something much different. The demand is running well ahead of supply. Much of what is contracted, be it the energy or the data centers, is with companies who are very lowly leveraged as well. It doesn't mean there aren't going to be things that go poorly, but I think it is a different dynamic. We're going to continue to keep our eyes on this as we're deploying your capital.

13:28 So now I'm moving ahead 84 years, January 22nd. I'm about to have my 14th birthday. I'm at Elm Place Junior High in Highland Park, Illinois. I’m at this game. I'm watching, I wish I was there, I'm watching this game. The Redskins under Joe Theismann are playing Marcus Allen in the Raiders, and it's not much of a game, 38-9. But the highlight actually is two commercials. The first is this one, the famous ad from Steve Jobs, 1984.

14:06 We Shall Prevail. On January 24th, Apple Computer will introduce Macintosh, and you'll see why 1984 won't be like 1984. This was an amazing, amazingly prescient ad, right? What did it talk about? It talked about the fact that computing should be democratized. You shouldn't have to use COBOL even with a PC.

14:36 You shouldn't have to have MS-DOS and all these crazy commands. You should just be able to use a mouse at a click of a button, and you should have this unbelievable computing power at your fingertips. And if you think about the analogy today, here we are, where we're talking about intelligence being available to everyone in the world. You could be in the smallest village in the world, and you have access to the greatest minds in the world.

15:03 You don't need to be in Cambridge, UK or Cambridge, Massachusetts. I think this is a very, very powerful message, but in fact, I don't think it's the most relevant one for the discussion today. This is the one that I'm focused on from that same Super Bowl game. That is Clara Peller, 81 years old, who asks a very important question. It certainly is a big bun. It's a very big bun. Big fluffy bun. It's a very big fluffy bun.

15:33 Where's the beef? Where's the beef? And the question here is we're spending trillions of dollars of CapEx. What's happening here? Why is there happening? Is there beef? Is it just a bunch of circular financing? People are going to lose all this money? That is the question we're trying to get at. So I want to move to where we are today. And the reason why I have confidence in what is happening in all this investment. And it really starts with what's happening in our companies, which is AI use cases are rapidly expanding, and it's pretty much everywhere.

16:07 It's in process improvements: IGI, diamond grading in India. Phoenix Tower, which is a mobile tower company we have, that is now processing their leases five times faster. They invested $4 million in a new AI process, and they're earning $4.5 million a year. Having done that, 100%+ annual return. Software engineering, virtually every one of our companies getting benefits. I mentioned Enverus earlier. They now have a new process to fix code.

16:40 They're getting an 18x return on the dollars they're spending with the model companies. This is why the spend keeps growing. It's customer service. Tricon, a rental housing company, can process your application 90% faster. Think about how much better that customer experience is. And it's content creation. Advertisements, design renderings. It's happening everywhere here as well. But it is not just productivity. This is an ad for our company, Chamberlain, a garage door opening company.

17:17 Introducing the MyQ Secure View 3-in-1 Smart Lock with face access. Don’t just view it, MyQ it. Brand new product. This company was an analog business with garage door openers. They figured out a couple of years ago they could use AI visualization techniques at the front door. Certainly at the garage door for your Amazon packages. Their business now is a $40, $50 million run-rate business that they think in the next five years, $500 million of run- rate revenue, according to the CEO I met with just a couple of weeks ago.

17:55 We're going to see more and more case studies like this as we move forward. And it's not just at our portfolio companies. At Blackstone, it's in legal and compliance, where marketing reviews are 50% more efficient. Software development. My personal favorite, CIO portfolio intelligence agent. What are we seeing there? Timelines that are reduced by 99%. Pretty amazing. Now it's starting to show up in the numbers as well. Annual US productivity growth, which for a decade, you know, averaged 1.5%, is now 2.6% over the last two and a half years.

18:36 It's hard to say how much is AI-driven, but something is happening out there and we're beginning to see it. Productivity has made a step function move upward. The tip of the spear is the hyperscalers. Their revenue per employee has grown 65% over a three- and-a-half-year period as they adopt these technologies. And of course, it's showing up in margins. These are the EBITDA margins for the S&P and our portfolio companies, up 500 and 700 hundred basis points over this four-year period of time.

19:11 And if you talk to our CEOs, they would tell you the next 12 months are going to be very positive as well. And not surprisingly, it's leading to very powerful earnings growth. What had averaged 15%, the last 12 months, 32%, even stripping out one-time gains. I don't know if it'll continue at that high of a rate, but there is no question that there is a powerful impact as this technology diffuses into the economy.

19:42 Now, it's going to move. We talked a lot about our investments in power and chips and data centers, and right now the technology is mostly just at our desk. That's where we're getting the benefit. But as the agents leave the desk and go into the real world in robotics, autonomous vehicles, national security, space, all these different use cases, there'll be more and more demand for compute, more and more productivity gains. If you look at Waymo, the driverless car company, its mileage is up 250-fold over the last two and a half years.

20:16 Again, it's pretty amazing what's happening here. This is happening, of course, because of the experience and the fact that serious accident rates are 94% lower. So this, I believe, will continue in a very meaningful way. Now, there's a lot of debate about AI, the impact on our lives. There are going to be challenges. There's going to be transition for society. But I do think it's worth talking about some of the real positives. One is job growth.

20:46 We're seeing a blue-collar job boom at QTS, our data center company. We have seen a tripling over less than two-year period of time of construction workers at our sites. By the way, overall employment is growing at our companies despite all the productivity from AI. New business applications have doubled over the last decade. It is much easier to start a company when you have an army of agents at your side. And of course, science is the most exciting thing from this technology.

21:19 This was a study from Nature that shows that Phase 1 trial success rates have moved up meaningfully as a result of using AI-discovered drugs. AI is definitely better in chemistry, Nick Galakatos will tell you it still needs to work on its biology, but this is promising, and this is as well: trial duration. This was a recent study from McKinsey on one trial, where they brought down the time by 40% by optimizing with AI.

21:49 Our portfolio company Advarra in this space is doing the same thing. This is the one thing that is most exciting for all of us as humans, what it's going to look like for our health over time. Now the question is, where are we going in light of all of this that is happening? And I would just say we have three foundational beliefs. One, that these use cases will proliferate, that it will drive enormous productivity gains. And it happens in medicine and legal, in the physical world as well.

22:23 As we get these productivity gains, people use this AI more and more, this will, of course, drive exponential growth in intelligence demand. That's what's happening. We think it will continue to happen. And the real block here is the physical world. The chips, the power, the data centers. Why is that the challenge? Well, here is me standing at a data center site, an AI factory, facing a bunch of real challenges.

22:54 One, entitlements. Communities are concerned. Unfortunately, there's a lot of misinformation because you can build data centers with virtually no water use. You can give community benefits. You can drive down electricity costs with new power. Nevertheless, people have concerns. That's slowing things down, and it's happening around the globe. Moratoriums. We've seen that right here in New York State. We could see more of that again, making it harder to build. Power, getting a turbine from GE Vernova, get in line, 2030, 2031.

23:30 That is a real challenge. Chip availability, that empty storage is a good indicator of what it's like to try to get chips. Today, if you order them, it'll take a long time. Why does that shortage exist? Well, the hyperscalers have increased their CapEx nine-fold in five years, and the chip companies haven't even doubled it. Now, in their defense, they've been in a cyclical business. Things go wrong. So they've been hesitant. But that's resulting in a significant shortage of chips.

24:03 And then one really hits close to home for us, the enormous capital that is needed doing these deals, these AI factories, including the power and the data centers and chips. $55 billion for a gigawatt. Blackstone is uniquely positioned at scale, given our ability to understand what's going on, to marshal the resources needed. But this is another challenge. This is why the supply is not keeping up. Now we're going to go back in time, just one more time, and talk about something that we thought was really important last year.

24:37 The thing that keeps us up at night the most: underestimating technological disruption. So that's what we were most concerned about last year. This started happening in the stock market a few months later. These are examples of professional services, software, information services companies. Their stocks went down. In many cases, their underlying business was fine. But the multiple compressed because market participants were worried about what is the future of their business going to look like.

25:09 Much greater uncertainty. And it happened in the private market. As you know, PE software deals went down 66%. If you looked at big deals, it would be down even greater percentage. Now, I would contend that when technology comes, the outcome isn't going to be the same for everybody, all on the same sea. But different ships can weather the storm in a different way. These are six retailers who faced the onslaught from Amazon, beginning 25 years ago, as a general merchandise direct to consumer model.

25:44 Of course. What happened? Kmart, Sears, toys R us ended up in the dustbin of history. Walmart, Costco, TJ Maxx. They managed to thrive. They had a value proposition. They were able to weather this storm. I would contend that in white collar businesses, there will be similar dynamics. They'll be companies that have really vital systems of record. They have management teams who change the model from seats to outcomes, and they end up doing much better. But there will also be plenty more of this.

26:15 And in the meantime, these companies will probably have lower multiples because of the uncertainty quotient. So there are also other winners in this world who are not technology related. Scarce assets, we'll call them. These are all companies, assets we’re invested in. A cricket team in India. Why do we like that? Well, there are 1.4 billion people in India. Cricket is their favorite sport. There are just ten teams and AI is not taking this away.

26:46 We have a great coffee chain, 7Brew. That's their Blondie, their signature drink. People are going to still need this. Still love these products. Beachfront property. This could be one of the best areas to go. Nobody loves real estate. Multiples are low, and yet these experiences aren't going away. These assets are irreplaceable. Or something like the Rome airport, where that infrastructure is essential. These are things that are interesting to invest in. Now, what are the greatest risks to this?

27:17 I would say a couple of things. First, cyber safety. This is obviously super topical. We're all focused on this. Something happens with a financial institution, critical infrastructure, that is likely to lead to a political dynamic, not just here in the US, but around the world. One thing I would say on this, if we see regulatory responses that slow the frontier, that ultimately is not going to change the diffusion that needs to happen because the technology today is already so powerful.

27:51 But obviously this is a very important risk. Evolving technology. These are data centers in space. We could talk about edge computing on our phones. I would say on this, all of this is likely to happen. But given that demand curve, all of the above, it's a little bit like power. You're going to need nuclear, natural gas, renewables. I think this technology comes, but terrestrial data centers for now are going to continue to be very important. Geopolitical tension, 90% of the advanced semis are made in Taiwan, a very politically sensitive area.

28:28 And of course, excessive valuations. If you think about, you know, companies that have barely started, no revenue with $10 billion values, that obviously gets you concerned. There are some defense technology companies at very high values. You've got to be mindful of what's happening in this area because your capital is so important to us. We want to get this right. This is what we're discussing each and every day as we make investments. That's part of the reason you see us focus on that seniormost part of the stack, the compute, where we have the most confidence.

29:02 So to finish off here, where is the beef in all of this? Well, we know what the bun is. It's chips and data centers and power in the trillions of dollars we're investing in this area. The beef is the return on investment This is really the heart of the matter. This is the return. We're going to continue to see these use cases go up. We're going to continue to see demand grow. And that is justifying this enormous investment, which is actually, I think, going to constrain the usage of AI over time.

29:39 So we finish here with the same slide as last year, our north star. What we say it's all about returns. From when Pete and Steve founded this firm 40 years ago, it's always been about delivering for clients. Steve reminds us of that every day. This is our focus, to deliver for the folks in this room. We remain committed. It's why we write the memos. It's why we read the memos on weekends. It's why we're spending so much time thinking about where the world is going.

30:11 So with that, I take my 1900s goodbye here with the double thumbs up. One value that is certainly timeless is gratitude. We have enormous gratitude for all of you in this room. I hope you enjoy the rest of the day, and thank you again.