Let's start with the big beat. Really strong numbers this quarter, John, just how much of that can you give us? An idea was powered by an AI wave?
Well, it definitely was a heck of a quarter, Danny. And 26% up in earnings on top of 25% last quarter. And it was definitely powered by that strategic decision we made to lean into AI in the AI ecosystem. It really started for us back in 2021 when we bought QTS, this big data center business, and it gave us a front row seat in terms of what's happening around AI. And we made a strategic decision, hey, we should lean in here, not just with data centers, but with neo clouds, large language models, electrical equipment, energy. And that decision is paying off and it's delivering for our clients, which is the most important thing. And what's also very exciting is we're not done here. Just in the second quarter, we announced partnerships with Google and TPUs with Broadcom and financing chips with Anthropic in deploying their technology. We continue to see a lot of opportunity in this space to generate favorable returns. We announced a couple of big sales in the last few weeks, one of data centers for $8 billion. We announced yesterday a battery storage company which sold for $7 billion. So that ecosystem and our strategic pivot has really made a difference for the firm and our investors.
Just on that pivot, Jon, I was really struck last month. You were at a Morgan Stanley conference, and you made the case that Blackstone is one of the least expensive ways out there to play AI. Today, Blackstone, of course, trades more akin to financial list multiples that look like that. Do you think Blackstone should be trading at tech multiples? Should you look more like an Nvidia or an Alphabet?
Well, what I would say about Blackstone is we're a company that operates with virtually no capital. We don't have any insurance liabilities. We have virtually no debt. We've shown incredible growth. Look at our earnings in the first half of the year. Today we're yielding 4%, which is one of the highest dividend yields in the S&P. And yet when we look out over the horizon we see real growth potential. We see it with our institutional client, with our insurance companies and with individual investors. And today we're trading at a discounted market multiple. Yeah, we see it very differently. But I think as the market begins to see the earnings power of this firm and what we can deliver as we grow, as we generate these great returns from this strategic pivot we've made, I do think the market will change things. That rerating will come. Sometimes we've got to be a little patient. The key for us is just keep executing.
I guess what I'm trying to understand is just to what degree Blackstone has been made into not just an asset manager, but a pure AI play. And, John, a lot of people might look at that and get concerned that you bet the house on one theme. How do you assuage investors who might be concerned about just concentration risk in AI bets?
Well, we've got a big diverse firm. Obviously, we've got a very large real estate business. We invest in secondaries. We're in the hedge fund space. We have all different forms of credit out there. Our private equity business owns lots of regular way businesses like Jersey Mike's. So there's more to this firm than that. But I think the decision here to lean in when you see this huge shortage of compute that is out there and an opportunity to earn favorable returns by aggregating capital at scale, having terrific platforms. To me, that makes a lot of sense. So there's more to this firm, by the way. It's a global firm as you know as well, with big strength in Europe, in India and Japan and around the world. But I feel very good about what we've done in terms of this concentrated focus, because this really is a new operating system for the world, and you want exposure to it. It would be a little bit like, hey, I don't want to own the S&P 500 because there's big exposure to technology growth. And I think for us and our investors, we want to deliver that premium return.
By the way, there has started to be somewhat of a market push back on just the amount of CapEx that's being spent by the hyperscalers. Many hyperscalers we've partnered with. You can see it in Alphabet results this morning, fears over just that big chunky number they announced in their spending. If those hyperscalers do start to pull back because of the market push back, how does that impact Blackstone's investments and deployment?
Well, that's not what we're seeing today. In fact, it's the opposite. If you look across our data center platforms, we leased one gigawatt in 2024. We leased two gigawatts in 2025. And this year, we're on pace to lease at least seven gigawatts. We believe, which is more than 100 billion of data centers and a couple multiples of that in terms of chips that go inside. If there is a slowdown, then yes, the pace of development would slow. The good news is you're not speculatively building. And I think that's really important when you think about the telecom boom in the 90s where people built fiber, something like 20 times more than was actually needed. In this case, we're only building to demand and long term contracts from very large creditworthy companies. That's really the focus for us. And to us, it seems like a disciplined way to do this. I think fundamentally what people aren't getting is how powerful this technology is at companies. We're beginning to see real time examples, certainly in customer service and coding. It's now starting to move to the revenue generation side where businesses are creating new lines, new revenue streams using this AI. And I think because this tool is so powerful, there's an underestimation of what's happening. We see the power of it and we're really focused on helping meet this.
John, because you use the tool every day too, both at Blackstone, I know, and in your portfolio companies, and you have partnerships like Anthropic. I got to ask, though, there's a lot of fear and perhaps excitement around Chinese open source LLMs that they're cheaper, they're more efficient, and they're just as good as the most advanced models. Have you guys tinkered around with any of them at Blackstone, and would you ever use a Chinese open source model?
