For all the talk of AI bubbles and market crashes, Scandinavian private-equity giant EQT says it’s making a bet that regardless of what happens with the frontier labs, AI is still going to need a massive amount of power, energy, chips, and data centers.

“All of a sudden, we find ourselves with $100 billion of energy assets, which are really the limiting factor for AI right now,” said Jean Salata, global chair of EQT, which has $400 billion in assets under management.

“A lot of people went out and signed tons of contracts with anybody — they call them ‘bragawatts.’ Butg something and delivering it on time, and with power connected, are two very different things,” Salata said. “Sophisticated investors understand that, but there’s a lot of things happening in the public markets where people are investing in things that they really don’t understand.”EQT is a unique animal in an industry where exits have dried up, a generational bet on software companies looks vulnerable to AI, and higher interest rates have broken the math of cheap leverage and multiple expansion. Since 2018, PE firms have asked their investors for $1.5 trillion more capital than they’ve given back, and the industry is sitting on nearly $4 trillion worth of businesses it can’t sell.

EQT has largely stayed out of the private credit and insurance boom that has turned its US rivals into asset-gathering giants — a “beta business,” said Salata, an artisanally crafted Wall Street dig. And it has an unusual backstop: The Wallenbergs, Sweden’s most powerful business dynasty, helped found EQT in 1994 and remain its largest shareholder. The family’s patient approach to ownership is close to the long-hold model that much of private equity is now trying to reinvent.

Liz Hoffman: For a long time, private equity firms could borrow cheaply, ride a rising market, and look like geniuses. It feels like the jig is up, and I know you have some thoughts on that.

Jean Salata: Our industry needs to remember what we are meant to be doing, which is delivering returns for our clients. If we’re not doing that, we’re going to end up in a very difficult place. There’s been a lot of attention on the growth in assets under management, a business model that has really benefited from the tailwinds of the private credit run of the last five years. But we’re in an environment where the tide is going out and that means you’re going where there is real value being created. There are something like 10,000 private-equity firms out there. What should that number be?

The industry already is consolidating and that will accelerate. This year the top 10 firms in the world have raised 50% of the capital. When you raise smaller funds, people don’t see the carry coming in and so they leave and you get zombie funds.

EQT has chosen not to chase those credit tailwinds you mentioned. It doesn’t have the big lending business that Apollo, Blackstone, and others have ridden to huge AUM piles. Do you feel good about that decision?

I think there’s a role for private credit, but the question is who’s best positioned to deliver that? We’re good at being an equity investor, being really close to the sectors that we invest in, bringing the best operational talent, bringing the technology and the AI capabilities that we have to the company. None of those are really relevant to the credit business. It’s an origination business. It’s a beta business. Best case scenario, you make 8% return on your money. So, the answer is yes, I think we are happy with the decision. But how do you resist the pressures to be a one-stop shop?

Eventually there is a feedback loop. Second-quartile [performance] is, maybe, good enough. But when you start talking about third- and fourth-quartile returns, clients will either stop or reduce the amount that they invest with you.

It’s hard to take anything public right now that isn’t a giant AI company and private equity’s track record on returning capital has been abysmal. How does the industry get out of this funk?

There are a lot of assets for sale that need to find a home and there aren’t that many buyers. At EQT, we’ve been able to take advantage of where there are pockets of opportunity and liquidity. We had $33 billion of distributions to our investors in the last 12 months, and as a result, we’re continuing to attract capital.

There’s this strange incentive structure in our industry, where you sell businesses that you otherwise wouldn’t want to sell because of this finite [fund] life. There’s a bit of frustration on the part of sellers that have sold assets because they need to go fundraising, only to see the buyers make like another 3x on it, which is why you’re seeing these longer-hold vehicles like evergreen structures, open-ended structures, continuation vehicles.

But then isn’t the market saying that the 10-year closed-end fund is outdated and should probably go away?

One of the reasons we bought [secondaries firm] Coller is we think there is a convergence happening between the primary market and the secondary market. There’s a case to be made for the secondary market essentially becoming the private equity market, as it is in the world of listed equities. Remember, the stock market is a secondary market. Yes, you have some IPOs, but the vast majority of the trading is in existing shares. Where we’re heading in private markets is not quite that extreme, but it’s going to be a much bigger part of the industry. There were $250 billion of PE secondaries deals last year. That’s doubled in the last three or four years and it’s probably going to double again in the next five.

One other tension in PE right now is that everyone seems to be in the same trade — the AI trade. Are you worried about correlation?

There’s no patents in finance. We’re all moving in a similar direction and it’s going to boil down to execution. A lot of this is about talent. There’s a shortage of really good talent that understands technology and AI, and the ability to attract, to afford, that talent becomes a question of scale. Tell me why all this AI investing is not going to end terribly.

There’s a lot of noise right now about data centers — ‘it’s a bubble, the market’s crashing.’But if you look at the fundamentals of what’s going on in the AI buildout with power, energy, chips, data center demand, we think it’s very fundamentals-based and a very, very attractive opportunity. I don’t want to say we stumbled onto it — the team deserves more credit than that — but we found ourselves in the middle of this in a big way because we were investing in power assets and in data center assets for cloud migration. The pre-AI generation of data centers.

Yes. So all of a sudden, we find ourselves with $100 billion of energy assets, which are really the limiting factor for AI right now. Now we are building the industrial AI core for the next generation of data centers, but we’re still only working with the hyperscalers. This is an important point; 92% of our contracted revenues in our data centers are with the hyperscalers. You’re not a fan of the neocloud credit?

You’re invested in Mistral, the big French AI lab. Is it important that Europe has a frontier LLM champion?

It used to be that capital was driven primarily by market forces. But now politics and policy are shaping the way capital is flowing in the world. So yes, there are going to be different ecosystems — obviously there’s a US ecosystem and the Chinese ecosystem. The European ecosystem, which logically would have been more part of the US ecosystem, but I think that trust has been undermined. People are feeling like they need to be more self-sufficient.