Skip to playerSkip to main content
  • 2 days ago
Transcript
00:00And the Financial Times reporting that Carlyle is set to make a more than five-fold return
00:05on the sale of a $2.6 billion data center power platform to EQT.
00:11Joining us now is Pooja Goyal, Chief Investment Officer of the Global Infrastructure at Carlyle,
00:16the woman behind this deal.
00:17Pooja, thank you so much for joining.
00:18And look, the reporting suggesting a more than five-fold return on this investment,
00:22I mean, that's huge.
00:23I feel like two times is even really widely celebrated in the world of infrastructure,
00:27let alone five.
00:28So you started building this thing in 2021 when AI was barely like a glimmer in our eye.
00:34Did you have any idea that this thing would grow to the behemoth and the attractiveness
00:37that it eventually did?
00:39Well, first of all, Dani, thank you for having me here.
00:42You know, we have a very thesis-driven approach to investing in infrastructure at Carlyle.
00:47And back in 2020, we had a thesis that there was going to be significant demand for power
00:52generation capacity in the United States.
00:55And so instead of investing in an existing platform at a bloated valuation, we decided
01:00to build our own platform from scratch.
01:02We started with four employees, a year one budget of $3.5 million.
01:07Obviously, you mentioned the purchase price right now, $2.6 billion.
01:11And we decided to launch Copia Power to develop, build, own, and operate power generation capacity
01:18in the United States at scale, but yet do it in a profitable and commercial manner.
01:24By the way, $3.5 million, that feels like a grain of sand on the beach compared to the
01:31numbers that companies are now spending building up massive power platforms.
01:35I wonder if that made sense at the time to kind of slowly walk this thing up and not throw
01:40a bunch of money behind it.
01:42And now you have to be investing lots of money.
01:43Or can you repeat this feat?
01:45Can you build something from the ground up yet again?
01:48So you're spot on.
01:49Infrastructure is a very capital-intensive asset class.
01:52When we started developing the power generation assets for Copia Power, we decided to be very
01:58strategic.
01:58And we decided to build what we call our energy campuses.
02:02So we said, let's retain the optionality to co-locate projects that could one day use the
02:09power that Copia would generate.
02:11Now, five years back, we thought that those projects would probably be manufacturing plants
02:16or industrial processes that need electricity.
02:19It obviously turned out to be data centers and the demand for compute on the back of all
02:24of this growth in AI.
02:26What about the rate environment?
02:28Because in 2021, we were hiking rates too, which makes infrastructure more complicated.
02:32And maybe we're kind of in a similar environment.
02:33We're not necessarily going to get hikes, but there is a concern that we will from this
02:37Federal Reserve.
02:38How are you thinking about building these platforms in that type of environment?
02:41So that is spot on.
02:43And that is exactly what we were very focused on when scaling up Copia.
02:47Like I said, we were focused on building a commercial and profitable platform.
02:52Not all growth is accretive when your risk-free rate is going up by 400 basis points.
02:57And a lot of the existing platforms at that point in time were doing what I call everything
03:01everywhere all at once.
03:03And we said, let's pick the locations where we can drive scale, but make sure that that
03:09scale is accretive given the interest rate environment.
03:12So you were early on this and early building out power capacity, but it feels like a trade
03:15that everybody is getting into.
03:17I mean, you hear of like ex-politicians getting in, leasing land from somewhere, trying to build
03:21things up.
03:22I wonder what you make of it, because clearly the demand for energy around AI is very real
03:27and very necessary.
03:28But on the other hand, is there a real risk that we might see some stranded assets?
03:32Yes.
03:33So look, when you are building infrastructure assets, especially those kind of data center
03:37assets that require tens of billions of dollars of capital, you have to be very thoughtful
03:42about your capital allocation decisions.
03:45We like front of the meter grid connected projects.
03:48You do not want to build islands or assets that could be stranded on a long term basis.
03:54So investors, both on the credit as well as equity side of the equation, need to be extremely
03:59deliberate when it comes to making these capital allocation decisions.
04:03So you've done the deal.
04:04What do you do with the proceeds now?
04:06Well, so the deal is still subject to regulatory approval because these are energy assets.
04:11So, you know, the deal has been signed and we expect to apply for those regulatory approvals.
04:17But we believe that investing in AI related infrastructure continues to be a very compelling investment
04:23opportunity.
04:24So we continue to see very attractive ways to deploy capital.
04:28The one thing to note is what we do at Carlisle is mid cap infrastructure that enables us to
04:34actually innovate, get creative, build platforms from scratch, or partner up with founders relatively
04:40early on in their company's investment life cycle.
04:43So we are able to originate deal flow that is differentiated from the kind of deals you
04:49might typically read about.
04:50How much of that is also just an acknowledgement that what exists out there now, maybe large
04:54cap, maybe some of the mid cap now, when it's not early, that valuations are maybe too
04:58frothy to be attractive?
04:59I do think that there is a fair amount of exuberance that's out there.
05:04So again, people need to be very discerning when making these capital allocation decisions.
05:09I would say that exuberance is not just on the equity side of the capital stack.
05:13It is also potentially on the debt side of the capital stack.
05:16So when you're investing in assets, be deliberate about understanding stranded asset risk.
05:21When you're investing in companies, be mindful about how much growth you're underwriting,
05:26especially given the current macro environment.
05:28Stranded asset risks.
05:29What are the other major risks that always go along with these kinds of investments?
05:33And when you're opening the books to somebody like EQT, what do they go and look at first?
05:37What is the main thing they want to know?
05:40So with infrastructure assets, these are long duration assets.
05:43These are not assets that you value only over a two to five year time horizon.
05:48The ultimate value of your company or your platform is going to be primarily driven by
05:53the value of your assets.
05:54So in addition to making sure they're not going to be stranded assets, you do want to take
05:59into account the quality of your revenue contracts.
06:02So your customers, in the case of data centers, it would be the leases with your hyperscalers.
06:07In the case of power generation assets, it would be the credit quality of the utility or the
06:12end customer that you're selling power to.
06:14And also you want to make sure that those projects have been prudently financed as well
06:19as built.
06:20Technology obsolescence can be a consideration.
06:22And you don't want to overlevel these assets as well because you want to get the cash coming
06:26off of these deals.
06:27We were speaking just quickly here.
06:29We were speaking with Christina Hooper of Man Group who said there is a real risk that some
06:32of the hyperscalers stop spending as much as they are.
06:35What happens to this whole ecosystem if you do see a hyperscaler pull back in less than
06:40a minute here, Pooja?
06:41So this is a completely different business model for hyperscalers.
06:45They've gone from having companies that are fairly acid light to fairly acid intensive.
06:50And I do think that you might hit some roadblocks where projects might get delayed and move to
06:56the right.
06:56But not all projects are created equal.
06:59And I do continue to believe that the fundamental demand for compute is still very strong.
07:03So you will see growth in the long run.
Comments

Recommended