T. Rowe Price's Mark Stodden analyzes AI infrastructure financing, highlighting debt shifts and collateral risks
Summary
Mark Stodden, a credit analyst at T. Rowe Price, discusses how hyperscalers like Amazon, Microsoft, and Google are shifting from internal cash flows to debt markets to fund massive AI capital expenditures. He outlines three primary funding channels: hyperscaler corporate bonds, data center SPV bonds, and emerging chip-backed debt structures. Stodden warns that chip-backed debt faces significant collateral risk due to the short 3-5 year economic life of GPUs, necessitating amortizing structures and credit enhancements. He also draws parallels to the 1990s telecom boom, noting that rapid technological advancement can render physical assets obsolete. Finally, he highlights distress signals in smaller 'neo-cloud' providers facing high leverage and rising rates.
Key takeaways
- 1 A one-gigawatt data center costs approximately $15 billion to build (shell and power) but $30-$45 billion to fill with chips, meaning electronics represent the majority of capital expenditure.
Unidentified speaker (speaker_3) · 5:59
- 2 Data center SPV bonds offer attractive carry in the 7%+ range with 15-20 year durations, supported by contractual protections regarding construction phases, tenant quality, and lease terms.
Unidentified speaker (speaker_3) · 10:05
- 3 residual value guarantee from the chipmaker, which will cover maybe a first loss position. You'll have subordinated tranches in the deal that'll absorb a loss and equity in the deal that will absorb loss. And then you'll have a layer of contractual protections from the off-taker. And by that I mean, you know, the chip will be installed and then it will generate tokens, right? It'll it'll process information and someone will lease that chip. That could be anthropic, open OpenAI, AI, or Amazon or Microsoft. The credit quality of the off taker can actually offer a lot of credit enhancement as long as the contract is really tight and they can't get out of it, right? So a combination of all those support factors can get you to a finable structure. It will not be simply the collateral and the lease, like an aircraft. a lease like that, because aircraft have a very established secondary market.
Unidentified speaker (speaker_3) · 13:56
- 4 Smaller compute providers ('neo-clouds') are showing signs of distress due to rate pressures and highly leveraged balance sheets, creating volatility that is spreading from high yield into the lower half of the investment grade market.
Unidentified speaker (speaker_3) · 20:45
Episode sections
- 0:02 – 3:19
Intro and Guest Background
Host Stuart Foley introduces the episode topic of financing AI infrastructure and welcomes guest Mark Stodden, a credit analyst at T. Rowe Price with prior experience at Moody's and ATT.
- 3:19 – 8:08
AI Infrastructure Financing Models
Stodden explains that hyperscalers like Amazon, Microsoft, and Google have shifted from using internal cash flows to issuing debt and using off-balance-sheet structures to fund rising CapEx. He notes that chip-backed debt is a nascent but potentially large segment where GPUs serve as collateral.
- 8:08 – 12:34
Investment Strategy Across AI Debt Structures
Stodden outlines three funding channels: hyperscaler corporate bonds (attractive yields but difficult timing due to supply), data center SPV bonds (attractive carry and contractual protections), and chip-backed debt (currently difficult to assess due to short asset life and lack of precedent).
- 12:34 – 16:32
Collateral Risk and Amortization
The discussion highlights the risk of technology becoming obsolete before debt is repaid. Stodden explains that chip-backed deals must be amortizing to match the short economic life of GPUs (3-5 years) and rely on credit enhancements like residual value guarantees.
- 16:32 – 22:05
Compute Cost Decline and Market Dynamics
Stodden discusses Jevons paradox, where falling compute costs drive demand but compress margins for providers. He notes that large, low-cost providers like Amazon are better positioned than smaller 'neo-clouds,' which are showing signs of distress due to high leverage and rising rates.
- 22:05 – 27:05
Portfolio Concentration Risk
The speakers address the difficulty of identifying hidden concentration risk when portfolios hold both direct issuer debt and off-balance-sheet SPVs backed by the same underlying tenant, such as Meta.
- 27:05 – 33:32
Telecom Infrastructure Lessons
Stodden draws parallels between the current AI build-out and the 1990s telecom fiber boom, warning that rapid technological advancement can render physical assets obsolete, though he notes differences in asset life and financing structures.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| AmazonAMZN | equity | mixed | So the big hyperscalers like Amazon and Microsoft and Google, et cetera they began to shift their capital allocation. |
| MicrosoftMSFT | equity | neutral | So the big hyperscalers like Amazon and Microsoft and Google, et cetera they began to shift their capital allocation. |
| AlphabetGOOGL | equity | neutral | So the big hyperscalers like Amazon and Microsoft and Google, et cetera they began to shift their capital allocation. |
| NVIDIANVDA | equity | neutral | And you've seen some announcements in the press from the semiconductor OEMs like NVIDIA and Broadcom that are working to establish deep funding pools for basically chip-backed borrowing. |
| Meta PlatformsMETA | equity | neutral | Example, Meta has a very large data center in Louisiana that they are building a $30 billion building, two gigawatt data center, and they're partnered with Pimco and Blue Owl to fund that. So it was a lot of headlines on that transaction. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Hyperscalers like Amazon, Microsoft, and Google began reducing buybacks and leaning on debt markets to fund capital deficits exceeding internally generated cash flows around the beginning of 2026.
Unidentified speaker (speaker_3) ·3:59
How to verify: Review Q4 2026 and Q1 2026 earnings reports and capital allocation statements for Amazon (AMZN), Microsoft (MSFT), and Alphabet (GOOGL) to verify changes in share repurchase programs and debt issuance volumes. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Primary sourceA one-gigawatt data center costs approximately $15 billion to build (shell and power) and $30-$45 billion to fill with chips.
Unidentified speaker (speaker_3) ·5:59
How to verify: Verify against industry reports from data center operators or construction firms, or check capital expenditure guidance in earnings calls of major hyperscalers discussing facility costs. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
About 60% of CapEx in a data center or hyperscaler goes towards buying chips and associated hardware.
Unidentified speaker (speaker_3) ·5:39
How to verify: Check the breakdown of capital expenditures in the quarterly earnings reports of major hyperscalers (Amazon, Microsoft, Google, Meta) for the period 2026-2026. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Data center SPV bonds are trading in the 7%+ yield range with 15-20 year durations.
Unidentified speaker (speaker_3) ·10:05
How to verify: Review bond market data from Bloomberg or ICE BofA for data center real estate investment trusts or special purpose vehicle bonds issued in 2026-2026. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Only one chip-backed deal had been syndicated in the investment grade market at the time of the interview.
Unidentified speaker (speaker_3) ·11:07
How to verify: Search bond issuance databases (Bloomberg, Reuters) for 'chip-backed' or 'GPU-collateralized' bond issuances in the investment grade sector prior to November 2026. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Meta is partnering with Pimco and Blue Owl to fund a $30 billion, two-gigawatt data center in Louisiana.
Unidentified speaker (speaker_3) ·23:47
How to verify: Verify through press releases from Meta, Pimco, or Blue Owl, or SEC filings (Form 8-K or 10-K) related to real estate investments or joint ventures involving Meta and these asset managers. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Primary sourceLong-distance phone calls cost 28 cents per minute when Mark Stodden started at ATT in the 1990s.
Unidentified speaker (speaker_0) ·29:29
How to verify: Verify historical telecommunications rate data from FCC records or historical news archives regarding ATT pricing in the early 1990s. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Neo-clouds are showing signs of distress due to rate pressures and highly leveraged balance sheets, with volatility spreading into the lower half of the investment grade market.
Unidentified speaker (speaker_3) ·20:45
How to verify: Review credit spreads and default rates for high-yield and lower-grade investment grade bonds issued by cloud infrastructure providers in the period leading up to November 2026. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
GPU economic life is approximately three to five years.
Unidentified speaker (speaker_3) ·28:55
How to verify: Verify against semiconductor industry lifecycle reports or technical specifications from chip manufacturers like NVIDIA and AMD regarding hardware refresh cycles. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Fiber optic cables have an economic life of 20+ years.
Unidentified speaker (speaker_3) ·28:55
How to verify: Verify against telecommunications infrastructure standards and depreciation schedules used by major telecom operators for fiber assets. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Do your own research
- Verify hyperscaler CapEx guidance in their latest 10-K or 10-Q filings to confirm the shift from buybacks to debt financing.
- Review specific data center SPV bond prospectuses for contractual protections on construction phases and tenant quality.
- Analyze GPU depreciation schedules and residual value guarantee terms in early chip-backed debt deals to assess collateral risk.
- Monitor interest rate trends and their impact on the leverage ratios of smaller 'neo-cloud' providers.
- Compare current AI infrastructure financing structures to historical telecom fiber boom financing to identify similar obsolescence risks.
- Check primary dealer reports for supply and demand dynamics in hyperscaler corporate bond issuance.
- Assess the credit quality of subordinated tranches in chip-backed debt structures against senior secured debt.
- Review Moody's or S&P ratings for any recent changes in the credit outlook of major hyperscalers or data center REITs.
Full transcriptexpand
0:02Unidentified speaker (speaker_0): Welcome to another edition of the Insurance AUM podcast.
0:06Unidentified speaker (speaker_1): AI is gonna transform how we do business.
0:08Unidentified speaker (speaker_2): You don't learn much from funding a transaction. You learn everything from being repaid.
0:12Unidentified speaker (speaker_3): I think it's a good time to invest. We have an unsure inflationary environment. We have an unsure monetary environment. There are positive times and there are negative times. The market just needs less volatility.
0:23Unidentified speaker (speaker_1): Risk is increasing and rewards are decreasing.
0:26Unidentified speaker (UNKNOWN): My
0:26Unidentified speaker (speaker_0): name's Stuart Foley. I'll be your host. Hey, it's a beautiful day in Fredericksburg, Texas, 83 and sunny, and this is the home of the world's smartest money, the insuranceAUM.com podcast. My name's Stu Foley. I'm very glad that you're with us today. Everybody wants to know the future of artificial intelligence, but somebody has to finance it. Somebody's got to finance the data centers, the chips, and the power to run it all. And this is the interesting question: what happens when the technology becomes obsolete before the debt gets paid off? Today's episode is entitled Funding the AI Infrastructure Build. And today I'm joined by Mark Stodden, credit analyst at T Row Price, where he covers the technology sector and leads a team of 16 credit analysts in the US and the UK. Before joining T Row Price, Mark spent eight years at Moody's and four years as a buy-side equity analyst. So this is what makes his background particularly interesting for today's conversation. He actually began his career as a planning engineer at
1:29ATT in the 90s. And Mark, everybody I know that was at ATT TNT in the 90s, like retired with a mansion. But uh we can talk about that later. I'm thrilled that you're on. Welcome to the show.
1:41Unidentified speaker (UNKNOWN): Thank
1:41Unidentified speaker (speaker_3): you, Stuart. It's great to be here.
1:43Unidentified speaker (UNKNOWN): Plus,
1:43Unidentified speaker (speaker_0): you have a great podcast voice. So this is great. Uh so where did you grow up? And what's a fun fact that your colleagues at T-Row would be surprised to know?
1:53Unidentified speaker (UNKNOWN): Uh,
1:53Unidentified speaker (speaker_3): I grew up in Pittsfield, Massachusetts, which is in uh western Massachusetts and the Berkshires. Uh, fun fact about me. Um I don't know. Um I did the Iron Man Triathlon in 2017. It was a big accomplishment and I have I have not done triathlon since, but uh that was uh something I I'm proud of. So like maybe that's something interesting to add.
2:17Unidentified speaker (speaker_0): I'll never forget the first time I was sitting in Imperial, Missouri and the w ABC's wide world of sports showed Dave Scott in a triathlon. And I was trying to wrap my head around that swim, that bike, and that run back to back. I mean, like I had never conceived of something like that. So hats off to you for sure. Thanks. So the conversation starter is who's gonna going to pay for all this? And and the scale of the AI infrastructure build out. is extraordinary, I think, by any measure. We're talking about the data centers, the semiconductor capacity, networking, power generation, and all the infrastructure that makes AI work. How is the AI infrastructure build out being financed today? And how do you see the funding model evolving as the capital requirements continue to grow? And in we've got a little bit of news in the market today. If you can wrap some of that in there, that would be super helpful.
3:19Unidentified speaker (speaker_3): Well, this journey began, I I think it started to accelerate maybe like a year ago, year and a half ago, where the scope of the investment started to come into focus. And what we observed, and this was like on a almost like a quarterly earnings cadence, is that the big players in AI were spending a lot of CapEx and their forward view of CapEx was rising at almost every opportunity for them to talk about guidance, they would increase the outlook. And became clear quickly that this number was going to exceed their internally generated cash flows. So the big hyperscalers like Amazon and Microsoft and Google, et cetera they began to shift their capital allocation. They started to reduce their buybacks, and they started to lean on the debt markets to fund the cash deficit. So into the kind of beginning of this year, we had this like this transition of these very lean
4:19net cash hyperscaler businesses transitioning to using debt and leverage, albeit at very very low relative to their scale levels. We also s have seen capital formation through off balance sheet structures where a lot of the hard assets for the the data center build out, kind of like the concrete and steel part of it, are being financed as real estate investments. And these are like special purpose vehicles and joint ventures that will lease the data centers to the compute operators like Microsoft or Meta. And that kind of channel has been growing throughout the year. It's gotten a lot of attention. And these these are very kind of complex, bespoke bonds that can be very big ticket numbers. And you know, they're sold to sophisticated investors. who can really dig into the terms and conditions. And the next kind of phase of this is you've probably seen in the news or heard about the chip-backed funding, which is a collateralized borrowing that uses the GPUs, which are the specific accelerator chips in the data centers, as collateral. And so the chip-backed chip back debt market is kind of the most
5:39nascent and could be the biggest segment when we're all said and done. About 60% of the capex in a data center or of a hyperscaler goes towards buying the chips and its associated hardware. And so kind of big picture a one gigawatt data center, which is a very large facility, costs about $15 billion to build. the shell and the and get the power. And it costs about $30 to $45 billion to fill it with chips. So significantly more to fill the data center with all the electronics than it is to build the structure itself. And so that last channel of financing the chip backed debt market is the one that's getting a lot of attention lately. It's going to be a very short duration market because the chips have a very short economic life. And it's going to have we're going to go through the you know the pricing and the relative value kind of finding the clearing level on that over the next few months.
6:43Unidentified speaker (speaker_0): And are you is that to say that if I'm a debt holder and the issuer defaults? that I'm literally gonna repo the chips? Is that my col that's my collateral?
6:56Unidentified speaker (speaker_3): Yeah, there'll be a servicer in the middle who will do that and there'll be, you know, some rules as to how that process gets administered, similar to other securitized debt markets like rental cars or aircraft or you know other more seasoned markets. So the mechanisms to do this exist. We're just going to apply those rails to a new kind of underlying asset type.
7:27Unidentified speaker (speaker_0): Where my brain went was like exactly where yours was, which is the shelf life on a chip isn't the same as the shelf life on a caterpillar tractor.
7:36Unidentified speaker (UNKNOWN): Right.
7:37Unidentified speaker (speaker_0): And so but anyway that's a another day. One of the things that makes this interesting for our audience is that AI infrastructure is not a single investment opportunity. You can look at the at the company, and you kind of touched on this, the the actual structure itself, the chips inside and so forth. How do you think about the different ways investors can play this m this infrastructure build build-out, out and how does it change the risk profile as you move across those structures?
8:08Unidentified speaker (speaker_3): Well, I think a lot of this is it's very dynamic. So I'll start with the first funding channel, which is kind of the traditional plain vanilla corporate bonds issued by the hyperscalers. So use Amazon as our example, double-A double A issuer that is because of the supply. they're pumping into the market, they're printing bonds at single A or maybe even triple B spread levels. So this is a market technical that is flowing through pricing. Like the supply of bonds is driving yields and spreads wider. You can buy a 30-year Amazon bond that has a yield in the high. sixes today, which is a really attractive for a double AA type of credit. But timing that market has been extremely difficult. And the ticket size that they show up when they want to issue more debt is very disruptive because the scale of these companies is just so disproportionate to the credit market, right? Like uh Amazon has a multi multi-trillion dollar market cap, but they have a hundred or so billion of bonds outstanding. So it's like a twenty to one difference between the equity cap structure and the debt cap structure.
9:25Unidentified speaker (UNKNOWN): For
9:25Unidentified speaker (speaker_3): them, twenty-five billion is not a big number, but for the debt market, it's a lot to take down in one deal. So it's been very challenging to have a an accurate view of when they're going to show up and print another $25 billion. bond. Because you don't want to buy before they issue, you want to buy after. We've ended up kind of taking more of a neutral stance on that trade because we acknowledge those yields are attractive for the credit quality, but the technical has become too difficult to get right. On the data center piece, we actually think this is a very attractive part of the of the AI uh debt. complex. These are relatively long duration, they're kind of 15 to 20 year bonds. They have a very attractive carry. So these are in the 7% plus range today. And there's a quite a bit of contractual protection in there. So we dig into the terms and conditions, and we have a framework where we analyze all the different characteristics.
10:27such as how long is the construction phase, because that's where a lot of risk resides before rent commences. Who's the tenant? What are the terms of the lease? How easy is it for the tenant to get out of the lease, if at all? And all of those things we kind of go deep into the weeds of the of all the agreements. And where we're comfortable, we think those are good bets. On the chip back deals, this we're having a hard time getting comfortable. There aren't a lot of deals to analyze, and I think the market is trying to really feel this out in aggregate. We've only seen one deal get syndicated in the IG market, and it was a fairly well-structured deal. We ended up participating in in a limited uh way, but that had a lot of contractual protections. and actually a third party guarantor behind it. And so that made us comfortable that we could step in. And that also priced at a significant discount to the underlying guarantor. So so that's how we got there on that one. We'll see what the the next generation of these deals look like. One strategy, we believe that the early deals in this asset class will have the most robust terms for lenders. And as more capital flows in, as investors get comfortable with the
11:47asset class, the quality will deteriorate and then we'll be able to determine if it still meets our standards. And so that's that's part of our thesis on this. And the early data center deals have quite a bit of contractual protections for us. And so we we think that's we're worthy of of a basket type approach. It's been volatile, I will admit, but it's uh we do think that the extra carry helps offset some of that. So
12:15Unidentified speaker (speaker_0): yeah I mean I think that phenomenon is pretty uh it's a pretty well established phenomenon where the the earlier you are in that market, before all you know you being early and in enrolling and rolling up your sleeves can can often pay dividends. And I think that's I think that's pretty consistent with what we've seen elsewhere, right? So the the one of the things that you that you caught my attention in the materials is that, you know, you talked about when collateral gets old before the loan does, right? Which is like it was a little sobering to me. We're accustomed to financing long lived life assets and when technology gets outdated, clearly it it is is declining in its value. What happens when the un when the economic life, the useful life of the technology is shorter than the debt financing it? And how should credit investors protect themselves? I know this is your this is the center of the fairway for you. And I think it really helps our audience to kind of understand this point.
13:13Unidentified speaker (UNKNOWN): Yeah,
13:14Unidentified speaker (speaker_3): we've done a lot of work. on this internally. My colleagues across the firm have a lot of expertise in different areas, and we have a lot of access to management teams and so forth. So we're very much a front-row seat as to how this is developing. And you've seen some announcements in the press from the semiconductor OEMs like NVIDIA and Broadcom that are working to establish deep funding pools for basically chip-backed borrowing. And the way they're doing this is a combination of credit enhancement. So it's more than just asset backed. There'll be other forms of credit enhancement that supplement the asset collateral. This could be the form of a residual value guarantee from the chipmaker, which will cover maybe a first loss position. You'll have subordinated tranches in the deal that'll absorb a loss and equity in the deal that will absorb loss. And then you'll have a layer of contractual protections from the off-taker. And by that I
14:16mean, you know, the chip will be installed and then it will generate tokens, right? It'll it'll process information and someone will lease that chip. That could be anthropic, open OpenAI, AI, or Amazon or Microsoft. The credit quality of the off taker can actually offer a lot of credit enhancement as long as the contract is really tight and they can't get out of it, right? So a combination of all those support factors can get you to a finable structure. It will not be simply the collateral and the lease, like an aircraft. a lease like that, because aircraft have a very established secondary market. It's not difficult to repossess a plane, and we know what they should be worth because transactions happen and it has a 20-year economic life. A GPU has a three-year economic life, maybe, and a lot of that is very uncertain because of the pace of technological improvement and you know how fast new chips are versus old generation chips. And so they could become technologically obsolete really quickly, such that they're not even economic to plug in, right? And that's part of where we will we'll be very careful on what our kind of collateral gap is in a it in a transaction. So
15:35these deals will be amortizing deals
15:38Unidentified speaker (UNKNOWN): such
15:38Unidentified speaker (speaker_3): that they get to zero by maturity. So the refinance risk will be reduced dramatically. And really what you're trying to do is you're trying to get the amortization down fast to match that accelerated depreciation schedule on the economic life, and use other credit enhancement to bridge the gap, to bridge that loan-to-value loan to value gap along the way. And there's a price at which you're willing to step into that risk. And that's what the market's going to figure out over the next six months, like what that extra spread is. And it's going to be really interesting because this market could be hundreds of billions of dollars.
16:16Unidentified speaker (speaker_0): Yeah, it's interesting. It's it's really instructive to hear you kind of step through it. You make it very clear, which really helps me. I learn it like I'm learning. it too, right? There's other people on this podcast that go, I'm glad he went through that. I'm like, me too.
16:32Unidentified speaker (UNKNOWN): Uh
16:32Unidentified speaker (speaker_0): so let's talk about something that sounds like good news for all of us, which is the cost of compute continues to decline, which is a little like the cost of gas is declining, but but I'm using seven times as much. So what is good for the AI adoption maybe isn't good for everybody who finance the infrastructure. So as the cost of compute falls, who do you think captures the economic benefit and who could find themselves on the wrong end of that equation?
17:04Unidentified speaker (speaker_3): Yeah. Well, that's no small question for sure. And it gets to the kind of the central debate in the market today around the cost curve. for compute, as you say, because in technology we have historically seen r rapidly declining cost curves, offset by rapidly increasing demand curves, right? We have very elastic demand. And so the bull case on this investment cycle is that as the cost of compute falls, so will the price and the demand, the volume growth, will still lead to overall revenue growth and profitability growth. And there's been a lot talked about this Javon's paradox, which to me is a fancy way of saying price elasticity of demand, which is a very traditional economic concept, right? Just with a technology angle on it, right? And Javon's Juvon's paradox is about steam engines and English. in England in the 1800s or something, right? We
18:05Unidentified speaker (UNKNOWN): can call this
18:06Unidentified speaker (speaker_0): the Staden paradox.
18:07Unidentified speaker (speaker_3): No, it's definitely not anything I invented. I
18:10Unidentified speaker (UNKNOWN): only, you
18:11Unidentified speaker (speaker_3): know, absorbed this from other conversations and smarter people at our firm. And so that's kind of the bull case, right? That we have the demand for intelligence and the utility or value of intelligence is such that people pay for it, but that the amount of value it delivers. is compelling such that the demand continues to skyrocket. I think it's a believable case. But I think in the value chain, the timing of these shifts matter. So if you're a player that sells compute, right, like Amazon, and the cost of compute is declining, that's great. But it's a competitive market. So the price of compute will decline as well. So you've got to outrun your competition. You've got to be a low-cost provider in that market, right? And so someone like Amazon, very well positioned to be that, to be a low-cost provider in a commodity market that has a declining cost curve. A smaller player would struggle in that. environment. Someone who has a higher cost of capital or less scale or some other constraint to having low unit costs. And then the AI model companies, Anthropic OpenAI, Google, Meta, whomever, they will benefit from declining cost of
19:31compute, both for training and inference, so long as the market will support. a positive margin on price for compute, right? So those are the variables that we don't know how they're going to play out. Uh we're very early in this cycle. So what we want to avoid are companies that lock in fixed costs, right? And least infrastructure isn't the most flexible, right? It doesn't really help you when you have a declining cost curve. So companies that are leasing a lot of their infrastructure are going to have a less flexible unit cost than companies that own it, that have owners economics and can really get the operating leverage on what they're investing behind. I think that's a big distinction. And I think that the scale advantages in that market for compute are going to dominate and the big players are going to really,
20:25Unidentified speaker (UNKNOWN): well,
20:25Unidentified speaker (speaker_3): you know, I we're and already kind of seeing that to the extent that the the current environment is is in such a supply-demand imbalance that some of those companies with cost disadvantages can survive because the price of compute is so high today. But as that falls, we could see a shakeout. In fact, just this week, we're starting to see some pressure on the low end of the compute market. Companies that we call neo clouds, which are these smaller compute providers, because of the rate pressures in the market and their highly leveraged balance sheets, they're showing some signs of distress. I think there's some questions on that business model, basically kind of what I just laid out around is this going to be a sustainable business in a market that's going to be so competitive in the future? And so we've seen that create a lot of volatility this week. It's bleeding through, not just in the high yield market where these neo clouds are mostly capitalized, but even into the lower half of the of the IG market, we're seeing some spread pressure from this exact thing.
21:30Unidentified speaker (speaker_0): And may ask you this: is it the decline in the price of compute or the increase in interest rates? that's causing the issue there?
21:41Unidentified speaker (speaker_3): The rates move is the catalyst. That's
21:45Unidentified speaker (UNKNOWN): because
21:45Unidentified speaker (speaker_0): they're highly leveraged, because
21:47Unidentified speaker (UNKNOWN): of
21:47Unidentified speaker (speaker_0): the capital requirement to provide the compute.
21:50Unidentified speaker (speaker_3): Yeah, that's right.
21:51Unidentified speaker (speaker_0): Okay.
21:52Unidentified speaker (UNKNOWN): So
21:52Unidentified speaker (speaker_0): insurers talk uh, you know, I mean, the CIO community, everybody else, right? Diversification is super important in the investment business. And diversification is really dependent upon correlation, because at a correlation of one, diversification is irrelevant, right? So you can own this stuff, this AI infrastructure a lot of different ways. But what the folks that show up on our CIO calls are really trying to figure out is how do I identify the concentration risk in AI infrastructure when it may not be obvious from the individual securities or borrowers? And honestly, like this exact question was on our most recent CIO call because not everybody in the CIO community is convinced that this is a panacea. And a lot of them have seen this kind of thing before. Like they got a bunch of gray hair
22:52like I do, and they go, I've seen this movie. And every time that somebody tells me it's different, uh, this time it costs me money. And so can you talk a little bit about how to identify that concentration risk?
23:07Unidentified speaker (speaker_3): Yeah, I think it's a really important point. It exists and it will grow because you know, a lot of the ways we manage portfolios is around ticker concentration, right? We have an individual bond with a issuer ticker, and we set ticker limits, right? And we diversify by company. And we, you know, under the broad assumption that different companies have different fundamental risk factors and different default probabilities, et cetera. But in this case, we have a lot of special. purpose vehicles being created to fund this infrastructure. And their credit enhancement, or their underlying credit, is coming from the off-taker. Example, Meta has a very large data center in Louisiana that they are building a $30 billion building, two gigawatt data center, and they're partnered with Pimco and Blue Owl to fund that. So it was a lot of headlines on that transaction. That ticker is different than Meta, but it's the same, it's meta risk, right? So as portfolio managers, we need to correlate those two tickers and kind of treat them as one security and set our assurer limits based on what those two positions are in aggregate and treat them as one. But we need to do that across the
24:27entire market, right? So we have to have a methodology to map these tickers to their underlying fundamental credit risk. And that's something we do through our systems. You know, we the analyst inputs the relationships and then our systems can aggregate. And then we can look at all of our positioning, market weight, net of benchmark aggregated by the underlying. ticker, right? And we can then determine like how much of that meta risk do we have versus the index. Now that that SPV bond for the data center is not in the index, right? And so by owning that, we're taking a fairly big active bet. It's not important for insurance accounts, right? Because they're total return accounts, but it still matters as far as the core as the total position size being aggregated together. and managed as one risk. Then we have another cut at it where we look at kind of AI factor risk, right? We map all of our tickers to different AI factors, and we can aggregate it at the portfolio level using those factors, AI winner, AI loser, and then AI winner within the data center, the chips, the power, et cetera, right? And we can kind of see where our exposures are to those factors And that's something that, you know, we can do in the database and and then all of our r risk reporting can flow downstream of that, which is helpful
25:47to kind of for a portfolio manager to see it. And that's comes from our our fundamental platform feeding into our internal systems and and allowing all that to work. But it is tricky. It took us a while to get that going and you know we're I think we're up and running with it, but it's it is a lot of work to make it make it happen.
26:05Unidentified speaker (speaker_0): It's one of those things that folks don't understand. Like that is a b like when p you know, you go, Hey, I'm gonna hire an asset manager and and look at the fees and whatever else. And you go, Yeah. And when it when it comes time to building a system like that, like you have got to have some scale and you have to have some some folks with some smart smarts to even take that on. I mean, to have AI factors and to be able to map the portfolio back, that sounds like, I mean, you know, what do I know? But at the end of the day, seems like a reasonable approach. And the fact that you've you've thought it out and it's not like you didn't like that that wasn't an answer that came up on the fly. Like that's something that sounds like it's baked into your process, which is uh which is really I mean it's a great point. So I want to finish with something that is particularly interesting in your background. You were a network planning engineer for ATT in the nineties, and I wasn't kidding when I tell you that everybody that I know worked at ATT in the nineties has got some fat house someplace. But when the telecom industry was building enormous amounts of infrastructure to support the internet, the demand was real to the innovation, and there there were some painful outcomes. So what lessons from that telco build out may apply to
27:25today's infrastructure boom? And where do you think the comparison breaks down?
27:32Unidentified speaker (speaker_3): So a lot of this is a really well-studied phenomenon, I think. I I was I in my 20s. twenties in when I was at ATT, so I did not I was not on the right end of the that uh Oh
27:45Unidentified speaker (UNKNOWN): you didn't get the you didn't get the
27:45Unidentified speaker (speaker_0): big the big house.
27:47Unidentified speaker (UNKNOWN): Yeah, yeah, yeah. I forgot that. That's right.
27:49Unidentified speaker (speaker_0): The most of those folks I know are like they're they're go they're even older than I am, so there you go. Okay. I'm all right, you're off the hook.
27:55Unidentified speaker (speaker_3): Uh but I I would say They, you know, in that case it the asset that drew the most investment dollars was fiber, right? And big conduits full of fiber, both long haul, metro, subsea, you know, the bigger the better. And then, you know, technology came along that could multiply that fiber, really much more less expensive than building it. And so it sat dormant because the electronics and the optics on either end could multiply it by a hundred times. And so that the lesson there, the one warning, is that the pace of technological advancement can overwhelm the asset value. And if if a cost curve is too steep, you know, you can undermine the economics to finance it. Right. So we see that with the GPU analogy today. Now, the assets are different because fiber has a very long economic life, 20 years. plus, right? And a GPU is a we think three, maybe five, if, you know. And so that's a a difference. And then I think, you know, we're financing these structures with amortizing debt and not and not trying to extend the economic life. We're trying to shorten as far as like the business case, right? We're trying to shorten the case and get it paid off sooner. But still, that same those same economics matter. And I think the
29:15other part, and this is where I felt this very personally, I was working at ATT Corp, which was the long distance business. And when I started working there, long distance phone calls were 28 cents a minute, right? If you remember those days.
29:29Unidentified speaker (speaker_0): I do.
29:29Unidentified speaker (speaker_3): And it went to zero, right? So the analogy there is that that was a really profitable business that was a monopoly, right? Almost an and it went to zero. And all that profit just spread out
29:43Unidentified speaker (UNKNOWN): into other
29:43Unidentified speaker (speaker_3): parts of the of the market, you know?
29:45Unidentified speaker (speaker_0): Yeah.
29:46Unidentified speaker (UNKNOWN): So
29:46Unidentified speaker (speaker_3): we have these big monopolies in search and social, right? That generate massive profits, and that's bankrolling a lot of this investment. And so I don't know, you know, that so far, Google has done a really good job of protecting that search business. And same thing with Meta, like they've done a good job of protecting social, but you see the urgency with which they're investing in AI. And you see how that is an existential threat to those legacy businesses. So we need to say their legacy, like they came within our lifetimes from zero, right? But they're they're still like legacy businesses in an AI sense. And so
30:25Unidentified speaker (UNKNOWN): the
30:25Unidentified speaker (speaker_3): the durability of those businesses to this disruptive threat, I think is an interesting parallel to the long distance phone business back in the day of how it wasn't enough. The runway was too short for ATT to invest into something more durable. And they ended up, you know, merging with the local company. And that's the business that exists today is that local wireless business, not not the long distance business. that I came up through. And so I d I I think that's interesting. But it if you push it too far, it starts to break down. And I I think that uh you know we're still in the early innings here and the utility of the product to me is the most compelling argument and how it's changing how you do knowledge work and even how you kind of it creeps into your personal life and how how amazing it is. So that's like a hard thing to forecast, but it's it's, you know, it is here. And it's making this job fascinating for sure. It's the most interesting year I've had maybe since I've been working the past twelve
31:31Unidentified speaker (UNKNOWN): months. Yeah.
31:32Unidentified speaker (speaker_0): Oh, that's cool.
31:33Unidentified speaker (speaker_3): Yeah. Well,
31:34Unidentified speaker (speaker_0): good that's awesome. And what a great education today, Mark. I really appreciate it. It's been good to have you on. You got a fun one for you in the way out the door. We always ask this a version of this. Describe your best real or fantasy dinner. Who's coming? What are you having? Where are you going to do
31:51Unidentified speaker (UNKNOWN): it?
31:53Unidentified speaker (speaker_3): I would say
31:54Unidentified speaker (UNKNOWN): I'm
31:55Unidentified speaker (speaker_3): a Red Sox fan, and on Yorke Way in Boston, right outside Fenway Park, they sell sausage. and pepper sandwiches from those little carts. And that to me is maybe the best meal on earth. Maybe with my son and my father, I would think that would be an amazing, uh, amazing experience.
32:17Unidentified speaker (speaker_0): That's awesome. It's been great to have you on, Mark. We've been joined today by Mark Staden, Stodden, who's a credit analyst at T Row Price, where he covers the technology sector and leads a team of sixteen credit analysts in the US and UK. Mark, it's been a pleasure. Thanks for being on.
32:34Unidentified speaker (speaker_3): Thank you, Stuart.
32:35Unidentified speaker (speaker_0): I want to say uh a quick word to our audience. Our audience is so loyal and so faithful to us, and we certainly appreciate that. Uh we have just been awarded another set of podcast awards, and uh there's gonna going to be a more formal announcement, but I just want to uh express my gratitude to this, to this audience and to everybody who listens. If you like what we do and we certainly appreciate it, uh please rate us and review us on Apple Podcasts, Spotify, or wherever you listen to your favorite shows. You can also view us or watch us and to see what we actually look like on the YouTube, insurance AUM Community on YouTube. Thanks for listening. This is the home of the world's smartest. money on the Insurance AUM Podcast.