414. The AI Capital Boom, Stubborn Treasury Yields, Office Bets, & Insurance Relief

0:05Welcome to the Trepwire podcast, the show where commercial real estate meets data and insights. This is our Week in Review for the week ending August 14, 2026. I'm Haley Keen with TRUMP, a data modeling and analytics firm for the CMBS, commercial real estate, and CLO markets. I'm with Lonnie Hendry, Chief Product Officer, and Stephen Bushbaum, Head of Applied Research and Analytics. Inflation was once again front and center this week, and the July Consumer Price Index landed almost exactly where economists expected. Headline CPI increased 0.1 % during July and 3.4 % from a year earlier, while core inflation increased 0.2 % for the month and 2.5 % year over year.

0:50That inflation theme also ties directly into our data point of the week. Later in the episode, we will examine how property insurance costs have changed for securitized multifamily properties since 2020, including how Florida compared with the rest of the country during the sharp run-up in premiums, and why the most recent results suggest that the trajectory may be changing, even as owners continue to operate from a significantly higher cost base. Equities and treasuries initially rallied following the CPI report, but not because inflation is back at the Fed's target. Investors have been bracing for something hotter, so the collective exhale when CPI printed in line with expectations gave markets a lift.

1:35AI-related stocks also supplied much of the lift in equities. CoreWeave, Neveus, and Supermicro all posted substantial gains during their earnings reports. While NVIDIA traded higher after announcing partnerships with several major investment firms to establish AI infrastructure financing programs targeting more than$500 billion of third-party capital. Stephen, we have a lot to cover this week, as usual. But when we combine the latest inflation report with last Friday's surprisingly weak jobs number, what is the market telling us about the Fed and interest rates? Well, I think the market is telling us two things.

2:16The Fed has more room to wait, but the pressure on long-term borrowing costs has not really gone away at all. Wednesday's CPI print did not give the Fed an all-clear on inflation at all. But it's certainly weak in the case for an immediate rate hike. That's particularly true after last week's non-farm payroll print. Payroll employment declined by 23 ,000 jobs in July, and the prior two months revised down, lowered by a combined 103 ,000 jobs. So that's interesting, considering that consensus was expecting a gain. Now, I just also want to mention briefly that we'll be getting an update here in September.

2:56And so what happened last year was we got the September update and it completely rewrote the entire story for job gains for the year. We could be facing a similar thing again this year. Now, the initial market reaction after the CPI print on Wednesday reflected that relief. Equities advanced, treasury yields moved lower, and the probability of a September hike fell. But inflation remains above the Fed's target. Energy prices are elevated. policymakers will receive another employment report and another CPI print before their September meeting. So there's still a ton of macro data left before the next Fed meeting.

3:34Now, the more interesting debate may be taking place on the longer end of the Treasury curve. OneCamp argues that the Fed may ultimately need to raise short-term rates to bring long-term rates down. The theory is that decisive action would strengthen the Fed's inflation-fighting credibility, giving bond investors greater confidence that inflation will return to target and reducing an additional term premium that's embedded in longer term yields. Now, the competing argument is that long-term rates are facing a broader supply and demand challenge. Treasury continues to issue substantial amounts of debt while the AI infrastructure boom is creating another enormous source of demand for long-duration capital.

4:14So in other words, the AI trade is so strong, it's soaking up a lot of demand for that longer term paper and is going to keep the longer end for treasury yields higher. And I think that carries a lot of weight. Now, interestingly, we received some timely treasury demands signal on Wednesday afternoon with the auction of$42 billion in new 10-year notes. The auction cleared at a yield of 4.683 % with a bid to cover ratio of 2.53 times. So indirect bidders took nearly 77 % of the offering. Primary dealers were left with less than 9%. So that's a very healthy demand story. But get this, the yield on the 10-year notes is the highest it's been since 2007.

4:59Buyers are requiring close to a four spot seven yield. So that's telling us long-term capital remains quite expensive relative to where we thought we were gonna be at the start of the year. Now, at the same time, AI-related companies have dramatically increased their borrowing. Four major hyperscalers issued approximately$195 billion of bonds during the first half of this year. And issuance from the five largest companies is projected to reach roughly$250 billion in 2026, potentially$400 billion in 2027. So NVIDIA's proposed$500 billion financing ecosystem reinforces just how much additional capital the industry may require.

5:38And then importantly, we also got a fund announcement from Bank of America or B of A this week announcing another$250 billion in fundraise or basically funding out there for this industry. So close to a trillion dollars in capital this week for AI infrastructure, energy, and just this broad AI buildup. It's absolutely nuts. For commercial real estate, the takeaway is that a Fed pause or even an eventual decline in short-term rates does not guarantee the 10-year treasury or CRI borrowing costs will fall by the same amount. Treasury supply, AI infrastructure, financing, inflation, all of this could keep the term premium elevated for longer.

6:19And that connects back to the data point of the week that we'll talk about in just a little bit with insurance costs. And this is going to be an interesting one this week, I promise. So Lonnie, are you surprised that we're talking yet again about AI? No, I mean, I think the takeaway here is just AI is the engine, the economic engine that just keeps giving at this point. And I think what it's doing is it's creating an ecosystem where you're going to have people that remain all in on this just based on the scale, the scope and everything that we're talking about here. But you're going to get a growing number of people that say this is too much.

7:01It's significantly impacting the broader macro economy. And we haven't seen or realized any real gains from this at a scale that would justify the spend. And I think we're going to see that debate start becoming more and more heated. You add on top of that the local moratoriums, and this gets to be very interesting very quickly. I mean, just looking at that$1 trillion, Steve, I mean, I comment on this every single week, it seems like, that it used to be a really good-sized deal in the tens of millions, and then it went to$100 million in the CRE space, and now it's$1 billion. These are in the$100 billion, the spend on these and the capital requirements.

7:42And it's just, it's the scale is incredible. And, you know, I don't know. I don't know what this means. I mean, I think to all of your points, and I'm not going to go back through them, I think you did a great job covering them. But, you know, the reality is the Fed has some more data that's going to come in before their next meeting. So I think it renders, you know, kind of what the takeaway is today somewhat mute in the sense that it's not going to be the latest data that they have to make a decision on. I do think that they're going to have to take, you know, some definitive action relative to at least guidance.

8:15If they leave rates the same, there's going to be even more questions. So even if they don't raise rates, I think there's going to have to be some sort of talk track from them, which has been missing in Warsh's first couple of meetings. And I don't know that he's ready to change his position on that. So now look, the employment stuff for them, I think, is good. Jobs cooling at some level gives them a little bit of a cushion. I think CPI, as you mentioned, was a little bit, you know, not as hot as what people had anticipated. I think some of the wild card stuff still is the oil and what's happening with Iran.

8:48You know, I think we all thought that that most recent act of war, whatever we're going to call it at this point, probably would have led to some sort of compromise. I don't think that's happened. It doesn't look like it's going to happen. So to me, the oil is the part that really could drive this in a way that maybe changes the narrative here. So, you know, like we tell our listeners all the time, Stephen, we're in a day-to-day business, week-to-week business, month-to-month business. And I think these data points just exemplify that. So, yeah, Lonnie, to your point, what will be really interesting to watch here is that the Jackson Hole conference is coming up.

9:24and I'm really interested to see what Warsh has on the table for this meeting, because there's a lot of people saying he needs to address X, Y, or Z, but then you step back and wonder, like, is he really going to cave? I mean, he seems fairly steadfast, but then behind the scenes, there's been some interesting collaboration between Besant and Warsh and some stuff with the Japan intervention coming in here. So there has been some capitulation behind the scenes very, very quietly to try and basically avoid some treasury selling. And I know there's pressure there, but I'll be curious to see what kind of a tone he comes out with at Jackson Hole.

10:04I think what he has to do is take back control of the narrative here. If he doesn't want to provide the guidance, that's fine. It's going to make markets more dependent on individual inflation, employment, growth reports, and the things that we look at. You know, there was a nice write-up in Reuters this week talking about, you know, when there's less visibility, reaction function, you know, is going to cause investors to demand additional compensation, which you highlighted earlier on their long-term bonds, adding to term premiums and borrowing costs. So if he's comfortable with that and he thinks that that's on the scale, he would rather stick to his script of no guidance and absorb the additional risk premiums that come through the market, that's fine.

10:47But I think he has to do a better job of presenting himself and the hows and whys and why he's comfortable with that. So if he doesn't provide guidance and he doesn't really provide any comfort to the markets around why he doesn't think it's necessary, it continues to lose control of the broader narrative. It's going to be a long road for him because, as you mentioned, the pressure in D.C. is not going away for the market participants, the lobbyists, the folks that want that guidance. They're going to be using all the pressure they can and back channels and other things to make sure that they get what they want.

11:23And so you mentioned some capitulation. I think you're going to see a continuation of that where, you know, in his mind, he probably holds firm on the not formally providing guidance. But I think there'll be, as you mentioned, you know, after the last meeting, some of the Fed governors or maybe some other folks that kind of let things trickle out in a way that the markets get what they need, which is not in such a formal kind of front and center approach that we're used to. So Lonnie, I saw this Reuters piece about soaring AI hyperscaler default edges, in particular, they weren't what they seem.

11:55So obviously, I had to dig deeper on this one. As we mentioned earlier, there's been a massive amount of debt issuance across hyperscalers with Microsoft reaching 250 billion 26 and projections for 400 billion in 2027 and combined issuance. It's just unreal what these numbers are reaching. So obviously, if we have this much corporate debt issuance in the hyperscale sector, there has to be some default hedging as well. So what the author of this article picked up on is that credit default swaps or CDS spreads in particular have increased fairly significantly. So long and short of this is that while some of this increase does have some concern to it, which I'll get to, really the broader point is that, look, we have way more debt trading in the market than what we used to.

12:46And so really, this is a recalibration of the market adapting to how much volume is out there. So the one spot that I picked up on, and we've talked about this in the podcast, the one clear area of credit concern they highlighted was Oracle. The company has approximately$130 billion of debt and its CDS spreads have widened from about 40 basis points one year ago to more than 200 basis points. And S &B Global Ratings recently lowered their credit rating to triple B minus one notch above speculative grade. And if you read some of the Michael Burry posts out there, Oracle is one of the main companies he's hammering on.

13:21He's hammered on a few others, but the one that concerns me most has been Oracle for their forward commitments and some of the questions about like, will they actually get there on revenue generation and be able to honor? Because right now it looks like there'll be a negative cashflow territory. And obviously S &P agreed with some of that argument as well. Look, Oracle has been the one that everyone's kind of been paying attention to for the reasons you mentioned. I don't think that that's news in the sense that people have been saying this for a while. I just can't get my head wrapped around some of this broader issuance stuff that you've covered.

13:53I mean, if you look at just 2024, the hyperscalers had about$16.7 billion in issuance. They're at about$195 billion in the first half of 26, which is 80 % more than their combined issuance during all of 2025. and then as you mentioned, Goldman estimating this reaching up to$400 billion in 2027. It's just phenomenal. But at the same time, it's great. I mean, this is what innovation looks like. This is what people are wanting to see, and the markets have an appetite for it. So it's just, for me, just seeing the numbers on the page is just, it blows me away just how large this is and has become and how rapidly we've gotten there.

14:35Yeah. I mean, the five hyperscalers plus Microsoft reaching$400 billion in 2027, that's a stone's throw away from the size of the CMBS market. I mean, that's what I'm saying. And literally, like when you were at$16 billion two years ago, three years ago, I mean, it's just insane. But this is where we are in today's world. And I think for all of us, we're hoping that this plays out. And I think it will. I'm bullish on this. But as you mentioned, maybe Oracle, maybe Oracle and someone else probably doesn't reach the level of success that everyone else does. And there will probably be some losses at some level here.

15:15I mean, that's just how the market, in every market, regardless of sector, you're going to see winners and losers. And as you mentioned, Oracle is probably set up right now to be on the short list. So while we're recording before the retail sales data we'll get this week, we do have at least a good proxy for what we could end up seeing later this week. The Chicago Fed has an advanced retail trade summary. It's called the CARTS data. It's the Chicago Fed advanced retail trade summary for CARTS. And it projects that retail and food services sales, excluding motor vehicles and parts, increased 0.1 % in July on a seasonally adjusted basis.

15:53After adjusting for retail price changes, this CARTs data estimates that real retail and food services, excluding autos, were unchanged during the month, which suggests that there really has been a modest increase in nominal spending that's reflecting higher prices rather than greater purchasing volume. And I find this to be a really interesting data set because it combines eight high frequency indicators. You have credit card transactions, retail foot traffic, gasoline sales, and consumer sentiment, along with previously published Census Bureau retail sales data. It provides you with a really good proxy and an early estimator of national spending before that official monthly report gets released.

16:35So what'll be really interesting to watch here as we continue kind of updating, going into the beginning of school or back to school push, and then the holiday spending, this will be really telling for how much purchasing power is out there. Because if you kind of step back and look at the inflation data and wage growth, really real wage growth has been negative. It's been advancing less than inflation has. So purchasing power has gradually been eroding for the last couple of months. And if that has become a binding constraint, it really wouldn't be surprising to see retail sales miss to the downside here.

17:15Maybe not back to school, but by the time we get to the holiday, how exhausted will the consumer look? Listen, I think the consumers, they've got themselves trained for a marathon, man. I mean, they just keep spending. You know, this is a great kind of preliminary proxy, you know, but there's few components that it doesn't have relative to what we're going to see later in the week. And that's, you know, it doesn't have auto sales and parts. And there's a few other components that, you know, are not going to be part of this. But I agree with you, Stephen. And there's got to come a breaking point or an inflection point or a moment when spending starts to go the other direction.

17:54And, you know, I think if people went back and listened to our show for the last two years, we're both probably on record thinking that we were going to get there a lot sooner. And, you know, the cumulative effect of this inflation and then to your point, you know, the realization of wage decreasing in an inflationary environment. It's tough out there. I mean, I'm in Arizona this week, and just regular unleaded gasoline was$5 a gallon at some of the stations. It's just amazing to me how different it is state by state. I mean, in Texas, it's like$3.50, and people think that's crazy. Arizona is like$5.

18:31California is$7. Obviously, the wage scale is slightly different, but not significant in most of those locations. So it's just tough, man. I think for a lot of people in the market, expenses have exceeded income for a while. And I think we'll start seeing this breakdown. Credit cards, auto delinquencies, other things are going to have to start peaking at some level because people just can't float it forever. So, Lonnie, I saw a story out this week, and it got me asking this funny question to myself. If you were picking a kickball team based on the top investment houses on Wall Street, who would be your ideal kickball team?

19:13And I'm pretty sure the partners on this announcement are it. It's like the all-star kickball team. So NVIDIA partners with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish AI compute infrastructure financing platforms to mobilize over$500 billion of third-party capital. That's a fantastic lineup. That's like the all-pro team. That's not even like all-district or all-state. This is like Hall of Fame worthy. Yeah, I mean, that's the who's who. And I think it just shows, you know, NVIDIA in particular. I mean, being associated with them in today's world brings a certain level of cachet.

19:52And all of these top players across the different landscape want to be part of it. And the momentum is real. And it's indicative, I think, of Apollo, BlackRock, these other firms. They're not wanting to miss out on some of these opportunities, even though some of it's unknown. They're all ponying up and they're trying to get themselves in the game and stay relevant through this new AI world that we live in. Yeah, we are going to see rapid advancements in this industry. I mean, it's just wild where we're going. So there were two other major AI infrastructure announcements here this week. Bank of America is going to deploy$250 billion to bolster AI and energy infrastructure.

20:34So similar to that prior announcement, we now have close to a trillion dollars in capital. I mean, shoot, what's another$250 billion from here? So it basically seems like everybody is getting in on this trade. And then as part of this, CME Group and Silicon Data just announced this week that they're launching compute futures on October 5th to unlock a new way to hedge AI risks. And so these new contracts will track Silicon Data Indexes measuring hourly rental prices for NVIDIA's H100 graphics processing units or GPUs and its newer Blackwell B200 GPUs. Each futures contract will represent one month of rental costs.

21:16So, I mean, this is just amazing. I'm imagining what's being financed by these massive pools of capital, right? And having it being linked to basically how profitable you're going to be. And then the ability to hedge the compute costs as part of this. This is fantastic. It's like the airline industry taking off. Sorry, that's a terrible pun. I couldn't help myself, but it's equivalent to that, basically. You know, I thought your pun was good, Stephen. For everything that feels new about this, it's not really new. It kind of always say history doesn't repeat itself, but it rhymes. But CME, you know, they said this is the development of compute trading, you know, is effectively like an evolution from oil being a physical commodity into like a global derivatives market.

22:00So I think a lot of these things and what we just talked about with some of these really large players setting up kind of an ecosystem for financing. I mean, like there are so many analogies that you could look back on and say, you know, in their time, they were revolutionary. this is just a different driver but it kind of is finding its way into some you know comfortable market type of uh dynamics where people are comfortable with this i mean like yeah of course we should set up some sort of future hedging marketplace for people to benchmark and trade why wouldn't we we do it for a bunch of other things too and it's just this ai everyone is It's like you've seen people that put just.ai in their URL or they change their company name to something AI, and the market rewards them with a higher valuation, whether when they're fundraising or they're trying to be acquired.

22:54It's just amazing how everyone wants to get their piece of this AI push. Yeah, so let's turn to another headline we saw this week. So we have banks, private capital, and tech companies all building financing structures around AI. And on the real estate side, we are seeing that same capital rotation show up pretty clearly in portfolios. There was a headline in BizNow that caught our attention about BREIT exiting self-storage as it invests billions into data centers. Yeah, this is great. They're just exiting one type of storage or another in a nutshell. I mean, that's a gross generalization. So BREIT is selling its final 79 properties, totaling approximately 5 million square feet.

23:42Transaction generated to 852.3 million of net proceeds and a$177.3 million net gain. The sale completes a multi-year disposition strategy that included the 2023 sale of a 9 million square foot self-storage portfolio to public storage for$2.2 billion. At the same time, BREIT invested$3.3 billion in data center development through QTS during the second quarter, bringing its first half investments in pre-lease projects to$5.7 billion. BREIT said the QTS developments are fully leased in substantially all cases to investment-grade tenants. QTS leasing activity increased more than 50 % year-over-year during the second quarter.

24:24BREIT owns a 35.7 % interest in QTS with a book value of approximately$1 billion. and Blackstone's joint venture acquired the data center developer for$10 billion in 2021. B-REIT also sold 20 rental housing properties and 27 industrial assets during the quarter. Combined with the self-storage dispositions, those sales generated$2.1 billion of net proceeds and just under$300 million of realized gains. So this is a really deliberate portfolio rotation toward sectors with strong growth expectations. This is going to be a really fun rotation to watch play out because we're still in the incredibly early innings of this AI build-out as we've been discussing.

25:09And I mean, heaven knows what the ultimate market cap is going to reach on this portfolio. Yeah, but I mean, as you're going through that story, Stephen, I mean, those numbers are a rounding error compared to what we talked about earlier and throughout this show. I mean, just, you know,$800 million,$500 million, just crazy comparing commercial real estate to some of this AI stuff. It's just two different leagues. But look, these guys, they rotate capital. They're reallocating their portfolio. They're trying to maximize where they're at. So we'll see what happens. There'll be more stories from them on this endeavor as we continue the show.

25:47Quick break. There are a lot of AI tools entering the commercial real estate market, and Lev is one of the platforms taking a more practical approach to putting AI to work for CRE teams. It's not just financing software anymore. Lev gives you ready-to-use AI apps and agents for research, comps, deal materials, outreach, documents, pipeline, and team workflows. Go to lev.com, that's L-E-V dot com, create an account, get$2 ,000 of free credits, and tell them Jake sent you for another$500 in credits. All right, let's turn to our digging through the data segment. Stephen, you said something earlier that I think sets this up really well.

26:32Slower inflation doesn't mean costs suddenly go back down. and property insurance might be one of the clearest examples of that for CRE. So in this data poll, we looked at securitized multifamily properties. And after years of huge increases, we found that Florida is finally showing some relief. So what does the data tell us on the insurance side? Yes, this upcoming TREP research piece is looking at the annual changes in insurance costs at securitized multifamily properties in Florida compared to the rest of the country. And the results are quite interesting. I mean, Lonnie, we've talked about insurance costs ad nauseum, and in particular, what we've seen in Florida.

27:16But now I think we have enough data through time to really paint the clear picture of what's happened, where we've come from, and where we stand today. And there's really three stages to this story here. Before I get into the numbers, let me just give you some quick detail on the analysis so we don't confuse it with some other types of inflation reports. What we did is we looked each year at the median year-over-year change in the insurance cost for all securitized multifamily properties, both in Florida and the rest of the nation. So what we're talking about here are annual medians. We also look at that combined or compound change or the compound annual change.

28:00But when we talk about those, it's not like the exact same thing as, say, a same-store analysis or a chained inflation rate. We're really just looking at the combined medians. So Florida consistently outpaced the rest of the country in this first stage, 2020 through 2022. In 2020, the median Florida increase was 13.4 % compared with 10.6 % across all other states. In 21, Florida increased another 13.5 % versus 7.9 % in all other states. In 2022, Florida accelerated to 17.2 % compared with 9.1 % across the rest of the nation. Then came 2023. The divergence was especially pronounced. We had the double, triple, quadruple check this number.

28:50the median insurance cost for a Florida multifamily property increased 42.4 % in 2023. We talked about this, Lonnie. We heard it real time from operators, from all sorts of folks across the industry. I mean, the 2023 repricing was painful. That was more than two and a half times the 15.9 % increase reported across all other states, and that created a 26.3 % point gap between the two groups. But that pattern reversed in 2024. Florida's median slowed to 7.7%, while the rest of the nation increased 11.1%. That was the first year in our series where Florida insurance costs fell below the comparison group.

29:35And then the reversal became even more significant in 2025. The median Florida property recorded a 6.2 % decline in insurance costs. So insurance costs actually fell in 2025 at the median, with all other states showing a 3.5 % increase. So we had a really modest, almost tracking inflation increase in 2025, the first year where it felt, well, kind of closer to what we were told to underwrite in our undergraduate programs, just underwrite the 3 % inflation on expenses, and you'll be fine, right? It's hilarious to me that the number hit so close to that age old magic 3 % number we pull out of a hat.

30:14so does that mean the insurance cost crisis for florida multifamily owners is over lying no it's not over in the sense that for this year's policy it's a great relief but to your point i mean like there's a cumulative effect and impact here i mean it's and people are going to be gun shy you know the fact that you started to see a reversal from an underwriting perspective from a diligence perspective if you're doing you know a refinance or whatever and you're the lender What number are you going to believe? I mean, you've got multiple years here of double-digit growth, one year that jumps off the page at over 40%.

30:49And now you've had a couple of, you know, pullback years in 2025 show the decline. Are you trusting that it's going to continue to decline? Or do you think that the decline is the anomaly here? I think nationally, you know, the narrative has softened. People are really excited about the fact that insurance has become more manageable and more predictable and underwriting isn't, it's not prohibited like it was. I think Florida always is going to be an outlier just because of the exposure to the hurricanes. And we fortunately this year haven't really experienced anything to date, but they're saying with this El Nino, we're going to have, you know, potentially more significant storms through the fall.

31:26If I'm someone in Florida, I'm, you know, counting my blessings that the numbers have gone back in a favorable position, but I would be hard pressed to say that I think they're out of the woods yet. I think this is something that's going to be a hot button issue, a topic, an area of concern for owners and operators from here on out. I mean, like I don't think until there's something that replaces the kind of antiquated insurance industry that we have, it's going to ebb and flow more dramatically than we've seen in the past because the cost to repair these things when they are damaged is significantly higher in today's market.

31:59Yeah, I think we've stabilized for the moment, and that moment is going to last until the next disaster. It's what happens in every case. As soon as we have another disaster, policies reprice. To some extent, the insurers have to basically recapitalize themselves. Policies get repriced. And sure, customers have to switch to try and find that lowest cost provider. But that median will show the pressure. And well, hopefully we'll be able to be the ones who give you the data to show it. Yeah, I mean, of course, we'll be tracking that as near real time as you can. Something to keep an eye on, Stephen.

32:31And it's not probably as headline worthy from a cost perspective because the damage hasn't been nearly as bad as what you see in Florida with hurricanes. But, you know, Washington state has been inundated with wildfires and Utah inundated with wildfires. At least to this point, they haven't really like materially damaged, you know, hundreds of millions of dollars or billions of dollars worth of property. But we kind of are lucky at this point. You know, we haven't seen some of the wildfire activity that we typically do in California. and the stuff that we have seen has been outside of fully developed areas.

33:05But I'm with you. I think insurance is so fragile right now because of just all of these increased, you know, catastrophes that have been impacting them that we'll take the decrease. Everyone will be happy, but it's going to be short-lived. It's a storm-by-storm endeavor. Even in Texas, I mean, like, homeowners insurance, auto policy insurance still going up. I mean, I got my homeowner's insurance policy renewed in July, and it was up 12 % year over year. Significantly less than what it has been, but still increasing. It's interesting you mentioned those wildfire issues, Lonnie. So I just saw a headline in BizNow today saying that 10 states alone carry$1.4 trillion in wildfire property risk.

33:50We should probably do a deeper dive on this, maybe have someone that lives in this world that understands it at a level that we don't. But it's not just the insurance costs, but like I saw some stuff on the news this week with Utah, the cost for them to employ the people to fly the planes and the helicopters to get the water to try to put the fires out. Like they're not budgeted for that, at least at this level. So in addition, obviously they'll get some federal relief through FEMA and there'll be some stuff for, you know, emergency declarations and money will come in. But beyond just the physical damage, there's a real cost associated with having firefighters, both local and non-local, on the job 24-7.

34:31Having these helicopter pilots and other pilots flying these airplanes that are dumping hundreds of thousands of gallons of things to try to put the fires out. The cost of that adds up very, very quickly. And some of the folks in Utah were just saying, the public officials were saying, we don't have the money to keep this up, but they haven't got the fire contained. All right, let's get into some more commercial real estate property type stories. And we'll start here with office. And apparently we're not done talking about AI just yet. We've talked about the capital going into the AI build out, but now we're seeing it reshape office demand as well.

35:09There was a headline in the Real Deal this week that AI companies occupy 10 % of San Francisco office space. When you hear that number, Stephen, what is your reaction? Do you think 10 % is, is that more than what you would have assumed or is that less or is that kind of where you thought it would fall? That feels about right. And I'm a little bit biased only because I think of like where SASB issuance for data centers has gone or overall CMBS. I'm like, yeah, that actually tracks about right. We've gone from 1%, 2%, 3%, to 5%, to now 10%. So, okay. I was probably surprised a little bit. I thought it would be higher than this, given all of the headlines that we'd seen about leasing activity in San Francisco in particular for AI tenants.

35:54So just to give you some perspective, this is according to JLL. They're saying about 8.5 million square foot comprises the San Francisco office market. They say that AI companies now occupy 8.5 million square feet across the San Francisco office market, which is representative, as Haley mentioned, that 10 % of their office inventory. You know, as we've noted multiple times, activities accelerated in 2026. If you look at what's taken place just this year, AI firms have leased about 2.4 million square feet so far in 2026. So obviously it's making this sector incredibly strong source of new office demand in that market.

36:34If you look at the AI footprint, it's probably larger than what's reported. You know, JLL's figures exclude some companies such as Google and Meta because those are typically classified as technology firms broadly and not necessarily just AI firms, even though both of those entities have substantial AI operations. And if you look at just the number of AI office tenants, so not looking at square footage, San Francisco now has 23 AI companies occupying space. San Francisco had just 23 AI companies occupying space in 2022 compared to today's number of 413 companies. So again, we talked earlier about just the rapid ramp up of debt issuance for the AI hyperscalers.

37:22Going from 23 companies, call it five or six years ago, to 413 just shows how insatiable this demand has been. So I don't know what your thoughts are on this, Stephen. Is this another one of those headlines and stories that continues this pace? Do you think that we see something next year that says it's up to 15%, 20 %? Or do you think that we're going to reach some level of equilibrium in a market like San Francisco that probably by definition is one of the strongest AI incubator kind of accelerants from a market perspective that we have in the U.S.? It wouldn't surprise me if we had another 5%, 7%, 10 % from here.

38:05I'm hard-pressed to believe that it'd be healthy for us to get beyond 20 to 25 percent by say like 28 to 2030 that feels like you're getting overly concentrated a bit frothy and maybe is setting up for some sort of a collapse yeah I think if if you see it surpass 15 or 20 percent I think it's going to have pushed rents so high that they'll be looking for cheaper alternatives so cities like Austin, maybe some of these other like secondary tech hub cities start to benefit more predominantly or pronounced than what they do today. Because I'm with you. I think you cross over 20, 25 percent. That's too much concentration.

38:44And in fact, in some of these buildings in particular, you know, if your tenant composition in an individual building is 60, 70 percent AI tenants, that gives you some significant concentration risk at an asset level as well. And sticking with office, we are also seeing investors increasingly willing to bet on the recovery. Let's talk about the headline that we saw this week that Real Capital Solutions has launched a contrarian office fund with$50 million from the CEO. Yes, they are targeting distressed Class A and Class B properties in primary and secondary markets. The fund is targeting$350 million of equity commitments, with RCS plans to combine with debt to acquire approximately$850 million of properties.

39:33And as you mentioned, the CEO is contributing$50 million. Another investor has committed an additional$50 million. RCS has already acquired 14 office properties across 10 markets for approximately$644 million since 2024. The company says those properties were purchased at prices averaging 18 % below replacement costs and more than 50 % below their peak values. The investment thesis is that for sales, limited financing ability, approaching loan maturities, and wide bid-ask spreads are creating unusually attractive entry points for buyers with available capital. So Lonnie, is this like the announcement coming after basically the majority of capital is deployed or how much runway is really left on this strategy?

40:18Well, first, I think it's interesting when you launch a fund called the RCS Contrarian Office Fund, I think you're trying to elicit a certain play on words that says, you know, you're running into the storm when others aren't and you have figured something out that others haven't and I would maybe say like listen I don't think it's that contrarian at this point like I think there's a lot of capital coming into this space and people are you know snatching up good well-located offices and cities where pricing has reset to a level that makes them attractive I take a little issue and it's not just with this story but it's just broadly like can we get past the the comparing your current sales price to replacement cost like does that metric even matter i don't think that metric matters at all i think it's antiquated it's old school it's what people say when they can't justify like any other way as to why they bought what they bought it's like oh we buy the discount to replacement cost i mean maybe if you push me and we were talking about brand new construction and supply constrained markets with high barriers to entry, there's some legitimate value there potentially.

41:30But for offices that are trading now at 50 % below their peak values, I don't think using a replacement cost metric gives you any credibility in the market beyond just saying, hey, we thought we got a good buy. So yeah, to your question though, look, there are so many offices across the US that are still in uncharted territory in terms of value decline, occupancy decline, NOI struggles, CapEx that hasn't been put into the building, et cetera, et cetera. I think this contrarian office play, if you want to call it that, will have legs for the next five to 10 years in some markets. I mean, the market hasn't recovered at all on the national scale from a sales price perspective outside of the high-end double A's.

42:12And I think, look, we've had multiple guests on our show in the last couple of months that have all said, listen, we think now is the time to buy office. I'm totally bought in on that in major markets. But there's a lot of Class B obsolete office buildings that still need to trade at significant discounts. And I think that you'll see capital flow into this for the next several years. Okay, so let's pivot back to multifamily here. Lonnie, there was a pretty significant transaction. We saw news that H &R REIT is to be acquired in a$6.7 billion deal. Yeah, I mean, again, another really strong headline with a billion-dollar lead-in.

42:51As you mentioned, H &R REITs agreed to be acquired and divided among Go Residential REIT, Blackstone, Crest Point Real Estate Investments, PSP Investments, and a company controlled by members of the family of H &R Executive Chairman and CEO Tom Hofstetter. Transaction is valued at approximately$6.7 billion, as we mentioned. This includes assumed debt. It implies approximately$3.4 billion of equity value and provides a 14.5 % premium to H &R unaffected closing share price before reports of a potential transaction emerged. This Go Residential will acquire 23 Sunbelt apartment properties, H &R's interest in the Jackson Park apartment complex in New York, and several mixed use and other office assets.

43:36The combined residential platform would own more than 13 ,300 units across eight markets. Blackstone, Crestpoint, and PSP mentioned earlier will acquire portions of H &R's Canadian industrial portfolio, while the Hofstadler family-controlled buyer will take the remaining non-core assets. This completes their multi-year effort to simplify the portfolio. This is for H &R and gives existing investors immediate cash consideration and continued ownership in a larger, more specialized residential REIT. So I think the broader takeaway here is that portfolio breakups and greater property sector specialization tends to unlock value when a diversified REIT trades below the perceived value of its underlying assets.

44:20And we've seen this play out. And I think this is a pretty good representation of what that takeaway is here. And so, Stephen, I don't know if you have thoughts on this. I don't think this will be the last deal like this that we see. And this one has some pretty good players attached to it. Yeah. I mean, this is just interesting to kind of see the reverse of the acquisition. We just got done talking about the other week, right? We had a portfolio sale in Southern California, or I guess across California more broadly. And now we have a Sunbelt portfolio changing hands. So this is an interesting one.

44:53I'll be curious to dig into it a little more and see if we can't back into what the implied cap rate is on this acquisition. Because part of me really wants to compare this to some of the other portfolio cap rates we've calculated and see where it compares to that kind of mid fives rate that we've been seeing. So pivoting here back to data centers, of course, we had a few quick hits and then one story that we wanted to dig into a little deeper, another billion dollar headline. Yes. So this first quick hit on the headline, well, we talked about capital earlier in the show. Well, now we're going to get to the sticks and bricks.

45:29But I don't know if it's fair to say sticks and bricks for a data center. I think we're a little bit beyond that. So I don't know, maybe copper and concrete. So Anthropic, McGuire, and GIC announced a strategic partnership to develop dedicated data center infrastructure at scale. So the platform will initially focus on projects in the US with Anthropic serving as the anchor tenant. And then funds managed by McGuire and GIC will own the platform and provide most of the equity required for each development. Paired with NVIDIA's announcement, the story shows that AI infrastructure financing is evolving along two tracks, the broad financing platform backed by compute assets and dedicated real estate platforms supported by long-term tenant commitments.

46:13Now, another quick headline here, we saw Blue Owl's development arm paid$66.4 million for a data center site in Culpeper, Virginia. Through its stack infrastructure program, the firm acquired a 71-acre site in Culpeper County for$66.4 million, adding to its growing pipeline of Virginia data center developments. And so what we're seeing here is that migration beyond the Ludon County ends, basically we're having to broaden out from that northern Virginia epicenter of data center development. And so this Culpeper market offered entitled land and improving power access. So in other words, this provided everything that's effectively needed to get one of these data centers built.

47:00Power availability is the key reason developers are moving into the market. Previous plans for the area called for two approved 300 megawatt Dominion Energy substations, making the site attractive for this hyperscale development. Now, finally, the headline we're going to dig a little bit deeper into is a tenant deal. Fermi finally secures Project Matador tenant in a$6.5 billion deal. This headline coming to us from BizNow. AI cloud provider TensorFlow signed a 15-year lease for a 222-megawatt data center at Fermi's Project Matador campus in Carson County, Texas. The initial lease is expected to generate more than$6.5 billion of revenue.

47:45The agreement includes two five-year renewal options, and TensorWave can expand into two additional data centers at the campus. The facility is expected to accommodate tens of thousands of AMD graphics processing units, with TensorWave taking occupancy in phases as the project is completed. Now, this is an important milestone for Project Matador. After a difficult period involving construction delays, management upheaval, and the loss of a previously anticipated investment-grade anchor tenant, the share price for this firm had increased more than 21 % following the announcement. So this is a massive deal for the company.

48:22Tenant credit remains an important consideration. TensorWave was valued at approximately$1.5 billion and raised$350 million in June, making the$6.5 billion projected lease revenue exceptionally large relative to the tenant's current capitalization. And this is this funny point that we're at here. This whole industry on the growth chart is steeper than that 45 degree curve. We basically cannot afford to fail on this build-out of AI infrastructure and the adoption and the revenue that's projected to go alongside it. So this is a really interesting timeline here we have ahead of us over the next two, three years as this industry gets up and running at scale.

49:05Are you calling this too big to fail 2.0, Stephen? Certainly what it feels like. I mean, because if it does fail, Michael Burry is going to make a lot of money. And a few last stories here in retail and lodging. In retail, it actually seems like we're seeing Simon benefit from the SACS closures. We saw a headline that said Simon is poised to rake in millions more in rent thanks to SACS global closures. Walk us through this story, Lonnie. Listen, I don't think this is unique to Simon. We've talked about this for a while now. With some of these retail bankruptcies, you're actually able to backfill the space at a higher rental rate and with a tenant that's not struggling.

49:49And it actually is a net positive. It's a little bit of an interim disruption. And so to your point, you know, with sex bankruptcy and store closures, they returned about 1 million square feet to Simon Property Group with nearly all of the vacated space coming from their SACs off Fifth Avenue locations. So despite the disruption, Simon's occupancy broadly remained unchanged from the end of the first quarter. And this is where the good stuff kicks in. Simon's already signed replacement leases covering approximately half of the vacated space. Those leases are going to generate more than the$18 million in annual rent that SACs previously paid for the entire portfolio of returned space.

50:27So again, this just highlights. They've only leased about 50 % of the space and they're already going to generate substantially more rental income and they still have half the space to lease. You know, if you factor it in, once the remaining leases are completed, Simon expects to increase annual rent from the space from$18 million to approximately$44 million, nearly two and a half times the previous amount they were getting paid from SACS. The leasing economics have improved beyond the rent increases. Initial base rent on new leases assigned through the second quarter was 17 % up year over year, while tenant improvement allowances declined 12%.

51:03So that's something we've talked quite a bit about, Stephen, on the TI and office and retail build-out. It's nice to see that, at least for Simon, I don't know if this is indicative of the broader retail marketplace, but at least with their scale, they've seen TI allowances decline about 12%. The results demonstrate the value of well-located mall space in a tight retail market initially appeared to be a major bankruptcy-related vacancy event that people were a little bit nervous about has actually turned into a significant upside play for Simon. And we've seen this with Big Lot, Stephen. We've seen this with Bed Bath & Beyond.

51:41We've seen this with Party Cities and some of these other, you know, chain tenants that have filed bankruptcy. The owners of those centers are actually in a much better position now with stores that are growing, footprint, and rents that are much higher than what they had with those failing tenants. Finally, our lodging headline of the week is that Ryman Hospitality is making one of the largest U.S. hotel acquisitions of the year. The REIT agreed to acquire the Grand Lakes Orlando Resort for$1.38 billion, or roughly $867 ,000 per room from Trinity Investments. There's another one that just jumps off the page at me.

52:21We're talking about an Orlando resort getting almost a million dollars a key on this acquisition. And Ryman has made headlines in the past. I mean, they don't shy away from making really big deals. And this certainly fits the bill. At over$1.38 billion, a pretty significant deal. This increases Ryman's convention-focused portfolio. The Grand Lakes includes just over 1 ,000-room JW Marriott, which is 1 ,010 rooms to be exact. 583 room rich carlton 320 000 square foot of meeting space a championship golf course water park and 14 food and beverage outlets the acquisition fits into ryman's strategy of owning large group oriented resorts they already own gaylord convention hotels and management said the deal strengthens its presence in orlando which you know to to maybe bolster their position here Orlando is the nation's largest meeting market for conventions, events, etc., etc.

53:21And we've been to several conferences down in Orlando, so I know that does play true. It creates cross-selling opportunities across its portfolio. The pricing reflects the strength of high-end hospitality assets. And Stephen, I think it was two weeks ago you did the breakdown of the really high-end lodging sector and kind of what people were paying, and we were shocked to hear it. But Ryman said it's buying the resort of the 6.6 cap based on the trailing 12-month cash flow. And if you look at our TREP data, we show the property generated about$109.5 million worth of NOI last year, implying a 7.9 % cap rate if you use the 2025 results.

54:01So, you know, I think for Ryman, they've demonstrated that they're not afraid to make a big, splashy acquisition. and they must have been reading the same data we have been relative to these high-end hospitality assets and they're making a pretty significant push with this deal. You know something strikes me on this which I guess I didn't really fully appreciate Orlando the nation's largest meetings market that's just curious to me like when when did it take over that place when did it get that honor. If you step back and think about it, Stephen, like who is it competing against really? I mean, like New York is, you know, the economic capital, but the cost makes it prohibitive for large scale events for a lot of different industries.

54:49I'm sure Texas is probably on the list there. You know, Dallas is being a pretty big hub for meetings. Orlando is generally good weather. It's, you know, significantly more affordable than some of the other major cities. and it's got a lot of other things to do. I mean, we've had conferences that are actually held at one of the Disney resorts, which was a little bit interesting. You know, Chicago maybe, but with everything that's transpired in Chicago over the last five years, like there's no, nobody's clamoring to be hosting events in Chicago. California cities are cost prohibitive. So, I mean, yeah, you know, if you look at Florida, Orlando is a lot cheaper than Miami and a lot less nightlife to get people to show up to the events every day when they're at conferences versus when people go to Miami and stay out all night.

55:39Okay, so we have some programming notes here before we wrap up. We talked a lot about data centers today and Stephen has a new report out that we released this week, which follows up on the data center deals that we discussed back on episode 410. So we talked about that segment, but then a lot of you asked us for the underlying data. So he dug into the four deals that priced between May and July. And the interesting part was that pricing varied by 55 basis points. One deal held its initial pricing, while the other three had to pay up. And the difference wasn't simply due to AI sentiment. It came down to the credit characteristics of each deal, including leverage and lease terms.

56:27So if you want to understand what's actually driving data center CMBS pricing, reach out to us at podcast.trep.com and we can get you a copy of that full report. Or you can find it on our Trep Inc. LinkedIn or our website under Research and Insights. Coming up later this month, we have a few webinars. So on Thursday, August 20th at 2 p.m. Eastern, we have a live product demo webinar called Connecting AI to Trusted CRE Data, where we're going to dig into our TREP MCP, which is our AI connector for CRE and CMBS data, and show you how we can deliver our trusted CRE data directly within the AI tools that your team already uses.

57:14We've had so many of you reach out to us looking to join this webinar, and we're really excited to walk you through different use cases and really show you how lenders, investors, brokers, and other players in the CRE market are leveraging the tool to accelerate their research, streamline underwriting, and really uncover any actionable insights faster through the tools that their teams have already built and are already using. So there's still time to join us for that webinar. Send us an email to podcast at trep.com and we will get you the link to sign up. Later in the month, we will have our monthly MarketPulse webinar.

57:53And this month, the theme is operating expense pressure, 2026 cash out activity and the student housing outlook just in time for back to school season. So on Thursday, August 27th at 2 p.m. Eastern, we will dig into line item financials. So we'll show you how property revenues and operating expenses are shifting across commercial real estate, which costs are rising fastest and where the greatest pressure on net operating income is emerging. We'll dig into the rise in noteworthy equity distributions from newly securitized loans in 2026 and how today's cash-out activity compares with prior years.

58:35And in that back-to-school special, we'll give you the latest on student housing performance and lending activity and what the data suggests for this sector's outlook. So this webinar is open to everyone. We'd love to have you join us. reach out to us at podcast at trep.com and if you get our emails you should be seeing that invite in your inbox shortly turning to shout outs jacob z told us he just finished listening to episode 412 and would love to join our ai and mcp webinar we got you the link to sign up and we're excited to have you dig in and see what you could do with the tool mitchell k said he's a regular listener of the TREP podcast and a subscriber of our newsletter and was also interested in our webinar.

59:23He said, as a commercial real estate broker who is also building tools for the industry, I'd love to learn how to integrate TREP data. So this is the perfect chance to do so. Randy E was interested in the webcast as well. So thanks for reaching out. William J sent us some ideas for what we can dig into around operating expenses. So thank you for the suggestion and questions, William. We will dig into your email and get back to you over email or on a future episode. Bobby R said he really enjoyed our episode with Larry Kravitz. It was so interesting to hear from a leader who has been in CMBS since the beginning.

1:00:00So another programming note, if you haven't heard our episode with Larry Kravitz, the head of U.S. commercial real estate and CMBS at Nomura, Go check that out. That was a great episode and so interesting to hear all that they're building at the firm. Susan D said she just found the podcast and can't believe there's so much she has to catch up on. She thanked us for the content and the continuous reporting. Margaret T is a huge fan of the show and looks forward to hearing from some of our upcoming guests. We have a pretty jam-packed schedule for August and September with guests from across the CRE and CMBS world.

1:00:39So keep an eye out for our guest episodes. Those typically drop on a Tuesday. So you may have two TREP podcasts to listen to that week, but we appreciate everyone for tuning in to the weekly shows and the guest episodes. Andrew M. said, great episode as always, and was interested in our latest CMBS delinquency report. Stephen R. heard the discussion on the delinquency numbers and was looking for more detail on the individual loans behind the increase. So we sent those over to you. If anyone's ever looking for more from us, please reach out and we'd be happy to share the details. And Daniel C. said, can I get a copy of the latest special servicing report?

1:01:19And thanked us for the show. So as always, thank you to our listeners. We know we're in the throes of August now. We're getting a lot of out-of-offices when we send out our emails. But we appreciate our listeners for continuing to tune in every week. And we have a lot more exciting shows, guests, topics. and events and happenings at TREP coming. So stick around. With that, we'll close. Thanks to our producer, Mariana Sobrana. Join us next week as we look at what's happened during the week and how it may be impacting you. If you have a question or just a comment, send an email to podcast.trep.com and subscribe to the TREP Wire podcast with your favorite provider.

1:01:58Thank you for listening and stay well. All right.

1:02:08Thank you.