The projects that Walker & Dunlop and Dominion have worked on together. And so, Peter, a lot has happened. The last three weeks have been incredibly tumultuous in the real estate capital markets. My phone has been inundated with Walker & Dunlop customers calling me and saying, 'Can we get you on a call with our team to figure out where are rates going? Where's the market going? And should I be running to refinance a property today? Should I be selling a property? Should I be concerned about being able to get out of this property a year or two from now?' And my feeling here has been that for the last three years, we've been in what I've called the hope trade — the waiting and hoping that inflation would come out of the system, that the president would replace the Fed chair with a potentially new Fed chair who would start to cut rates in line with what the president has been calling for for quite some time, and that we would be sitting in a sort of 4% ten-year, maybe even best case 3.5% ten-year. And you were very straightforward in saying at the beginning of 2026, the inflation prints looking good, if you strip out a lot of various things such as housing from the CPI and other inflation prints, that we should start to see cuts, and you predicted that we would see two rate cuts by the Federal Reserve in 2026. You made that prediction not knowing we would invade Iran. You made that prediction thinking that oil would be somewhere between $60 and $70 a barrel rather than $90 and $110 a barrel. Other than the Iran war, what did you get wrong?
Well, that was the main one. Really, that was absolutely the main one. When did the ten-year start really moving? It was as the Iran conflict began and picked up speed. Anybody who's not called you Willie, I think, just called me and asked the same questions. And they probably called me because your phone was busy, so I was the second choice to ask those questions. I can tell you the spirit of what I'm answering people: We are the highest ten-year in now 25 years. That's an abnormal situation. A lot of the people who are saying it's going to stay high forever were the same people in 2000 — not you, but the same people that were in 2018 and '19 saying, 'Yeah, negative interest rates make sense.' And they don't. Negative interest rates don't make totally sense. This doesn't make sense. I think a lot of people, in my view, real estate is an 8 to 12-year investment horizon. And that's why the research I've done tends to focus on just performance over a longer period. You got to be around and allow adjustments. And people are judging it after sort of two to four years. That's like judging who won the NBA championships this past year based on the halftime score. And it wasn't the New York Knicks that won based on halftime scores. By the way, take another example: the Philadelphia Eagles were winning with four and a half minutes to go. Unfortunately, they played four and a half more minutes. That's why they play the full game, and they lost. And what you've got to do with real estate, I think, is play the whole game. Don't judge it at halftime. Don't judge it at the middle of the second quarter.
But let me jump in there for a second. First of all, so everyone knows, my password to access the online copy, the PDF of this Linneman letter is 'Go Birds 2,' which killed me as a Washington Commanders fan and not a Philadelphia fan. So I appreciate the fact that you push those birds all the time. But the issue here is this: many people's game on a specific asset ends in 2026 or ends in 2027. They have a refinancing coming up. They have an investor who wants to recycle capital. So while I get your analogy of, hey, this is an 8 to 10-year game, unfortunately for many people, they don't have another four and a half minutes to play as it relates to what to do with a specific asset — whether to sell right now, whether to refi right now — because the clock is running out on it. And so to them, what do you counsel them to do now that their game is coming to an end on that specific asset?
Don't sell if you don't have to in the current market. Therefore, talk to you and see about arranging a two-year or some such product, a three-year. By the way, and as you know, I'm generally a person that says 7 to 12-year financing on these things, but I play the slope a little bit here right now. And are you exposed as a result of that? Yeah, but you're dead the other way. And better to be dead two years from now than dead today. I'd rather be alive today and the option value of it improves. That's what my advice is.
So you would go on shorter-term floating-rate financing to refi your property today, then go and either prep equity or new equity if you can find it, and put longer-term financing on, even though that would obviously do something to your ownership of that asset?
Yes, that's — and by the way, that's very contrary to my normal view.
And you say that because you think rates are coming down.
I don't see them going up. They're at a 25-year high. Let's take a look at inflation. Why I say that: inflation for the 30 years average inflation prior to COVID, which totally reshocked the debt, right? That's why I'm taking prior to COVID was 2.2% for the things people actually purchased. For the two years prior to the Iranian situation beginning — by situation I mean bombings, etc., right, March — okay, for the two years prior to that was 2.3% for those same items. Okay, we had kind of stabilized. We then had the oil situation. The oil situation has increased inflation. You go from $60 a barrel to $100 a barrel, let's say $118 a barrel. Do the mathematics, and you get a higher inflation rate. But it doesn't stay inflated. Why? Let's suppose oil stays at 100, right? Let's just suppose for the next two years, inflation, assuming oil stays at 100 — what's the inflationary impact of oil over the next two years if it stays at 100? Zero. What will that do to measured inflation? Pull it down.
Yeah. It's a weighted average. Yes, except for the fact, as you and I have discussed, of the 41,000 goods and services that are on the CPI print, 70% of them are impacted by the price of oil. So what you're not showing is that those 70% move up, and then once they move up, you're flat. Then you're flat. That's all I'm saying. I'm just reminding people of mathematics. So even if oil stays up, it will become non-inflationary. Okay? And much less if some reasonable solution occurs and it falls. Now, why is that relevant? What do I think happens when it stops being inflationary? The Fed will at some point lower the rate. And when they lower the short rate, we saw what it looked like in March — this should be in February. If you started adding your sum rate in February, people thought, yep, it's at 4%, short rate stays the same or falls, and that long rate was at 4%. What happens if there's no inflation that the Fed says they have to tame? And I think you go back down, much less if oil prices fall, because that will really cool down inflation. The Fed will declare success, even though they had nothing to do with success, because they don't control the strength of our moves. At least last time I checked. I'm not — by the way, did anyone notice — you got a big audience — did anyone in the audience notice that immediately after the Fed increased rates by 25, the Iranians immediately said, 'Oh, we're repudiating any attempt to go nuclear, we're destroying all our weapons, the Republican Guard is being disbanded, we're opening the Strait of Hormuz, and we're sending lollipops and marshmallows to all American children on their birthdays'? Of course not. They have no effect on those things. And yet they pretended as they did. And that has roiled the market. There's no doubt that it's confused me. It's confused everybody. So it is not Rip Van Winkle. There's an important thing to make here. You and I talk about the federal deficit for years and outstanding debt. There is this silly narrative that for the last 40, 50, 60 years, the investor has been asleep to the fact that the deficit occurs and that the debt outstanding grows, and suddenly like Rip Van Winkle, they awakened and saw the deficit and scrambled and said, 'Oh, we've got to increase the rate we demand.' That's silly. It's just intellectually silly. If it were true, why is it that every country has about the same increase in the long rate, rather than being very nuanced depending on their own fiscal situation? So that's why I want to go short. I don't want to go short — let me clarify that. I don't — if you know, it's like I don't want to give rent concessions. I really don't. But the reason I give rent concessions is to get me from here to there.
That's the same thing on the people who have to finance. If I had listened to the Walker webcast in June when you said they're going to do two rate cuts in the subsequent two meetings — you actually said that they weren't going to cut in July, they were going to come back in September. So if I listened to you then, I said, 'Great, I'm going to float because Peter says that rates are going down. Surely the curve's going to go down. I can play a SOFR plus 200 loan and I'm going to do well.' Well, if I followed you though then, it's cost me 100 basis points between that point and today to have floated rather than fixed in a long-term option at that point. So my only question to you is, what confidence should we have? And obviously, I'm asking you to try and predict rates, and nobody can predict rates. But why is it, do you think, that we are at the all-time high when if I listened to you four months ago, I probably would have made the wrong move?
Well, because of how dramatically it's moved since then, right? I mean, I don't know how else to say it more obviously.
I don't think there's anybody — I don't know of anybody — I'm sure there's some hedge fund somewhere — I don't think anybody when we talked in June would have ever predicted, I don't know, we're at today 5.35 for a ten-year.
I don't think anybody would have picked it. Right? You might have had a few people saying, 'Oh, upper fours and five.' And so the fact is, it's run — it's just run to a point that's not justified by the economics. That's the simplest way to say it. And I think even the Fed — and I stress even — can figure that out.
So as you look forward, you just talked about sort of, if you will, a normalized — we get out of Iran, oil comes down, rates come down, the Fed doesn't see inflation pushing rates and therefore they can cut. Do you think that there's a scenario where something breaks? And if you think that something breaks to get them to cut, what do you think breaks?
Well, obviously something dramatic in Iran. And it's interesting, you're saying the same stuff I did. In the first few months, every little announcement, every head fake on looks like we made progress, oil would drop like $78 and then it would go back up, and the rate was kind of bouncing with it. And there's almost — people have gotten desensitized to that information, and it's kind of staying at 100 and or a little above 100 a barrel, and the rate saying, 'I don't want to screw around and figure this out. I'm insuring myself against the Fed being stupid.' I think it's actually come down to the low 90s in the last couple days because of the release of the G7 oil reserves. Which I would underscore the fact that China has released so much of their reserves and lowered their oil consumption since the beginning of the Iranian conflict, and now the release of the reserves on the G7 is a big reason why we actually have $90 barrel oil and not $120 barrel oil. And not something that is sustainable, at least on the G7 side of releasing reserves. The Chinese very clearly seem to have built up enough reserves and lowered their demand for oil to be able to stay in their current demand situation for quite some time. And it's kind of ironic that we can look to thank the Chinese for what they've done to keeping oil prices [down]. Last night I filled up on Sunday and I was $4.10 a gallon. I'm not at $4.10 a gallon in Denver, Colorado, if the Chinese haven't pulled back their oil consumption on a global basis as significantly as they have.
But Peter, let's switch for a moment to unemployment because I think that a lot of what you're focused on is GDP growth and the overall acceleration of the economy. GDP growth — you're actually looking at '26 GDP growth higher than what you'd predicted it would be in '25. But on the unemployment side, your unemployment calculation is about almost 2x what the official unemployment numbers are. Talk about that latency or that underemployment in the economy and whether that is people who have permanently retired — so they've, they're a 55-year-old person who's just removed himself or herself from the labor force and therefore generally speaking is just going to be a drag on the economy going forward. No criticism of someone who's my age who's decided to step out of the labor force — but or is someone who is 22 years old living with mom and dad and sitting there not employed but has the ability to come off the sidelines and enter the labor force that actually could give us lift from a GDP standpoint?
The reason my number, so to speak, is higher is all it is is saying, look, there's a — just to remind people, the unemployment rate comes from a survey that says, are you employed? And if you answer yes, fine. And if you answer no, they ask you the next question: are you actively looking for work? That's it. And the numerator is those who say they're not employed but actively looking for work. The denominator is the sum of employed and those who say they're actively looking. It's that simple. What it misses are: I'm not looking, I'm not employed, and I'm not actively looking. And that is still fairly high, and it's a bit abnormally high. And there's a whole field discussing why that is the case, ranging from rank entitlement to parents that spoil their kids to there are no opportunities to — there's a lot of labor out there and it's not working. It's not unemployed. It's not because of wages not occurring. There's a lot of wage activity that's occurred. I don't think anyone, including me, fully understands why it's as high as it is, but it is. And it's a drag on the economy. It is a drag on the economy.
So it's a drag, not a potential catalyzer of significant economic growth.
Could be, but you have to wonder why are they not there after such a prolonged period? Because, as you know, it's been up there for some time. It's, you know, the recovery has occurred. We went through a very hot period. Wages have been up in real terms, etc. It probably is much more a drag. Now it's interesting, unemployment — the numbers are really strong. If you look at people getting fired, because nobody's getting fired. If you look at unemployment insurance claims, new unemployment insurance claims, nobody's getting fired. That's great, right? That's a strong labor market. If you look at new jobs created, it appears fairly mediocre. Kind of a run rate of about 900,000 this year, which is not terrific.
It's actually super slow.
And in comparison to the last — I mean, since ChatGPT was launched in November of 2022, we've added 4.6 million jobs, I believe, and that goes to like 2.1, 1.8 in '23 and '24, and then '25 is somewhere around 700 or 800,000. And if you add up all jobs in Trump 2, it's a whopping 700,000 jobs. Let's go wide. Let's assume it's going to end this year around 800,000. Not a great number, not a great cover number. We're down probably two to two and a half million legal and illegal immigrants that came in in a particular year. If you say half of those would have gotten a job, which is not wildly off, that would have been another million workers. If you add the million workers to the 800,000, you'd say a million eight. Not spectacular, not bad, right? But absent the immigration, there's a whole bunch of people not looking for work. Just not looking for work. And you say, 'Well, why doesn't that —' Remember, immigrants get jobs and they take jobs, but that also increases the demand for jobs. Why? They buy sweaters, shoes — they ride Brooks nine-and-a-half shoes, for example. You know, people say, 'Oh, immigrants come in and it's an additional supply of labor that destroys and takes away opportunity.' What they forget, and what the research shows, is they also get jobs and they do what everybody else does when they get a job. They add to the US economy's productivity. They buy sweaters, shoes — it creates jobs.
But yeah, I get that. But that would show up then in inflationary pressure on wages.
No. No. Because roughly there's a positive net effect, but it's not huge. And the reason is their supply creates its own demand. Just think of immigrants as a new country of two and a half million people. Okay? New country comes in, the two and a half million people go to work, the two and a half million people get hired because they make stuff that they want. There's a circularity to it, and we're not giving them — so we drop them out, right? And you're right, we lose the labor side, but we also lose the labor demand side because they're not buying as many shoes and they're not buying as many — you know, just go through the list. So the demand is also down. That's why you don't see the spikes.
And you made this great point very articulately, I think, at your MIT presentation — I'm doing this from memory — that you get asked everywhere, are there labor shortages? And basically the answer, and it's the same answer the hotel people give, not terribly. A little here, a little there, not terribly. And that's because yes, the supply got reduced, but the demand got reduced because those immigrants were getting jobs and working and buying stuff and that coursed through the economy. So we've dropped the demand and we dropped the supply. Let's switch then to a sector that is using a lot of labor right now, which is data centers. If we back up to January of 2025, you and I were together at the Constitution Center in Philadelphia, and I said to you, 'What is the best sector for you to invest in?' And you said, 'Well, if I want to stay rich, I'd invest in multifamily.' I then said to you, 'Well, what's the get-rich asset class?' And you said, 'Office.' And then I said, 'Well, what's the get-poor asset class?' And you said, 'Data centers.' So in this quarterly letter, for the first time ever, Peter has these canaries in a coal mine — to anyone in the room who doesn't read the Linneman letter on a quarterly basis — and he goes through and typically has 55 different opportunities for a canary to be dead as sort of a leading indicator of trouble. I think the lowest it's ever been since I've been reading the letter, Peter, is like two or three canaries. I don't think you've ever had zero canaries.
That over the last three years it's gone from 12 to 16 to 18 in the most current Linneman letter — 18 out of 55 potential dead canaries. But this is the first time you've had data centers as a dead canary. And on each one of these sectors, if you will, there can be a potential of five canaries. So Peter's just starting to feel, if you will, the vapors coming out of the coal mine, or the canary hasn't shown that we've got a real issue down in the coal mine yet, but it's coming along. Peter, the growth in data centers — we've gone from, you know, in the last five months, we've gone from 570 data centers in the United States to something like 1,300. We've got 2,000 data centers planned to be built. So with that amount of capital — and the hyperscalers are putting $800 billion into data centers in 2026 — so with that amount of capital going to it, do you think something breaks as it relates to just there's default, there's not demand, something happens in the data center AI construct that breaks? And then the follow-up to that is, if all that capital continues to flow there and you don't have anything break there, what is suffering from the amount of capital that's going to data centers that would in a traditional way be flowing going to other asset classes?
Remember how good apartment demand was, rent-occupancy-wise, in what, 2021, first half of '22, something like that?
Boy, do I remember it fondly.
No supply. No supply because COVID had shut it off, right? And demand — pretty robust demand from credit tenants, namely residents. Pretty credit, right? They pay their rent mostly. How did that end? That ended by a huge supply spurt for a lot of reasons. You still have demand, and the supply spurt overwhelmed. What's my concern on data centers? You see anything that gives that good a yield, sooner or later it's going to get overbuilt. I don't know when. And how do I know something? If you look at what's happening in data centers — this is overly broad — everybody says these hyperscalers are credit, you know, credit tenants, credit. And by the way, you're old enough to remember just barely when people were building shopping centers saying Sears and Kmart were credit tenants. And by the time they opened, or a year or two after they opened, they're — one's dead and the other's almost dead, right? Eroded for a lot of reasons, competitive and otherwise. And I just — so here, if one of the hyperscalers came to one of our developer friends and said, 'Build me an office building or build me a warehouse,' what kind of yield on cost would they negotiate it down to on a 15-year lease? Six, seven?
Let's just say six to seven percent yield on cost.
Because they're who they are, right? They're going to negotiate it down, maybe even a little lower. Okay, on a data center, what are they doing it for? Ten. And by the way, a nine, nine —
So what does that tell you? It tells you they're desperate for that product and they're doing something they normally would not do. No credit tenant would normally do it. Why? They view they're in a race that challenges their existence, which means they're afraid they go out of business if they don't win. Okay? That's why they're doing it. They want to get it. And we know they're paying a premium. Now, by the way, let's go to the other capital part. Let's go the capital part. Lenders. Lenders are lending 90%, 85–90% first mortgages. Okay? Why? Because, oh, they're credit tenant, even though the tenants are worried about their existence. That's why they're willing to pay that rent. And you go, hm, that's interesting. So the lenders are viewing it as no risk. How has that generally turned out in history, that when lenders view something as having no risk, it turns out well when they provide 90% debt or 85% debt? Why are they doing it? They have no exposure to it currently in their portfolio. They got a lot of apartment exposure, got a lot of warehouse exposure, they have a lot of office exposure, a lot of this. So they all say, 'Okay, we're going to go get it and we're going to get it while the getting's good.' So they're flooding into it and they're doing something we don't. So we see tenants doing something we don't normally see. We're seeing lenders do it. And then we see equity providers saying, 'Hm, I got crushed in real estate in the last three and a half years on everything else. Let me try there.' Especially with 90% leverage. And by the way, the developer makes the development fee. They have to work hard for it. I'm not taking anything away from it, but the developer's development fee with 90% leverage, they can't lose as the developer. So I think it gets overdeveloped in a strange way. Remember the office market in the 1980s? You had very good demand growth, and yet somehow vacancy got to 18% in spite of great demand growth. That's because unbelievable lending, right? That's why I have concerns. If the tenants are worried about — just think of this for a second — the tenants are worried about their existence. Okay, why aren't the rest of the capital stack worried about their existence?
Well, they also though — I mean, they have multi-trillion-dollar market caps. I mean, you know, I hear you about their quote-unquote worried about their existence. And by the way, I mean, I would — your concerns as it relates to the hyperscalers, if you listen to the All-In podcast, which I listen to on a weekly basis, so just — I'm just stealing from the guys from All-In. This is not my own research, but I listened very intently last week. And in their discussion last week, they talked about a couple things that concerned me on exactly this topic, Peter. The first one was they looked at frontier models versus open models. And 90 days ago, frontier models were 80% of AI usage in the United States and open models were 20% of AI usage in the United States. 90 days ago. Today, it's the inverse. It's 20% frontier models and 80% open models. So at Walker & Dunlop, do you think that has anything to do with why they're now saying, 'Oh, let us get together, let us get together with the government and regulate and set up barriers' — who think they're going to set up barriers against —
1,000%. It's called regulatory capture. It's let's put up barriers so that we can create a whatever monopoly, duopoly, triopoly you want to put together. But let me keep going on this for two seconds, because at Walker & Dunlop we use ChatGPT. We have a closed environment of ChatGPT that makes it so that we can keep all of our data, run all of our analysis inside of our own —
environment. It's protected and I love it except for the fact that we spend a tremendous amount of money on tokens with ChatGPT and my CTO is going to come to me sometime in the next 6 to 12 months and say if we moved from ChatGPT to an open model, we pulled a server and put all of this on our own servers, we would drop our token usage from X to Y and we would save Walker & Dunlop X millions of dollars a year. And by the way, we don't need to be using frontier models at Walker & Dunlop. I'm trying to process mortgages. For everyone in this room, if I were trying to solve cancer, I got to stay on ChatGPT or Anthropic and keep on that frontier model. But for W&D, my big issue right now is not necessarily the capabilities of the technology. It's the implementation challenges I have of taking the technology and putting it into our work streams to make it so I can process loans for all of you quicker, more efficiently, and more cheaply.
And so as I think about that shift in just the last 90 days of the frontier models, you listen to the All-In hosts talk about it and they sit there and they talk about if these frontier models don't stay 6 months in front of the rest, their value goes to zero. Yeah, that's their words, not mine. That's their words.
That's the existence concern.
