The AI Demand Is Real. The Accounting Games Are Growing. How Long Can Both Be True?

0:00 Cold open — the through-line, and a running gag about Jack the action hero

• Zeigler's framing of the format: "I am endlessly amazed by the through lines that emerge when we pull these up. I can't believe how connected some of these clips are."
• Roughly the first three minutes are a bit about a viewer comment ("Jack comes across as a grizzled veteran boxer coming off that private jet") on an AI-generated Last Call intro video. No content — skip to 3:23.
• The week's slate is named at 3:23: Dan Niles (first appearance), Ben Hunt (second episode of Why Am I Reading This Now?), and Cameron Dawson + Dave Nadig for Clip Beta.

3:52 The setup: "these are the smartest people in the world — how could they be wrong?"

• Forehand states the bull argument in its strongest form before attacking it: "there's very very smart people running all these companies… they're spending massive massive amounts of money, but the people that know the most about this are spending the most money. How could they be wrong?"
• That is the question actually put to Niles on the interview, and the clip that follows is his answer.

4:16 Niles' answer: pull the Cisco release from May 2021

• The specific, checkable rebuttal: Cisco's own CEO reporting bookings going "from 70% year-over-year growth in several months to negative 30%" — at a company that had been "the most valuable company in the world."
• Niles concedes everything before disagreeing: "are these really really smart companies? Absolutely. Are they the best companies that we have right now? 100%. But do massively smart, big companies get it wrong? Yes, all the time."
• On bubbles as an asset class: "they're great to be invested in. You can make a lot of money if you stay with them, but obviously you have to get the timing pretty well, because you never know — is this the actual break or is this just a pause that refreshes?" He dates his own two calls: a June 20 note flagging "a speed bump," and a July 29 note saying "we're near the bottom."

5:47 The three phases — training, inference, agentic — and a dated clock

• The framing that both hosts single out as the most useful of the week: AI has run through distinct phases — "you had the training phase initially, then you had the inference phase, and then on January 30th of this year, you had the agentic phase."
• The quantitative claim underneath it: "the agentic phase uses 10 to 100 times more tokens than the chat-based AI phase."
• The conclusion, with a leash on it: "I think you've got a long way to go, at least another year for stocks to go higher. But… like what I said about the Chinese memory companies, I do think you're going to have to get more selective and then watch the data like a hawk to make sure that you're not deluding yourself either by being too negative or being too positive."

6:55 Forehand's self-diagnosis: the 90s CEO would have sounded just as convincing

• "I wish there were podcasts back in the 90s where I could have listened to the CEO of Cisco doing the same thing and I would have been like, this makes complete sense."
• The inference: "everybody gets trapped in this bubble and everybody sees what's in front of their eyes. And that doesn't necessarily mean that we're not in a bubble just because all these tech people are saying very positive things."
• Zeigler labels it: "it's the Bob Farrell-ism. It's a common sentiment thing… people saying the same thing or the same version of the same thing. That's the biggest contrarian itch I feel when I look at these S-1s, when I look at these companies going public, when I listen to stuff off of the earnings calls."

8:23 The discipline: concede there may be an intelligent person on the other side

• The episode's cleanest piece of method, from Forehand: "it's easy to look at these arguments and… just say, 'oh, those guys are over optimistic. They're idiots.' No, they're smart people… you should concede the idea that there might be an intelligent person on the other side of this and they might be right."
• And symmetrically: "this doesn't mean the tech guys are wrong either. It just means you need to balance both sides and understand the facts on both sides, because there are things that are different about AI."
• The two things named as genuinely different from 1999: "there is massive demand for it now, which is very different in the late '90s where we were building in advance of demand," and "there are things like governors that are limiting it that might make it not as much of a bubble as it could be."
• Self-aware caveat: "I can't sit here and listen to the All-In podcast and drink the Kool-Aid the whole time and think this is going on forever, because that's just not accurate."

9:38 Zeigler's field evidence: clients are feeling the agentic lift this week

• The anecdote that makes the Niles clock actionable rather than rhetorical: "I'm seeing this talking to clients, talking to companies… they are just now feeling the agentic lift in a bunch of areas… that happened literally earlier this week talking to a client."
• Why that implies a second capex leg: a customer who just saw the return "is willing to spend more budget… on doubling down on some of these tools."
• Forehand agrees against his own prior: "this is the part where I do drink the Kool-Aid a little bit, because… the Excess Returns podcast has seen an increase in the ROI in the agentic phase." He caps it: "it doesn't mean we're not in a bubble. It doesn't mean this whole thing doesn't end badly. It just means we are probably going to start seeing those results."
• The adoption-lag analogy: "it took forever for anyone to adopt Excel 1997… what you and I are doing is not what the average person does." Zeigler ties it to crossing the chasm: if the late adopters in software start seeing returns, "there's another investment cycle that now follows."

11:56 Ben Hunt on credibility: the teacup you can glue but never repair

• The professional-ethics version first: "in this business your reputation is everything… there will always be places where there'll be a temptation to cut corners… and you can't do it. There is no possible advantage on a trade or with a client that should ever let you do anything that chips at your reputation."
• The image: "you break or you chip the teacup and you can glue it back together and it can still be a functional teacup, but it's never the same."
• The generalization he then applies to policy: "whether you're a Fed chair, whether you're a financial adviser… credibility, your believability, your reputation is the most important thing."

13:54 Where the teacup broke: Warsh's end-of-July press conference

• "The teacup got broken with Kevin Warsh at his press conference… at the end of July, where they didn't hike rates. He was talking like he was going to hike rates and they didn't do it."
• The counterfactual: "it would have been easy for him to do, and then if he had, we wouldn't have any of this now. You would be cemented in everyone's perception as… an inflation fighter… and that's when the market does the work for you, as he likes to say."
• What was heard instead: "new boss same as the old boss, lots of words, lots of talk, but they will never do anything to raise rates if they can possibly help it — then that's when gold skyrocketed."
• The bind: Jackson Hole "last Friday" was correctly read as "a very hawkish speech," but "it'll just be more words if he doesn't actually pull the trigger on hiking rates in September" — and "no one in this administration or in markets wants him to raise rates."

16:42 The Volcker pedestal, and why action is now the only credibility currency

• Zeigler: "there's Paul Volcker and then basically everybody else… that is the pedestal we look to when somebody talks a big game. And nobody has talked a big game and delivered on those words since Paul Volcker."
• The transition read: "we were getting all these things about rates should be higher, I would hike, the critical stance — okay, here's your chance to be this much Paul Volcker. And he didn't do it.… He is seven steps down. And so if he does have to hike, even if he does have to cut, it's going to be viewed in a very different way by market participants."
• Forehand supplies the mechanism: "in a world of no forward guidance, the way you maintain this credibility now is action… the only way you show seriousness is action." Even a token move would have done it: "he didn't have to do anything that would break the market. He could have turned around and cut rates at the next meeting… and that would have still preserved more credibility in his seat."
• And it is priced: "when we look at the price response in other categories, mainly in gold… this has felt across it in a way that you can actually map into prices."

19:23 Cameron Dawson: financial shenanigans are cyclical, and they cluster at the crest

• The organizing claim, delivered in front of the Financial Shenanigans book on her shelf: "you typically don't see shenanigans as bad, as prolific, as widespread when things are in the earlier stages of an economic boom. You see the most shenanigans at the end, at the crest, as you're going over the apex… then things start to clean up. You have credit issues get washed out, and then you reemerge the other side."
• The structural point: "the shenanigans that are popping up are all over the three different financial statements" — income statement, cash flow, balance sheet.

20:47 Exhibit A — Nvidia's receivables against the hyperscalers' payables

• "Nvidia extended its accounts receivables in order to help its customers. Well, its customers are the hyperscalers who had extended accounts payables, which flatters their free cash flow."
• The netting: it "made Nvidia's free cash flow look worse, but it made the hyperscalers' free cash flow look better… if you then normalize that, we would have probably seen even more negative free cash flow numbers coming out of the hyperscalers."
• Note what this is: a cross-company cross-check. The same transaction shows up with opposite signs on the two sides, so the pair is checkable even when either side alone looks explicable.

21:37 Exhibit B — capex reclassification, and the SPVs that never appear at all

• "Microsoft lowered its capex guidance because it reclassified financial and operating leases that move certain parts of its capex down from operating cash flow into investing cash flow… the capex calculation that they use looked to be lower than it actually was."
• "Then you add on top of that, there's a bunch of stuff that's not even hitting the balance sheet or cash flow statements because it's all happening in these SPVs… super esoteric and very murky, and we don't know what's going on in them, we don't know all the different players, and we know that they're private."
• The heuristic she draws from the pattern — the single most portable line in the episode: "as you start to see more games being played, and then the clarion calls saying no, it's not circular financing, it just ends up causing us to raise a lot more eyebrows… If things were as great as you say that they are, you should not have to be playing the games that you are playing."

22:33 Exhibit C — disclosure opacity, the equity method, and the 2027 comp

• Segment opacity: an FT piece "effectively saying Microsoft has done a terrible job at disclosing its AI business and it hides it in with a bunch of other businesses."
• The accounting mechanic worth knowing: two ways to carry an investment in another firm, and "one is called the equity method, which has to be consolidated into the income statement… Microsoft owns 25% of OpenAI and so it accounts for it within the equity method. We're going to be having to see a lot more disclosures about these companies over time."
• The dated consequence: one-time securities gains "are skewing all of these GAAP headline earnings numbers… as we go into '27, we could be talking about negative year-over-year growth in a quarter like the second quarter, simply because you have a big huge gap of missing one-time gains." Google and Amazon report GAAP only, so they carry the distortion straight into the headline.
• Her closing line: "accounting is not something that people care about till the peak of the cycle."

24:23 The counterargument the hosts steelman: Damodaran on young growth companies

• Forehand puts the other side up: Damodaran's point "about like who cares about how depreciation is being done for a young growth company, that should be not even the top 10 of things you consider."
• His resolution: "if the growth keeps going on and if it's legitimate growth, this stuff probably will go to the wayside for a period of time, but we also know it's going to escalate to a point that it's going to be a problem… when there's money in front of people and there's a big revolution like we're having, it happens every single time."

25:48 Zeigler's reading method: turn the knob all the way to 11 on each statement

• How he learns a system: "I learned this with music equipment… I have to basically take a knob and turn it all the way up and all the way down to start to understand what it does."
• Applied to Financial Shenanigans: the book walks the income statement, then the cash flow statement, then the balance sheet, and on each one shows "how to turn that knob extra Spinal Tap all the way up to 11… you start to see the examples of what can a company do when they need to push that extreme side of it to see, can we get just a little more out of it."
• Why that explains the escalation: "we have all these examples across the income and the cash flow statement and to a degree the balance sheet of push that border to just get a little bit further in the next quarter… that push is going to lead to a bad ending at some point."
• Forehand's version of the same worry, with an absurdity limit: "eventually we're going to get to a point where it's like we will not depreciate H100s because they will live forever… but along the way the knob has been slowly slowly turning."

28:18 Niles again: "buy and hold forever is completely moronic"

• The claim: "saying there's some stocks you just need to buy and hold is completely moronic, because you don't know that you have a lot of survivability."
• The evidence is an absence: "you hear from the people who said, 'oh yeah, Apple was a buy and hold.' Well, you never have people on that say AOL was a buy and hold. Yahoo was a buy and hold. Nokia was a buy and hold. Cisco was a buy and hold… IBM was a buy and hold. The list is really long of market share leaders that then went into trouble."
• The one survivor, named as an exception rather than a rule: "Microsoft has done great through three different decades, right? But that's one company."
• The rule he substitutes: "you have to have strong conviction but loosely held… a very flexible open mind, and as the facts change, you want to change too." His own live counterexample is Nike; his heirloom-stock counterexample is Disney ("put it away for your grandkids… that hasn't worked out").

29:40 Survivorship bias, sell discipline, and the books that already say so

• Forehand: "there's a bunch of survivorship bias in this idea that you just needed to buy and hold Amazon… it's great for the people who got it right, but… if you got the wrong companies in the drawer, you got a big problem."
• Zeigler reaches for two books physically within arm's reach — Chris Mayer's 100 Baggers and Ian Cassel's stock-picker book about giant holdings from micro caps — and points out that neither author is a buy-and-hold-forever person: "even if you hold it and you get the hundred bagger, there's probably still a time to sell.… Sell discipline."
• The structural argument that closes it: "the people we point to, the Warren Buffetts… even they say no, please don't do that. Even your S&P 500 index fund does not do that. It changes, it reconstitutes, it rebalances, it reweights."
• And the mortality argument: "at some point somebody else is going to sell it for you because you're dead. So you might as well have some awareness of this while you're alive." The estate version: "it is no small miracle that my grandmother… doesn't have a bunch of scam stocks… forever there to be cleaned up by… the next generation."

33:21 Hunt's breakthrough: narratives have a measurable life cycle

• The upgrade: "now we're not just able to track how loud or how quiet a given narrative is but we're really able to track its life cycle, because all narratives have life cycles."
• The stages: a burst when a new story arrives, which "may or may not lead to a change in what I like to call common knowledge — what everybody knows that everybody knows"; then contested ("people are arguing whether that's true"); a possible/building stage; and finally confirmed.
• Why staging matters more than volume: "if you know where you are in the life cycle of a narrative, then you're much better off to be able to say, well, how does that impact prices or policies that have to come out next?"
• The signature being tracked in the shown slide is "central bank losing credibility," filtered to the US, over five years — and what it caught in late July was "an enormous burst like a supernova… that transformed almost immediately into a confirmed narrative regime of the Fed has lost credibility."

34:38 How it is measured: semantics, not word counts

• The input: "we're reading all the news in the world, all of it overnight… looking for the presence of somebody saying, 'okay, the Fed's losing credibility.' This isn't word search. This isn't sentiment analysis. We're able to track the underlying… alert semantics."
• Zeigler draws the contrast with the old word clouds: "that's a generic version of sentiment analysis where you just go, how many times did positive words show up… semantics are actually understanding the sentence construction, the argument construction — how did they position it when they talked about inflation, whether it showed up a hundred times or not at all?"
• Then a density score across venues: "oh, it showed up in the Journal. Oh, it's all over Reddit… that narrative density stacks up with a semantic signature that can say not just is this high or low relative to history, but how is it moving. Is it all of a sudden screaming?"
• On actionability: "there's actionability and it's pretty much always like a step removed. You as the person figuring out what the actions are need to understand what's correlated to it to then make the decision." Hunt has separately argued (in an earlier Excess Returns episode, "narrative as an investing factor") that this meets the Larry Swedroe criteria for a factor.

39:10 The discovery: split central banks by country, and gold explains itself

• The methodological fix that unlocked it: "at first they were just tracking central bank credibility… They realized they shouldn't be looking at central banks as a whole. They should be looking at central banks by country."
• Then the correlation appeared: "this helps explain, for example, the price of gold. Because… gold is basically the Brent Donnelly definition of like one divided by trust… gold is going up when people don't trust central banks."
• Three episodes mapped onto one series: (1) Trump attacking Powell produced "a burst of just negative sentiment… that led to a lack of credibility for the Federal Reserve… that was the run-up in gold a little over a year ago"; (2) once "Powell was on the way out and Warsh was on the way in, all of that lost credibility… recovered," and "the price of gold basically goes sideways"; (3) into and after Jackson Hole, "we've got another burst of it reversing again, because when Warsh didn't make a move… he is not a Volcker… and now everybody knows that everybody knows it. Gold starts moving again."

41:11 Dave Nadig: the danger is the grind, not the blow-up

• The worry stated inside out: "I worry that we're not going to have something that is catastrophic and horrific that shakes everybody out and gets everybody smart."
• The mechanism: "it is rarely the big loss that's the problem… It's not that they're losing it all on a Super Bowl bet. It's that they're betting on a hundred baseball games a summer and they're slowly losing 10% a week… then they have a 20% win and they feel like heroes and the cycle repeats."
• Why the shape matters: "that kind of grinding despair in a financial situation is much harder to recover from than the one big mistake… I feel like this is a grinding tax on math literacy that we're not going to be able to turn around just by willing it."

43:09 The market-structure worry: gambling and investing in one app, soon one exchange

• Zeigler is explicit that this is not prohibitionist: "I am not anti-gambling… I fully believe people should be allowed to worsen their situation if it's genuinely what they want to do."
• The actual concern is venue: "all these products live right next to financial products literally in the same Robinhood app, or pretty soon on the same exchange at the CME… if there's these qualities of gambling risk that are different than investment risk, polluting those against each other just seems like the opportunity for problems."
• The normalization point: from "sports betting is taboo… 10 years ago" to "the play-by-play announcer is telling you parlays you can do, it's flashing on the side of the screen… the studio host that you've trusted the whole time" is now the one recommending the bet.

44:43 The same shape inside a portfolio: small accumulating losses are the deadly ones

• Forehand generalizes Nadig's point to strategy design: "any strategy where the losses are just very small and they accumulate over time, those are the most deadly strategies, because you don't see that moment where you feel the loss."
• The named example: "overtrading your account… you're hurting yourself a little bit over time, and when you don't see that, sometimes by the time you see it it's too late."

45:31 Zeigler's planning test: is it coming out of income or out of assets?

• The first question he asks a client doing anything speculative — "and this is both with investing and gambling, and… we'll throw trading in there too": "are you doing this out of income? Is this coming out of cash flow? Or is this coming out of assets?"
• Why it separates a habit from a problem: a scratch ticket bought "on the way home" after payday is "additive income, not a big deal"; "if however you're emptying out your savings account to buy six-packs and scratchy lotto, it's a different type of problem."
• The second-order test is recovery: "if you're spending assets down but income is coming in to replace it… it might be papering over the hole. That could work all the way up until you retire and then that could sandbag your whole thing."
• His forecast for his own practice: "risk capital for things like gambling is going to become as important and as frequent a conversation in the next 10 years of my career as concentrated stock was" — the same psychology, the same question ("that number's too high; what's the right number for you to start to back this off?"), and the same failure mode: "it can make for an irrecoverable loss."

48:20 Close

• The takeaway they give themselves: "be careful with the gambling, people. That's our takeaway here."
• Housekeeping: the Excess Returns Substack, the full network at excessreturnspod.com, and the standard disclaimer — "no information on this podcast should be construed as investment advice. Securities discussed in the podcast may be holdings of the firms of the hosts or their clients."