Perkins and Beamish discuss AI CapEx peak, credit stress, and macro regime shifts amid Nasdaq volatility
Summary
Dario Perkins and Freya Beamish analyze the current market environment, citing peaking AI capital expenditure, high-yield credit stress, and collapsing token prices as key risks. Perkins argues that while the technology remains valuable, the industry faces commoditization and that a crash would likely be a demand shock rather than an inflationary supply shock. Beamish highlights a macro regime driven by re-leveraging into tech amid negative supply shocks in the non-tech economy, noting that rising interest rates are now exposing leveraged balance sheets. The speakers discuss historical parallels, such as the dot-com era's market concentration, and contrast US dynamics with diverging trends in the UK and China. They conclude that while systemic risk may be limited due to hyperscaler safety, significant rate cuts could alter global capital flows and currency valuations.
Key takeaways
- 1 In early October, he told us about a report he was going to write and on what the AI Capex crash would look like. NASDAQ drops 4%. four percent. He then published said report on the twenty second of October and we lost a further four percent.
Unidentified speaker (speaker_0) · 1:05
- 2 the CapEx is now clearly peaking... these companies are saying, well, they're you know beginning to revise down the estimates for next year very early
Unidentified speaker (speaker_1) · 3:47
- 3 Companies that leveraged up rapidly and took on significant debt are now exposed to rising interest rates.
Unidentified speaker (speaker_2) · 15:21
Episode sections
- 0:02 – 3:23
Intro and Dario's Blog Success
Andrew Slazenger introduces the topic of an AI crash, noting a 9% Nasdaq drop and $1 trillion daily loss. He asks Dario Perkins to apologize for his accurate prediction of a 4% drop in October, which Dario declines, citing his blog's success.
- 3:23 – 7:27
Market Signals and Fed Policy
Dario Perkins identifies earnings misses, peaking CapEx, and credit stress as troubling signs. He argues that interest rates matter, citing high yield stress and praising Fed Chair Walsh for being less ego-driven and more transparent.
- 7:27 – 9:31
AI Commoditization and Historical Parallels
Perkins compares the current AI boom to historical tech revolutions, suggesting that while the technology remains, the builders may not profit due to commoditization and collapsing token prices. He mentions using Claude AI to rewrite his blog.
- 9:31 – 18:57
Freya Beamish on Macro Regime and Leverage
Beamish clarifies their bearish stance is based on macro mechanics, not tech trust. She describes a 'mini-cycle' driven by re-leveraging into tech amid negative supply shocks in the non-tech economy, noting that Fed rate hikes are now exposing leveraged balance sheets.
- 18:57 – 21:43
Soft Landing and CapEx Levels
Perkins questions if a 'soft landing' for CapEx exists, noting that reducing CapEx risks revenue contraction in the ecosystem. He highlights that 82% of industry revenue comes from CapEx recycling, making a transition to consumer/business revenue difficult.
- 21:43 – 27:15
Systemic Risk and GDP Impact
Perkins argues against a subprime-style disaster, stating that tech CapEx leaks into global supply chains (Korea/Taiwan) but cancels out in GDP calculations. He notes that while standalone model developers may default, hyperscalers are safe, limiting systemic risk.
- 27:15 – 28:28
Yields and Recession Outlook
Perkins predicts yields will fall in an AI crash because it is a demand shock, not an inflation-driven supply shock. He believes the Fed has room to cut rates, unlike in the 1970s or during COVID, and expects no massive fiscal response.
- 28:28 – 29:11
Crowding out and rate cuts
Speaker 1 argues that high interest rates have crowded out sectors like housing and consumer durables. They believe lower rates would crowd these sectors back in, allowing the Fed to cut rates significantly with a quick economic recovery.
- 29:11 – 30:09
Wage growth and labor market signals
Speaker 2 notes that wage growth is bottoming out in the US, excluding education and health sectors. They highlight that job-to-job transfer wage growth is picking up, which serves as a leading indicator for broader labor market improvement.
- 30:09 – 31:28
AI spillovers and unit labor costs
Speaker 2 identifies construction as a sector seeing positive spillovers from AI demand, exacerbated by reduced labor supply due to de-immigration. They warn that rising unit labor costs could pressure profit growth, potentially shifting profits rather than expanding them.
- 31:28 – 32:59
UK economic divergence
Speaker 2 contrasts the US with the UK, where wage growth has dropped and productivity growth has surprised to the upside. This creates a favorable profit cycle for UK companies to invest, provided energy prices do not remain structurally high.
- 32:59 – 33:48
US fiscal constraints post-midterms
Speaker 2 asserts that significant US fiscal expansion is unlikely regardless of midterm outcomes, citing the reversal of tariff revenue promises that previously supported fiscal hawks. They note no major changes regarding France's fiscal situation.
- 33:48 – 34:49
US tech concentration and decoupling theory
Speaker 1 compares current US tech concentration (45% of stock market) to the dot-com era (40% threshold for negative outcomes). They theorize that decoupling is plausible because other markets lack similar valuations, except for tech supply chain nations like Korea and Taiwan.
- 34:49 – 42:59
Investor instinct vs. rate cut benefits
Speaker 1 suggests that while decoupling is theoretically sound, investor instinct in a crash would likely be to sell equities for bonds. However, they argue that global rate cuts following Fed action would weaken the dollar and boost non-US economies.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| Crude oil | commodity | neutral | After all, futures are up today, oil is down slightly, and Trump has promised peace with Iran once again. |
| NVIDIANVDA | equity | neutral | from Wagon, which okay, now I'm using actual names of companies, which is uh related to um NVIDIA. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
The Nasdaq is down 9% from highs.
Unidentified speaker (speaker_0) ·0:03
How to verify: Check Nasdaq Composite index historical data for the period leading up to November 2, 2026. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Tech stocks lost one $1 trillion in a day.
Unidentified speaker (UNKNOWN) ·0:21
How to verify: Check daily market capitalization changes for major tech indices or S&P 500 for the relevant trading day. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Nasdaq dropped 4% in early October.
Unidentified speaker (speaker_0) ·1:05
How to verify: Check Nasdaq Composite index performance for early October 2026. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Nasdaq dropped a further 4% after the report was published on October 22nd.
Unidentified speaker (UNKNOWN) ·1:21
How to verify: Check Nasdaq Composite index performance for the period following October 22, 2026. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
AI CapEx is peaking and companies are revising down estimates for next year.
Unidentified speaker (UNKNOWN) ·3:27
How to verify: Review Q3 2026 earnings reports from major hyperscalers (e.g., Microsoft, Google, Amazon) for CapEx guidance revisions. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Interest rates matter and high yield credit stress is present.
Unidentified speaker (speaker_1) ·4:07
How to verify: Check ICE BofA High Yield Index performance and default rates for Q3/Q4 2026. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
By the end of next year, AI CapEx will be $2.5 trillion.
Unidentified speaker (speaker_1) ·19:28
How to verify: Compare against consensus analyst forecasts for AI CapEx published in late 2026. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
82% of industry revenue comes from CapEx recycling.
Unidentified speaker (speaker_1) ·20:27
How to verify: Review Perkins' blog or report for the specific methodology and data source used to calculate this ratio. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Data center employment will be 800,000 by the end of next year.
Unidentified speaker (speaker_1) ·23:43
How to verify: Check US Bureau of Labor Statistics (BLS) data for 'Data Centers' or related NAICS codes. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
The mid-1970s recession was caused by inflation.
Unidentified speaker (speaker_1) ·27:39
How to verify: Review Federal Reserve economic data and historical analysis of the 2026-2026 recession. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Interest rates have crowded out consumer durables, housing, and construction unrelated to data centers.
Unidentified speaker (speaker_1) ·28:28
How to verify: Check Federal Reserve economic projections or GDP component data for residential construction and consumer durables sales trends during the high-rate period. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Job-to-job transfer wage growth is picking up and serves as a leading indicator.
Unidentified speaker (speaker_2) ·30:09
How to verify: Access the Atlanta Fed Wage Growth Tracker or BLS data on job-to-job wage differentials. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Wage growth for low-income employees is increasing due to AI data center construction demand and de-immigration reducing labor supply.
Unidentified speaker (speaker_2) ·30:29
How to verify: Review BLS Occupational Employment and Wage Statistics for construction sectors and Census Bureau migration data. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
The US stock market is currently at 45% concentration in tech, up from 40% last week.
Unidentified speaker (speaker_1) ·34:07
How to verify: Calculate the market capitalization of the top tech stocks (e.g., Magnificent 7) relative to the total S&P 500 or US equity market using Bloomberg or Refinitiv data. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Historically, a 40% stock market concentration in one sector is a cutoff point where bad things start to happen.
Unidentified speaker (speaker_1) ·34:07
How to verify: Research historical sector concentration data from BIS or academic papers on market bubbles (e.g., dot-com era) to verify the 40% threshold claim. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Korea and Taiwan are part of the tech supply chain and would be destroyed in an AI crash.
Unidentified speaker (UNKNOWN) ·34:49
How to verify: Analyze export data from Korea and Taiwan to the US tech sector and their respective stock market correlations with US tech indices. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Europe's reliance on gas for energy pricing dropped from 90% in 2022 to an estimated 60%.
Unidentified speaker (speaker_2) ·40:05
How to verify: Review European energy market reports from ENTSO-E or IEA on marginal pricing mechanisms and gas share in electricity generation. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Local government loan issuance in China picked up by 60%.
Unidentified speaker (speaker_2) ·38:41
How to verify: Check China's Ministry of Finance data on local government financing vehicles (LGFV) bond issuance or shadow banking statistics. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
The Nasdaq was down roughly 8-9% in the last month.
Unidentified speaker (speaker_0) ·36:10
How to verify: Check Nasdaq Composite index performance data for the month prior to the podcast release (October 2026). For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
European equities held up 3.5% while the Nasdaq fell 8-9% in the last month.
Unidentified speaker (speaker_0) ·36:10
How to verify: Compare STOXX Europe 600 or Euro Stoxx 50 performance against Nasdaq performance for the relevant month. For recent events, verify 2026-08-02 through 2026-11-02. Use another period only when the speaker explicitly names it.
Do your own research
- Verify AI CapEx forecasts from major hyperscaler earnings reports and industry analyst consensus (e.g., Goldman Sachs, JPMorgan tech research).
- Check high-yield credit spreads and default rates using Bloomberg or ICE BofA indices to assess credit stress levels.
- Review Nasdaq Composite performance data for October 2026 to confirm the reported 8% total decline.
- Analyze token pricing trends and compute unit economics for major AI inference models to assess commoditization risks.
- Examine US job-to-job transfer wage growth data from the Bureau of Labor Statistics (JOLTS report) as a leading labor indicator.
- Compare US tech sector market cap concentration against historical dot-com era data to evaluate decoupling risks.
- Assess UK productivity and wage growth divergence using Office for National Statistics (ONS) data.
- Review European natural gas price formation metrics to verify the claimed reduction in gas's role in energy pricing from 90% to 60%.
Full transcriptexpand
- 0:02
- Unidentified speaker (UNKNOWN): As
- 0:03
- Unidentified speaker (speaker_0): the band The Strokes once asked, Is this it? Is the AI crash happening? With the Nasdaq down 9% from highs, we got headlines like Tech Carnage, AI stocks plunge, and AI doomsday fears arrive at worst possible time. And my personal favourite:
- 0:21
- Unidentified speaker (UNKNOWN): tech
- 0:21
- Unidentified speaker (speaker_0): stocks just lost one $1 trillion dollars in a day. A gothic horror test for the moats. Buffett taught Wall Street to love.
- 0:29
- Unidentified speaker (UNKNOWN): Of
- 0:29
- Unidentified speaker (speaker_0): course, this podcast headline will be completely balanced. After all, futures are up today, oil is down slightly, and Trump has promised peace with Iran once again. So is this actually it? Let's answer. Is this it?
- 0:45
- Unidentified speaker (UNKNOWN): What
- 0:45
- Unidentified speaker (speaker_0): is the soft level soft landing level of AI CapEx? Number two, are you sure? yields would fall in an AI crash? What about the fiscal situation? And finally, is global decoupling from an AI crash really plausible? Right. First of all, listeners, if you want someone to blame for this, Dario is your man.
- 1:05
- Unidentified speaker (speaker_0): In early October, he told us about a report he was going to write and on what the AI Capex crash would look like. NASDAQ drops 4%. four percent. He then published said report on the twenty second of October and we lost a further four percent. Dario, can you start by apologizing?
- 1:21
- Unidentified speaker (UNKNOWN): And
- 1:21
- Unidentified speaker (speaker_0): then answering, is it happening?
- 1:24
- Unidentified speaker (UNKNOWN): Well
- 1:24
- Unidentified speaker (speaker_1): I don't think I need to apologize, you know, for having a wildly successful blog. I mean that there has been some chat like on social media and stuff that um you know because I'm suddenly like really big and really famous, like maybe I'll stop doing these podcasts
- 1:39
- Unidentified speaker (speaker_1): or I might get big headed or something. Like, I mean, that's just not gonna happen. You know, I
- 1:44
- Unidentified speaker (UNKNOWN): just want to know like
- 1:45
- Unidentified speaker (speaker_1): whatever like fame I get or like fortune or prizes, you know, the Nobel Prize for Economics. I mean, surely that's like obviously gonna be mine now. I'm still gonna keep doing this, and you know, I I think it's a bit like you know, the rap game, like when you make it big, you like take all your homies with you. You know, I'm not gonna like leave you behind and like
- 2:05
- Unidentified speaker (UNKNOWN): You're gonna ring
- 2:05
- Unidentified speaker (speaker_2): us along.
- 2:07
- Unidentified speaker (speaker_2): Thank you, Dario.
- 2:09
- Unidentified speaker (UNKNOWN): And
- 2:09
- Unidentified speaker (speaker_1): like even like even the ones that were like on holiday when all this was happening, like when when I had to write the blog myself and then like distribute it myself and marketing it myself because everybody was away. Like maybe
- 2:22
- Unidentified speaker (UNKNOWN): the producer was on
- 2:23
- Unidentified speaker (speaker_1): holiday or like Freya was off gallivanting, God knows where. I
- 2:27
- Unidentified speaker (UNKNOWN): was
- 2:27
- Unidentified speaker (speaker_0): I was in Sicily driving around in a very, very brilliant BYD car. That was I have no complaints about. I'm sorry, Dario. I I can't back up or verify what you said. It's a fantastic car. It served me very well. Uh visiting my brother in law and his wife. It was it was wonderful.
- 2:44
- Unidentified speaker (speaker_1): So I'm just you know, all I'm saying is like I'm not gonna change, you know, I'll still be the same. Don't worry.
- 2:51
- Unidentified speaker (speaker_1): It's discontinues.
- 2:53
- Unidentified speaker (speaker_1): Is this it? The critically acclaimed blog said that it wasn't. Like this was just like the taster of what would come. And I think there's definitely like an element of positioning here because like we saw this from clients, like all the clients we talked to during the summer, they're all like massively bullish on AI, but they're all calling it a bubble. And so like any sense of
- 3:13
- Unidentified speaker (speaker_1): something was start
- 3:14
- Unidentified speaker (UNKNOWN): to
- 3:14
- Unidentified speaker (speaker_1): go wrong, you just felt that they'd be like desperate to get out of those positions. Um, so it's not the first time we've seen something like this, like a big correction.
- 3:23
- Unidentified speaker (UNKNOWN): I
- 3:23
- Unidentified speaker (speaker_1): guess this time there's some more like troubling signs. So
- 3:27
- Unidentified speaker (UNKNOWN): the
- 3:27
- Unidentified speaker (speaker_1): obvious one is that we're in an earnings season and we've seen some earnings misses. That hasn't happened before, even in the tech sector. Now, some of it is like accounting losses and the fact that the stuff they invested in has started. to go down. But yeah, when it was going up, it was bullish though. Now it's going down. We can just ignore it. Sure. Um, the other bit is
- 3:47
- Unidentified speaker (speaker_1): like the CapEx is now clearly peaking. This is not good. You you can see like the the I mean, it's only just starting. So these companies are saying, well, they're you know beginning to revise down the estimates for next year very early, but this is like beginning. to sort of peak. Um, the credit stress is obviously there. This is what
- 4:07
- Unidentified speaker (speaker_1): like a big feature of that critically acclaimed, widely popular blog that somebody published.
- 4:13
- Unidentified speaker (UNKNOWN): Um,
- 4:14
- Unidentified speaker (speaker_1): and like there's some parts that are like really vulnerable. And this goes back to the debate we had like all year when people were saying interest rates don't matter, the Fed doesn't matter. Well, actually, yes, they do. Look what's been happening in high yield. And it's not just that the Fed raises interest rates, it's that like for like Walsh actually got something right last week in the press conference.
- 4:34
- Unidentified speaker (UNKNOWN): No, I
- 4:34
- Unidentified speaker (speaker_0): thought you'd you
- 4:35
- Unidentified speaker (UNKNOWN): actually like him
- 4:36
- Unidentified speaker (speaker_0): now, Dario.
- 4:36
- Unidentified speaker (UNKNOWN): I'm starting.
- 4:41
- Unidentified speaker (speaker_1): I'm not sure if this is like Stockholm syndrome, like you start to believe Leave, like you have to be nice to him. Like all the journalists like in the press conference that are suddenly all like being really nice to him and sharing. jokes and stuff, and they've all got these in jokes. Maybe I'm falling victim of that. Like maybe that's part of it. Um, but I do think he's got less irritating. That's like
- 5:00
- Unidentified speaker (speaker_1): that's a good thing. Like compared
- 5:02
- Unidentified speaker (UNKNOWN): to a few months
- 5:03
- Unidentified speaker (speaker_1): ago, his like actually giving a view on the economy, and he's actually sharing that view with us. Like maybe now he's had time to form a view, which he didn't have a few months ago. I don't know. But he's like there's definitely there's less ego. I mean, I like I started this podcast being there's a little bit of ego, but there's less ego coming from him, which is quite nice. Um,
- 5:23
- Unidentified speaker (speaker_1): so I I do like prefer these press conferences now. Fay is just trying not to
- 5:27
- Unidentified speaker (UNKNOWN): well
- 5:27
- Unidentified speaker (speaker_2): that does feel wildly hypocritical.
- 5:30
- Unidentified speaker (UNKNOWN): All right, all right, yeah, but it's a joke.
- 5:31
- Unidentified speaker (speaker_1): When I do it, it's a joke. When he does it, it's serious.
- 5:35
- Unidentified speaker (speaker_2): All right.
- 5:35
- Unidentified speaker (UNKNOWN): So I like
- 5:36
- Unidentified speaker (speaker_1): a little bit less like irritating. I'll try to. be less irritating for the rest of this. And I
- 5:41
- Unidentified speaker (UNKNOWN): won't talk
- 5:41
- Unidentified speaker (speaker_1): about the blog or like the critical claim or the nice comments I got and linked. None of that, right? Um, but we're seeing like the stress, and he was right, like interest rates, monetary policy get into all the crack. So you're beginning to see this like feed across the ecosystem. So that's definitely happened. The other bit A is like the correlations are there now that weren't there before. So before it was like the
- 6:01
- Unidentified speaker (speaker_1): hyperscalers would sell. off and the semis would go up or they'd do the reverse. And it was like money rotating from one to the other. At the moment, they're both going down, which suggests that it's like a genuine sort of or a sort of generalized loss of confidence in this narrative. But I do think it's mostly confidence. I do think like the Capex thing is only just starting. So
- 6:21
- Unidentified speaker (speaker_1): it wouldn't surprise me if we'd sort of recovered from this and then like the big crash that the note, the really widely acclaimed, critically, you know, important note, the one that probably wins a Nobel Prize, but let's not jump ahead of ourselves. That
- 6:35
- Unidentified speaker (UNKNOWN): note
- 6:35
- Unidentified speaker (speaker_1): was predicting like the crash for the beginning of 2028. So
- 6:39
- Unidentified speaker (UNKNOWN): I
- 6:39
- Unidentified speaker (speaker_1): stick to that.
- 6:40
- Unidentified speaker (speaker_0): Yeah. But with all these things, uh you know, you have a prediction which which we've we've had. I mean, I know I'm gonna say it again, but it's been well over a year that we've been predicting this and laying out the framework for it. So this report was and and the blog that Dario is referring to is widely available. You can go find it and have a read. Uh but the the other report that you wrote, Dario, Wargaming, the
- 7:00
- Unidentified speaker (speaker_0): AI crash, was was fantastic as well and way more in depth. And I guess I just had a question about especially i the the link with the railroads um and how that worked out and and your point around that sort of technology, new technology blowing up in a crash but ultimately sticking. with us and being okay. I mean is that how is that how you envisage it? It's sort of a slightly a couple of years
- 7:20
- Unidentified speaker (speaker_0): that this plays out or could this happen all quite quickly over like a short period of time considering how accelerated everything is these days?
- 7:27
- Unidentified speaker (speaker_1): Yeah so that I mean that was from a different note but it was basically the the the argument that if you look through history you have these sort of big sort of world changing technologies. But the companies that are sort of doing the build out aren't always the ones that make the money from it. And you know, with something like this, where you know, token costs are collapsing, but prices are also
- 7:47
- Unidentified speaker (speaker_1): collapsing, and most of the benefits seem to be getting pushed onto consumers. Um, there is that risk of commoditization of this industry, which I think is, you know, the risk that we're talking about here. And, you know, if that plays out, then it just goes the way of all these other technological revolution. revolutions, which is that you get like something big, something that changes the world,
- 8:07
- Unidentified speaker (speaker_1): but you get a wave of defaults and you know the earnings don't quite live up to the expectations. And you you sort of look back and you say, Well, we're all better off because of this, but those companies that started this aren't necessarily. So yeah, I mean it's a different story, isn't it? The difference, I mean, I don't particularly like the technology, but that's a different question. Like I do use it every day. I just
- 8:27
- Unidentified speaker (speaker_1): don't quite trust it. And not because I think it's trying to kill me or stupid
- 8:34
- Unidentified speaker (speaker_1): when I was writing that that wonderful blog that everybody likes, I thought like I thought, because I wrote the note first, which was like 20 pages full of charts, and I thought I can just sort of give this to Claude and he'll just change it into a really funny blog and it'll be just like Citrini. But he didn't like what he produced was just rubbish. Like none of the jokes was funny. Not as funny as me anyway. Obviously.
- 8:54
- Unidentified speaker (speaker_0): That's why we
- 8:54
- Unidentified speaker (UNKNOWN): need you.
- 8:55
- Unidentified speaker (speaker_0): We need you.
- 8:56
- Unidentified speaker (speaker_1): And and so I ended up like having to spend hours and hours just rewriting the whole thing. And that just sort of annoyed me. And then the other bit that annoyed me was that he just like wouldn't use any of the names of any of the companies, particularly Anthro Anthropic. It was like all the way through, it was like, Are you sure you want to say anthropic? Don't you want to say standalone model developer? And I was like, No
- 9:16
- Unidentified speaker (UNKNOWN): And
- 9:16
- Unidentified speaker (speaker_1): he's like, But I should warn you that I'm made by Anthropic. And I'm like, I don't care.
- 9:21
- Unidentified speaker (UNKNOWN): And he
- 9:22
- Unidentified speaker (speaker_1): said, But I've got no inside information. I was like, this isn't about inside information. This is you just not wanting to criticize this company that's doing all this this crazy stuff.
- 9:30
- Unidentified speaker (speaker_0): Different view?
- 9:31
- Unidentified speaker (speaker_2): Uh yeah, I have a different view on the tech, which we know. So just to be clear, our reason for the circumspection around tech valuations and the outlook for tech has nothing to do with Dario's trust issues. Um a r on on the the tech his use of the technology. Uh I used the tech. Okay,
- 9:51
- Unidentified speaker (speaker_2): so here was my weekend, and don't judge me for this having been my weekend. The kids were away and this is what I chose to do with my weekend.
- 9:57
- Unidentified speaker (speaker_1): You build like
- 9:58
- Unidentified speaker (UNKNOWN): a
- 9:58
- Unidentified speaker (speaker_1): fort in the garden or
- 10:00
- Unidentified speaker (speaker_2): no, I built a virtual fort. to house economic research and data
- 10:06
- Unidentified speaker (speaker_2): using Unreal. Ooh, good idea. Yeah. Using Unreal Engine, which I downloaded on a virtual computer
- 10:16
- Unidentified speaker (speaker_2): from Wagon, which okay, now I'm using actual names of companies, which is uh related to um NVIDIA. Um, and I got a bunch of commute compute on there, a bunch of stuff. storage on there. Uh, I had a little agent set up, wandering round London like a complete madwoman, talking to the agent on my phone and
- 10:36
- Unidentified speaker (speaker_2): telling it what to do as it was building this space using Unreal Engine, which is a uh a game development um engine. So I did that on my own time as an experiment because I thought it'd be fun, and it was awesome fun, and it's kind of cool. Um
- 10:53
- Unidentified speaker (UNKNOWN): this is a financial
- 10:53
- Unidentified speaker (speaker_0): markets podcast. We're trying to help clients invest and that's all well where where is this going?
- 10:58
- Unidentified speaker (speaker_2): Where is this going? That's what how we're gonna develop our that's gonna how we're we are going to deploy our research to clients going forwards um through a video game interface. Okay, I'm not actually being serious yet, but maybe um that's
- 11:12
- Unidentified speaker (UNKNOWN): right. Just real realize like
- 11:13
- Unidentified speaker (speaker_1): players Raya's gonna have like this little character of herself, like wandering around doing that.
- 11:18
- Unidentified speaker (UNKNOWN): Totally. And
- 11:19
- Unidentified speaker (speaker_2): we can like have avatars of ourselves in this in this world. I won't have to travel as much, will be
- 11:24
- Unidentified speaker (UNKNOWN): which will be
- 11:25
- Unidentified speaker (speaker_2): fantastic. But that so then let's think about who captures the value for that. Will everyone in our space do the same thing? Probably not, because they're not all as mad as I am, but maybe there will be, you know, everyone is gonna going be to using AI more. in our space. And so competition will, you know, stay fierce in the space. So
- 11:45
- Unidentified speaker (speaker_2): maybe we don't capture as much as that. Obviously, I shouldn't have said that on online here because now that gives us less pricing power. But if we're able to differentiate ourselves and get some kind of a first mover advantage, then maybe we will get some of the value from uh from that technology
- 12:00
- Unidentified speaker (UNKNOWN): and
- 12:01
- Unidentified speaker (speaker_2): from those developments and the deployment. of the technology. And that brings us back to this idea of
- 12:07
- Unidentified speaker (UNKNOWN): the
- 12:07
- Unidentified speaker (speaker_2): mini cycle. Um, and I just want to circle back to something that Andrew was mentioned, which was that we've been call we've been talking about this for well over a year. Just to be clear, we haven't been calling an AI crash for well over a year. We've been talking about these mechanisms and thinking about the point in time where uh it would all sort of come to bear.
- 12:27
- Unidentified speaker (speaker_2): And what seems to be happening over the past kind of two, three years, even, is this sort of asset price mini-cycle, which is a function of the fact that we're starting this macro regime with a lot of capital left over from the previous era, from the 2010s. We uh uh we saw a lot of asset price inflation
- 12:47
- Unidentified speaker (speaker_2): during that period of time, and then the features of this new. regime are essentially that it's a re-leveraging cycle, which at the moment is centered around this new technology, but i in the right cyclical environment we will see that spreading elsewhere to to other parts of the economy and globally. It's a capex cycle relatedly, and again it's not just in
- 13:07
- Unidentified speaker (speaker_2): AI, uh more broadly when you don't have the the the the supply side shocks, we do start to see evidence of of Capex rising as well. a percentage of GDP globally. And then on the supply side, we have restricted uh labor supply. When you put all that all that all that together, um, you get a demand-resilient
- 13:26
- Unidentified speaker (speaker_2): economy uh with a lot of capital. And then the final feature of this new regime is that there are negative supply shocks. And those negative supply shocks thus far, um, or more recently at least, have hit the rest of the economy, as in the not tech non-tech sector, a lot more than they've hit the tech sector. So all of that capital that we we've
- 13:46
- Unidentified speaker (speaker_2): started this new regime with gets forced into the tech space. Um, that happens to be a US environment, happens to be. That's a little bit unfair. That is a US um call uh which pushes everyone straight back into the US hyper-exceptionalism, hyper exceptionalism. uh, everyone drops. their their hedging ratios on the on on the FX uh
- 14:06
- Unidentified speaker (speaker_2): side of things. Um and everyone's crowded into that sort of top right hand side of the dollar smile where uh the only way of of playing a positive tr tech trade is to buy the tech complex. So we've right back to that stage uh in the past
- 14:22
- Unidentified speaker (speaker_2): until the past kind of few weeks or so that's exactly where uh where everyone had been forced back into by the fact that there have been these sort of negative supply shocks um hitting the rest of the of the economy. But it's not a very stable equilibrium for a number of different reasons, some of which Dario has already sort of pointed to. One of which is
- 14:42
- Unidentified speaker (speaker_2): that that that that process is demand positive um and creates the risk of overheating in the US labor market. And so the Fed has to respond to that. And I think one of the really good things that has happened over the course of this half is that Walsh has stepped up, the Fed is still the Fed, and the monetary
- 15:02
- Unidentified speaker (speaker_2): transmission mechanism is still working. There was a risk, as Dario highlighted earlier, there was a risk Um that the the tech space would just be completely oblivious to the fact that uh the the Fed was trying to to uh remove a dose of accommodation um and to tighten monetr monetary policy.
- 15:21
- Unidentified speaker (UNKNOWN): Um
- 15:21
- Unidentified speaker (speaker_2): but it turns out that the those companies that have leveraged up very rapidly, that have taken on a lot of debt with respect to their own balance sheet are exposed. And that that is part of what is is causing this kind of slowdown at this moment in time alongside the other issues that that Dario has mish mentioned as um as triggers.
- 15:40
- Unidentified speaker (speaker_0): So so you're saying that if
- 15:41
- Unidentified speaker (UNKNOWN): the
- 15:42
- Unidentified speaker (speaker_0): Fed being more hawkish and hiking rates, if that just hadn't affected the the AI stock
- 15:48
- Unidentified speaker (UNKNOWN): that would have been a big problem. That would have
- 15:49
- Unidentified speaker (speaker_0): been really worrying.
- 15:50
- Unidentified speaker (speaker_2): Because because the last thing we want is for the the tech space just to completely obliviously continue leveraging up, continue investing for the next kind of two or three years. Bigger,
- 16:00
- Unidentified speaker (speaker_0): bigger, bigger bubble.
- 16:01
- Unidentified speaker (speaker_2): Bigger and bigger, bigger bubble and then something tips it off, and then you really do have a global recession. Um whereas, you know, when we get into the the the rest of the the the questions, um what we're going to find is that that's it's there is um less of a spillover at this moment in time to to the real economy. And I think that the the really kind of positive
- 16:21
- Unidentified speaker (speaker_2): thing that has happened over the course of this half is that there's been one of the th there's been another kind of speed. check, um, and that that prevents, you know, that that sort of slows that process down and gives the economy a little bit of a chance to sort of get on a on a more even footing. And we see that process of market discipline starting to to to work through
- 16:41
- Unidentified speaker (speaker_2): through the system. And effectively what that's doing is just saying are are investors currently um attributing the future earnings gain. gains from this marvelous new technology to the right companies? And there's a bit of a questioning of that right now. Yeah. Um so we're in another process of that mini-cycle. The problem is that the rest of
- 17:01
- Unidentified speaker (speaker_2): the the economy is still facing quite a lot of of difficulties with regards to the the the energy shock. And so there's a bit of a a pressure there. There's not that much uh evidence that I can see at least in the major economies, that there's any kind of stress with regards to the the rest of the real economy and the increase in
- 17:21
- Unidentified speaker (speaker_2): interest rates or even that increase in energy. There's been a lot of sort of uh transition away from the exposure to gas prices
- 17:31
- Unidentified speaker (speaker_2): in Europe, uh and even the sort of the old guard sectors of the US seem to be trucking along relatively nicely despite the fact that. diesel has has risen so much. There doesn't seem to be a massive um problem in terms of delinquency or or kind of delinquency rates. And so the economy is kind of kind of trucking along. And really to to to sort of
- 17:51
- Unidentified speaker (speaker_2): round out um the next steps for the the the the asset price mini cycle, if we stay in risk on, the condition to stay in risk on is that we start to see profits broadening out. from the tech sector into other parts of the economy. That's the only way that the labor market can really continue this process of reacceleration, which is a
- 18:12
- Unidentified speaker (speaker_2): necessary condition for private consumption to continue recovering and for growth to remain in a risk-on environment. But if the condition for us staying in risk-on risk on is to see profits broadening out, then the upside surprises are going to come from outside of. the tech space. If we go into risk off because this AI story, um, or if we sort of
- 18:32
- Unidentified speaker (speaker_2): stay in risk off because you
- 18:33
- Unidentified speaker (UNKNOWN): know
- 18:33
- Unidentified speaker (speaker_2): things are getting a little bit hairy right now, um, because that some of these triggers start to sort of come through, um, then we anyway wouldn't want to be in tech. So I think we've gotten back to that stage where we're looking for other um opportunities Communities outside of the tech space. That doesn't necessarily mean that this is the crash, really because.
- 18:51
- Unidentified speaker (speaker_2): there isn't enough systemic build-up of leverage yet for this to be a really serious um problem.
- 18:57
- Unidentified speaker (speaker_1): The issue is right that it seems like Bayes describing a soft landing. But the the issue is that the CapEx is just so big already.
- 19:08
- Unidentified speaker (speaker_1): I take the point that if we take some of the heat out of this now, um the bubble won't get so big. And so then when it stops, we won't have. such a big collapse. I think that's true. The problem is that we're talking about like CapEx on a massive scale. I mean, this is like by the end of the next year, it's going to be what, two and a half trillion? I mean, these are massive numbers.
- 19:28
- Unidentified speaker (speaker_1): And so to like, and we get this question a lot from clients over the past couple of weeks. It's As been being like, what is this soft landing level of CapEx? Like, because they assume that if the hyperscalers are speaking. spending less, um, their free cash flow starts to return. And that must be a good thing. And that was a reason to be bullish. Um, but you know, the other
- 19:48
- Unidentified speaker (speaker_1): side of that is if you're putting less capex into the system, then revenues across the system are going to start to you know contract or come under pressure. And so you have to find a level of capex that restores free cash flow to some degree, but doesn't undermine the revenues. of the entire industry. And I and the only way that really happens is if you can reduce
- 20:08
- Unidentified speaker (speaker_1): CapEx slowly and at the same time, you've got huge amounts of revenue coming into this ecosystem from businesses and consumers using the technology. So you remember that chart that we've been using for like six months now that shows the share of revenues in the industry that come from just CapEx recycling. Um that's still like
- 20:27
- Unidentified speaker (speaker_1): 82% or something. And it hasn't really changed. Like over the last six to twelve months, it's been fairly stable. So it's still like revenues are going up, but it's mainly because CapEx has been stronger. Um, so if you had a huge increase in that sector that's you know of the money coming in from businesses and consumers, there may Maybe, be, but it that that sort of, you know,
- 20:47
- Unidentified speaker (speaker_1): passing the batter from hyperscaler capex to business and consumer subscriptions is going to be really difficult, particularly if you're in this environment or for this open way, you know, free stuff. So I don't know. I'm I'm a bit more pessimistic on whether there's this sort of or like our star, but it's like a capex star, like AI Capex star,
- 21:08
- Unidentified speaker (speaker_1): like some equilibrium level of Capex that we can just hit this kind of gold locks environment. I'm not sure. Um
- 21:14
- Unidentified speaker (UNKNOWN): it seems
- 21:15
- Unidentified speaker (speaker_1): more likely to me that this will be you know some correction that will shake out sentiment. The sentiment will come back, the capex will continue. Maybe the stocks you start to melt higher again because they only seem to either go up a lot or down a lot. It's not like they just
- 21:30
- Unidentified speaker (speaker_1): and then like you know that very influential blog ends up being correct, which is that this just turns into a nasty crash. And I think where I'm where I'm more optimistic about this is actually on the impact of the crash on the the economy.
- 21:43
- Unidentified speaker (speaker_2): Right.
- 21:44
- Unidentified speaker (speaker_1): Because I don't think this is going to like the people that are bearish about this stuff, um, and you see them on Twitter, there's not that many of them because Twitter's main mainly full of people sort of AI boosterism, you know, this is all great, fantastic stuff, all the stocks are going to the moon. Yeah, the people that are bearish on this are all saying, well, this is going to lead to some massive collapse. It's
- 22:04
- Unidentified speaker (speaker_1): going to be like um sub it's going to be like subprime, you know, 2.0. And you know, you even have people talking about this being the biggest misallocation of capital in history. And there's this one guy I used to work with, Julian Garan. Um, he's he uses this thing called a Wixilian coefficient. I don't know if you know what this is, but
- 22:24
- Unidentified speaker (speaker_1): basically you take a measure of R star, which as we know. is just complete nonsense. Um, you compare the interest rate to that R star, and then you sort of calculate the cumulative difference over time. And so on his measure, you've got like the biggest resource misallocation in history because you've got this massive um difference, sort of um
- 22:44
- Unidentified speaker (speaker_1): sum difference of. our, I mean, it's all nonsense because R our star is nonsense. So, you know, when you start like compounding the gap between the interest rate. rates and our star, you end up with like nonsense squared. I mean, literally. I don't think I don't think it's going to be like that. I think like the stocks would go down a lot, the capex would go down a lot, there'd be quite a big hit in terms of capex. What people forget, and
- 23:04
- Unidentified speaker (speaker_1): that like you you can show them the chart of like uh like Korean and Taiwanese High easy exports, and suddenly they're like gone up 20% of GDP. Um, that's where all the a lot of this stimulus is going. It's like leaking out into the supply chain. Um, if you just do like a simple map of the cumulative impact on
- 23:23
- Unidentified speaker (speaker_1): GDP from imports and tech capex, they basically cancel each other out. Um, now it's not it's not really true in reality because you know you take these these um chips and then you add value to them domestically, like NVIDIA is a US company and it adds. value to them. But if you look at like the value added in GDP that comes from the tech
- 23:43
- Unidentified speaker (speaker_1): sector, it's been trending higher, but it's on the same trend it's been on for ages. So it's not like there's some obvious like massive hit to GDP that's coming. And you like take data center employment, it's what 500,000, maybe at the end of next year it'll be 800,000. thousand. That's like half a percent of unemployment. So like these aren't like massive
- 24:03
- Unidentified speaker (speaker_1): systemic things. Like the all the debt, I mean there's like a lot of debt. Um most of it is unwritten by hyperscalar balance sheets. They're not going to default on that debt. But there are some weak sectors of the the ecosystem that probably would default, including certain standalone model developers that have taken on a load of debt. But again, you know it's not systemic. Like the worst case scenario I
- 24:23
- Unidentified speaker (speaker_1): can come up with is like a 1990 nineteen ninety star style recession because like the debt looks a bit like the leverage buyout boom in the late eighties, like a lot of financial engineering, a lot of the sort of private debt stuff. Um but it's very difficult to tell like a subprime style disaster from this.
- 24:41
- Unidentified speaker (speaker_0): Right. So I think that that's pretty clear. So this is probably not it. This is probably just a healthy pullback, a healthy correction, and then we
- 24:47
- Unidentified speaker (UNKNOWN): we
- 24:47
- Unidentified speaker (speaker_0): get another melt up into the end of
- 24:49
- Unidentified speaker (UNKNOWN): the year.
- 24:50
- Unidentified speaker (speaker_0): Yeah,
- 24:50
- Unidentified speaker (speaker_2): I think we continue to be in this asset mini cycle, which is basically like rotation and re-rotation. Um but the the tech stocks sort of blister higher when they blister higher and they they sell off um when you know they sell off a bit harder when they sell off. Um but it's just sort of like this process of of uh finding where that value's going
- 25:10
- Unidentified speaker (speaker_2): to be captured. And when I say sort of like discipline and the market discipline that's been exacted on uh on the situation over the past kind of few weeks or so as a result in part of of uh interest rates rising and yields rising. It's more so about the potential, I think,
- 25:29
- Unidentified speaker (speaker_2): longevity of the expansion rather than us really being able to say that this pace of increase in debt and capex is not going to end in um a more significant sell off. So I think ultimately what happens is you continue to have these mini cycles. Um and then finally the there's
- 25:49
- Unidentified speaker (speaker_2): a a more significant a more significant kind of reckoning that happens. Um and you just hope that that happens before the the um the scale of the build up has been big enough to really cause a a m serious recession. I think the worst case scenario for the US at the moment, like say this sell-off really intensifies.
- 26:09
- Unidentified speaker (speaker_2): Um the worst case scenario for the US at the moment is just we go back to that question that we've had multiple times over the past few years as to whether the US is going to go into recession. And if that happens at this moment in time, we'll probably be saying, you know, probably probably not this is not big enough to uh to to really push the US into a serious recession. As
- 26:29
- Unidentified speaker (speaker_2): Dario says, it's more Korea and Taiwan that have the issues there, although we really do need some of those Korean ships to get energy flowing again. So maybe Korea has a has another sort of source of demand for the economy. But those that those charts on on Korea career and Taiwan, in it's hard to sort of look at those and not
- 26:49
- Unidentified speaker (speaker_2): think there's a Dutch disease. problem building there, leaving them very, very exposed
- 26:55
- Unidentified speaker (UNKNOWN): in a
- 26:55
- Unidentified speaker (speaker_2): macro sense.
- 26:57
- Unidentified speaker (UNKNOWN): Very
- 26:57
- Unidentified speaker (speaker_0): good. Right. We're slightly behind on time. So let's crack on to yields. Are you sure yields would fall in an AI crash? What about the fiscal situation? Right, Daria, let's uh kick off with you. We've talked a lot last pot last two podcasts were mainly about bond yields. So so what where where are we tie how do we tie this in AI crash?
- 27:15
- Unidentified speaker (speaker_1): So I think I'm pretty sure that yields would fall in this scenario. I mean, what we're talking about is a classic demand shock, like an asset price problem bursting, um, capex would drop quite sharply, there'd be some job losses associated with this, um, you know, GDP probably contract.
- 27:31
- Unidentified speaker (UNKNOWN): But
- 27:32
- Unidentified speaker (speaker_1): you know, people tend to look at the past and say, well, there's been recessions in the past where yields have gone up. There's only really one, and that's the mid-1970s.
- 27:39
- Unidentified speaker (speaker_1): And that's because the recession was caused by inflation. So if you get a recession that's caused by inflation, then yeah, yields probably go higher. This is going to be quite
- 27:47
- Unidentified speaker (UNKNOWN): different.
- 27:48
- Unidentified speaker (speaker_2): This is a demand shock.
- 27:49
- Unidentified speaker (UNKNOWN): Yeah,
- 27:49
- Unidentified speaker (speaker_1): this is a demand shock, not a supply shock. Um so and I don't think the term premium is is going to sort of blow out and offset that um for the similar reason. Um, but also like I don't think this is gonna going to require some massive fiscal response. either. Like, we we've got plenty of room to cut interest rates. Like, this isn't like COVID where we had literally no
- 28:09
- Unidentified speaker (speaker_1): monetary ammo. Um, the Fed has raised interest rates quite a lot, probably will have raised interest rates a bit more by the time this ends. Um, you'd have a lot, you compare it to like past recessions, you've got plenty of sort of real interest rate room to cut rates. And I think that would work. Like, that's another reason for thinking that this. would be quite
- 28:28
- Unidentified speaker (speaker_1): mild, is that there's lots of bits of the economy that have been crowded out by high interest rates, or in the sense that you know, interest rates, they're sensitive to interest rates. So consumer durables, um, housing, construction that isn't related to data centers, all that stuff. Like people talk about the K-shaped economy, and they're very focused
- 28:48
- Unidentified speaker (speaker_1): on the top of the K and oh my god, what were the top. would of the K do if equity wealth started going down? They've forgotten about everyone else, the rest of the K. And actually that's the bit you could crowd in with lower interest rates. So I think the Fed could cut rates quite a lot. That the economy would respond to that. You get a a recovery quite quickly. But you know, before that happens, I think yields
- 29:08
- Unidentified speaker (speaker_1): would drop quite a long way.
- 29:09
- Unidentified speaker (speaker_0): Yeah. Very interesting, Fred. Freya.
- 29:11
- Unidentified speaker (speaker_2): Yeah, I mean I think we're very aligned on this. Just on on the bottom part of the K, not so much in response to um an AI crash or you know yields coming down more broadly. I think even without that, we could be starting to see signs that um those reliant on
- 29:31
- Unidentified speaker (speaker_2): labour income in the middle and and bottom sections of the distribution could start to have a bit of an easier ride. And that's our sort of where we're getting the forecast from
- 29:42
- Unidentified speaker (speaker_2): the near term, absenting an AI crash,
- 29:47
- Unidentified speaker (speaker_2): which is essentially that wage growth does look like it's starting to bottom out if we strip out education and health, which don't really contain a lot of signal for the wider economy. Wage growth looks like it's bottoming out in average hourly earnings. The job-to-job transfer wage growth is picking up, and that tends to be a leading
- 30:07
- Unidentified speaker (speaker_2): indicator.
- 30:09
- Unidentified speaker (speaker_2): Then when we look at low-income employees, which aren't really a kind of a cyclical driver, but it's interesting in the sense that that might be one of the only places where you see direct spillovers from AI into the labor market in the US, in the sense that that demand for data structure. data center construction
- 30:29
- Unidentified speaker (speaker_2): is a positive spillover into construction. And at the same time, we've had a lot of de-immigration, which reduces labor supply in that sector. So wage growth for the bottom section of the distribution is already starting to pick up if you look at the Atlanta Fed breakdowns. So we've got wage growth increasing, Um, and then we've also got a
- 30:49
- Unidentified speaker (speaker_2): very high bar for productivity growth because. of things that we've discussed at infinitum on this podcast with regards to last year, and potentially um unit labor cost growth starting to increase in the US, which is not what you really want. Um, you know, there has been a big decline in labor share of
- 31:08
- Unidentified speaker (speaker_2): income again uh in this most recent expansion, but this could be the moment where we start to see that turning around for round the US. So again, when we're talking about rotation, profits broadening out, I said, is a condition for for the the macro economy story remaining on track. Um but that doesn't necessarily mean in fact it I think
- 31:28
- Unidentified speaker (speaker_2): it's not consistent with at this moment in time profit growth accelerating, profit growth improving in the aggregate, um, because maybe unit labor cost growth is going to start to pick up finally. And that's not the case. So that means rotation within the US, but that that um cyclical story is is very different in other places, with the
- 31:48
- Unidentified speaker (speaker_2): UK, funnily enough, being in a completely different place for the first time in a really long time, and wage growth has has dropped, has continued to drop, and productivity growth is actually surprising to the upside. Um, so finally we have the right moment in the in the profit cycle for
- 32:08
- Unidentified speaker (speaker_2): companies to start investing. Um, obviously, not ideal that energy sh energy prices have risen at that moment in time. That's a structural problem that the UK has. But cyclically speaking, um, this is probably a good place for the for the UK for the first time in a long time, as long as it can demand can remain resilient to the increase. in energy prices.
- 32:28
- Unidentified speaker (speaker_1): Is your your wage growth forecast for UK is that adjusted for wildly successful blogs that feed through into wages?
- 32:37
- Unidentified speaker (speaker_0): Just to say I'm looking forward to carrying your bags at the Nobel Prize ceremony next year. Okay. And Freya's looking forward to coming along too and maybe
- 32:46
- Unidentified speaker (UNKNOWN): I'll take your
- 32:46
- Unidentified speaker (speaker_2): coat and hat when you get there.
- 32:49
- Unidentified speaker (speaker_0): Um and on the fiscal side, I mean we you've you've you've both kind of addressed it already. I mean that the you wouldn't have any other uh any major concerns around that on the fiscal side around France or or maybe the UK or or other countries.
- 32:59
- Unidentified speaker (speaker_2): Yeah, I think the fiscal side, no matter what happens with the midterms in the US, it's hard to see a big fiscal expansion from here. The reason why even under uh you know, a republican uh Congress they were able to to to push through one big beautiful bill was because um there was the promise of of tariff revenues. So you can't really
- 33:19
- Unidentified speaker (speaker_2): bring the fiscal hawks on broad on board um anymore because that's like literally that's reversed, right? That's reversing right now. So um it's hard to see big kind of fiscal expansion from here for for the US. And then, you know, we talked about I think we talked about France last time. I don't really have anything different to say. Um this time
- 33:39
- Unidentified speaker (speaker_2): around.
- 33:40
- Unidentified speaker (speaker_0): Fantastic. Right. Kraken on to question three. Is global decoupling from an AI crash really plausible, Dario?
- 33:48
- Unidentified speaker (speaker_1): I I think that's probably the most difficult bit of the analysis that I did on this. Um because I in theory it should be. Like I mean, this isn't anything like dot com. So like in dot com, um a lot of the well sort of fully participated in the dot com mania. Um, you look at sort of BIS
- 34:07
- Unidentified speaker (speaker_1): corporate debt data in the late 90s, it was rising rapidly everywhere because everybody was doing this sort of telcos build out. Um, you look at the stock market at the peak of dot com for all of these countries, it was all like forty percent. I mean, forty percent seems to be like the historical cutoff point where bad things start to happen when one sector takes up forty 40% percent of the stock market. Um
- 34:27
- Unidentified speaker (speaker_1): US currently at 45%, what it was till last week. I haven't updated it yet. So uh like in theory, you should get decoupling. Like the the fact that you haven't got like extremely bubbly stock markets everywhere else, they should be relatively immune. Obvious exception is all of those countries that are part of this you this um tech supply chain, because they they'd
- 34:47
- Unidentified speaker (speaker_1): get absolutely destroyed in that.
- 34:49
- Unidentified speaker (UNKNOWN): Korea,
- 34:49
- Unidentified speaker (speaker_0): Taiwan, et etc. cetera.
- 34:50
- Unidentified speaker (UNKNOWN): Yeah,
- 34:50
- Unidentified speaker (speaker_1): exactly. In theory, yes. I I think the first instinct of investors is not going to be let's take all our money out of the US and put it into Germany. Um I think probably the first instinct is going to be, oh my god, let's sell everything and buy bonds. But I think that um, you know, to the same degree that Europe doesn't have this sort of
- 35:10
- Unidentified speaker (speaker_1): direct exposure to this, even in the stock market or in the real economy, if the Fed is cutting interest rates, you know, the other bit that's sort of to the extent there's been that crowding out that we talked about in the US and different sectors, you could say you've had that globally as well. Like lots of Europe, um, interest rates have just been too tight for too long. Um, even though interest rates, even though yields have
- 35:30
- Unidentified speaker (speaker_1): been rising for sort of good reasons, because it was sort of global resilience and US strength, um, it's still been squeezing those economies. So you've got a lot of room to sort of cut interest rates in line with what the Fed would be doing. I think you'd get um a much weaker dollar as well from these sorts of dynamics. Um a lot of potential for rate
- 35:49
- Unidentified speaker (speaker_1): cuts. Uh so pretty quickly I think you would start to see some decoupling between the rest of the world and the US. It's just that that's not going to be the first instinct. And we we had a taste of this like in the last couple of corrections of US tech. like global equities haven't done brilliantly. So I so yeah, I think there is I think the scope for decoupling is just not the
- 36:09
- Unidentified speaker (speaker_1): first instinct.
- 36:10
- Unidentified speaker (speaker_0): Yeah, you say that, but in the last month, I mean I was talking about the Nasdaq's being is down roughly what, eight, nine percent. But you know, Europe's held up three and a half percent, even even the UK, three percent, places like India almost three percent. So you know the uh up, sorry, up three percent. So They I I I mean they have held up in the last month whilst whilst AI has crashed. They they've you know
- 36:30
- Unidentified speaker (speaker_0): maybe
- 36:31
- Unidentified speaker (UNKNOWN): I
- 36:31
- Unidentified speaker (speaker_1): don't think Europe's been brilliant over the last week or so. Um UK has been a little bit better than some of the bigger European countries, um bigger sort of core European countries. I j I just think like it's a bit a tough sell to like sell like a broad range of global equities when the US is potentially you know facing some sort of recession,
- 36:51
- Unidentified speaker (speaker_1): even a mild one. Well, you know, I think if you bought them on a sort of depends on your horizon, doesn't it? Like if your horizon is two weeks, then you're not going to want anything to do with those markets. If you've got a slightly long horizon, there there could be loads of value in those bits.
- 37:05
- Unidentified speaker (speaker_0): Yeah. And then um, Freya, just broadening it out then to EM and to China. I mean, obviously it's been a rubbish, awful six months for China. Well, What what are we looking at now in in connection? to maybe the AI story in the US?
- 37:17
- Unidentified speaker (UNKNOWN): Yeah,
- 37:17
- Unidentified speaker (speaker_2): just quickly on Europe, we do see uh
- 37:21
- Unidentified speaker (speaker_2): improvements in the credit cycle. As soon as the conditions are right, we see credit picking up again. So like January, beginning of this year before the energy shock, um, Europe was basically proving that it is getting on board with this re-leveraging story. Um so yeah, again. like Daria says, it's not gonna be
- 37:41
- Unidentified speaker (speaker_2): an immediate um rotation that's gonna that's gonna save you in that scenario, in that kind of wargaming scenario. But um in terms of the ability of these other economies to sort of pick up quite quickly, um, that that does seem to be a positive there. Then with with China, um look, they're
- 38:01
- Unidentified speaker (speaker_2): just having a horrible kind of a transition year. Um, in the sense that last year they did a lot of uh fiscal for consumers. Those that rolled off. A lot of kind of consumer vouchers rolled off. And it does seem as though maybe at the beginning of the year there was a bit more of a a positive story around And
- 38:21
- Unidentified speaker (speaker_2): household incomes, wage growth starting to sort of bottom out and and pick up again as um uh the economy came out of sort of PPI deflation, uh, that gave a little bit more space for for wage growth to start picking up again. And I think that's a you know really positive secular development, but it's nowhere near
- 38:41
- Unidentified speaker (speaker_2): big enough to uh counterbalance the base effects in fiscal policy. They have now recognized that local government um loan issues. issuance picked up by a crazy 60%. Anyone else, that would be insane. But that's kind of like the cycle that they have in China. So that will
- 39:00
- Unidentified speaker (speaker_2): start to show through in the economy. It's it's kind of a odd juxtaposition between if you take sort of how Chinese fiscal policy operates versus how German fiscal policy operates. When they turn it on, it works. it gets you some quarters of growth in China, but it's not going to catalyze anything. It's the complete
- 39:20
- Unidentified speaker (speaker_2): opposite in Germany. When they turn it on, it takes, you know, a year longer for it to actually sort of hit the economy. But it does we do think it will will start to sort of catalyze subject to when there are uh energy shocks. Um, And that's unfortunately something that you that that Europe is just more exposed. to. But as I said earlier in
- 39:40
- Unidentified speaker (speaker_2): the pot podcast, um and Davide's been doing a bunch of work on as well. Uh it does seem like the economy in Europe, the Euro area and the UK actually have seen a uh transition to some degree away from exposure to gas. Um I think sort of ninety 90 in
- 40:00
- Unidentified speaker (speaker_2): 2022 in the UK, 90% of the time. the estimate is that
- 40:05
- Unidentified speaker (speaker_2): prices, energy prices were set off of gas, which is the marginal supplier. The estimate is, and it's hard to kind of know exactly what the what the true figure is, but it seems like it's now only 60%.
- 40:18
- Unidentified speaker (speaker_2): So there has been that transition there, and a lot of energy-intensive and industry has frankly just given up. So um the the base that we're starting from is less exposed to energy this time around.
- 40:31
- Unidentified speaker (speaker_0): Great. Thanks, Freya. You can read Dario's Purchini report for free. Search it out actually. I'll put it I'll put a link into it uh in the in the show notes. So so do do do have a read. And if it's sparkling,
- 40:41
- Unidentified speaker (UNKNOWN): hang on,
- 40:42
- Unidentified speaker (speaker_1): links in the show notes. Is it have we done that before? Do you know how to do this? I mean this is like a new skill.
- 40:47
- Unidentified speaker (speaker_0): Don't undersell me. Come on. Uh yeah. So I
- 40:51
- Unidentified speaker (UNKNOWN): I'm sure
- 40:52
- Unidentified speaker (speaker_2): everyone in the world has read it already, so it's probably useless to
- 40:55
- Unidentified speaker (UNKNOWN): to do that at this stage. Can I
- 40:57
- Unidentified speaker (speaker_1): can I also say that like I've had some questions about this? We're not going to change the name of the podcast. Like I think it should still be Perkins versus Beamish. I mean, some people are saying like it's no contest anymore because of that blog or it should just be called Perkins and Friends.
- 41:10
- Unidentified speaker (UNKNOWN): like joining. Perkins and Friends was what
- 41:13
- Unidentified speaker (speaker_1): the suggestion was. But no, I I think think just keep it for now. Like
- 41:17
- Unidentified speaker (speaker_2): arrogance
- 41:18
- Unidentified speaker (UNKNOWN): is very
- 41:18
- Unidentified speaker (speaker_0): unattractive, Dario.
- 41:21
- Unidentified speaker (speaker_1): Like
- 41:22
- Unidentified speaker (UNKNOWN): this is not
- 41:23
- Unidentified speaker (speaker_0): doesn't sound like a joke to me. Sounds like you've got some issues you'd work on.
- 41:26
- Unidentified speaker (speaker_2): Yeah, you need to work that through, Dario.
- 41:29
- Unidentified speaker (speaker_1): Use maybe
- 41:30
- Unidentified speaker (UNKNOWN): if I
- 41:30
- Unidentified speaker (speaker_1): use Claude more and Gemini less, because like Gemini Gemini's is constantly, Oh, that's fantastic, you're brilliant. How did you come up with that? It's genius. Whereas Claude is like, oh I wouldn't have done it like that if I were you. Nah, this doesn't seem right.
- 41:42
- Unidentified speaker (speaker_2): No, I think you need the the the positive feedback. I'm
- 41:46
- Unidentified speaker (UNKNOWN): gonna let
- 41:47
- Unidentified speaker (speaker_2): right from now on. Let's start the podcast with a list of all the fantastic things that Direct me and Andrew will start that with a list of all
- 41:54
- Unidentified speaker (UNKNOWN): the
- 41:54
- Unidentified speaker (speaker_2): fantastic things.
- 41:55
- Unidentified speaker (UNKNOWN): We've only had
- 41:56
- Unidentified speaker (speaker_1): like 45 minutes. I mean,
- 41:58
- Unidentified speaker (speaker_2): yeah. Maybe we should make a whole separate podcast for
- 42:02
- Unidentified speaker (speaker_0): a Q. Purkinian
- 42:03
- Unidentified speaker (speaker_2): Friends, where it's just
- 42:07
- Unidentified speaker (speaker_1): special distribution that's just like my mom. My wife.
- 42:14
- Unidentified speaker (speaker_2): And that very rude person that suggested that we change the podcast name to Purkinian Friends. That does hurt actually. Retract that statement, whoever said that.
- 42:23
- Unidentified speaker (speaker_0): Anyway, you can read that report. I don't know why you'd want to, but you can. You can read that. report. Or you can get in touch, get a trial, read some of other other work that we're producing. Far better work from Freya, from Davide, from Rory. You know, really good stuff rather than the rubbish Dario's churning out.
- 42:40
- Unidentified speaker (speaker_0): But yeah, do get in touch with me, ask for a trial. Happy to extend one. Um, but that's all we
- 42:45
- Unidentified speaker (UNKNOWN): got
- 42:45
- Unidentified speaker (speaker_0): time for. Freya, thank you very much. Dario, goodbye. bye.