Warren Pies argues AI fundamentals remain strong despite technical sell-off, citing GPU scarcity and hyperscaler economics.
Summary
Warren Pies, founder of 314 Research, explains that while technical indicators have triggered a downgrade in equities, the underlying fundamentals of the AI sector remain robust. He highlights tight GPU availability, rising rental prices, and improving revenue per dollar of compute spend as evidence of strong demand. Pies contrasts this with macro risks, noting that the Federal Reserve faces a unique challenge as AI infrastructure costs now drive inflation rather than oil prices. He suggests the current market sell-off is a technical liquidation rather than a fundamental supply shock, though he warns of potential credit stress if interest rates rise during the AI capital expenditure cycle.
Episode sections
- 0:00 – 3:22
Introduction and Allocation Philosophy
Host introduces Warren Pies and his methodology of balancing fundamental conviction with technical risk management.
- 3:22 – 7:23
Technical Triggers for Downgrading Equities
Pies details the technical indicators leading to the equity downgrade, including market consolidation and leadership shifts.
- 7:23 – 10:27
Market Sentiment and Positioning Analysis
Discussion on why the market has not reached capitulation, citing elevated vault target positioning and lack of extreme pessimism.
- 10:27 – 15:02
Semiconductor Dominance and Market Structure
Analysis of the outsized influence of semiconductors on the S&P 500 and systematic trading correlations.
- 15:02 – 17:36
AI Fundamentals: GPU Availability and Rental Prices
Pies argues AI fundamentals are strong due to tight GPU availability and rising rental prices for advanced chips.
- 17:36 – 18:39
Interpreting Contango in GPU Rental Markets
Explanation of GPU rental market contango as a positive signal of capacity demand rather than weak outlook.
- 18:39 – 22:13
Hyperscaler Economics and Anthropic S1 Report
Evidence of strong AI economics, including rising revenue per compute dollar and dropping inference costs.
- 22:13 – 29:32
Supply Side Monitoring and Political Risks
Use of satellite data to track data center construction and assessment of political risks in Texas.
- 29:32 – 31:25
Macro Shift: From Oil to AI Inflation
Discussion on how inflationary pressures have shifted from oil to AI infrastructure, complicating Fed policy.
- 31:25 – 34:03
Fed Policy, CapEx Cycles, and AI Funding Risks
Analysis of the unique risk of Fed rate cuts coinciding with ramping AI capital expenditures and external funding needs.
- 34:03 – 35:12
VC Market Impact on AI ARR
Estimation that VC startups account for 25% of AI Lab ARR and the risk of a VC funding collapse.
- 35:12 – 37:53
Geopolitical Game Theory and Fed Constraints
Argument that geopolitical imperatives constrain the Fed from crashing the AI trade.
- 37:53 – 39:45
Economic Sensitivity and Recession Outlook
Outlook that a cooling AI trade leads to market rotation rather than recession due to persistent fiscal deficits.
- 39:45 – 52:59
Iran War Scenarios and Oil Market Dynamics
Analysis of US strategies in the Iran conflict and their impact on oil markets and market direction.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| Crude oil | commodity | neutral | Um, and during that time, you know, the the oil market and global supply chains are a dynamic system and are gonna gonna adjust. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
The market entered a consolidation phase defined by a 6-7% trading range over a 63-day period with no new highs or lows, occurring about 13% of the time.
Unidentified speaker ·3:43
How to verify: Check S&P 500 price action data for the period leading up to the podcast release (Nov 2026) to identify consolidation patterns and frequency.
Consumer staples rose by 14% in the month of October.
Unidentified speaker ·4:52
How to verify: Review performance data for the Consumer Staples Select Sector SPDR Fund (XLP) or equivalent index for October 2026.
Hyperscaler credit spreads broke back above the levels seen during the 'Leopold liquidation lows' in July.
Unidentified speaker ·5:32
How to verify: Analyze credit default swap (CDS) spreads or bond yields for major hyperscalers (e.g., MSFT, AMZN, GOOGL) comparing current levels to July 2026 levels.
The spread between S&P 500 performance on days when semiconductors are up versus days when they are down is over 105% year-to-date.
Unidentified speaker ·11:07
How to verify: Calculate the cumulative return of the S&P 500 on days where the SMH ETF (or equivalent semiconductor index) had positive returns, minus the cumulative return on days where SMH had negative returns.
GPU availability for Blackwell and B200 chips has been zero.
Unidentified speaker ·16:18
How to verify: Check cloud provider (AWS, Azure, GCP) availability dashboards or third-party GPU rental market reports for Q3/Q4 2026.
Historical data center economic models assumed a 25% annual decline in GPU rental prices.
Unidentified speaker ·17:04
How to verify: Review financial models or analyst reports from 2023-2024 that projected data center ROI and GPU rental trends.
Anthropic's revenue per dollar of compute spend increased from $0.57 in Q2 of the previous year to $2.08.
Unidentified speaker ·19:27
How to verify: Analyze Anthropic's S1 filing (SEC document) for specific metrics on revenue and compute spend.
The Fed had priced in 75 basis points of cuts before the war, but later priced in as many as six hikes out to 2027.
Unidentified speaker ·21:39
How to verify: Review CME FedWatch Tool data or interest rate futures contracts for the relevant period to compare pre-war and post-war pricing expectations.
Satellite data models suggest an additional 27+ gigawatts of data center capacity coming online over 2027.
Unidentified speaker ·23:12
How to verify: This is an internal proprietary claim by 314 Research; verification would require accessing their specific research reports or contacting the firm directly.
GPU rental prices are now in contango, with forward-dated rentals trading above spot prices.
Unidentified speaker ·17:51
How to verify: Check term structure data from GPU rental market providers (e.g., Silicon Data, CoreWeave, etc.) for forward contracts vs. spot rates.
The correlation between the NASDAQ and Treasury tenures is now over 90%, while the correlation between tenures and oil has decreased.
Unidentified speaker ·30:24
How to verify: Calculate the rolling correlation coefficients between the NASDAQ Composite index and US Treasury yields (e.g., 10-year) using financial data providers like Bloomberg or FRED.
This is the first time in history the Fed has cut rates while Capital Expenditure (CapEx) was ramping.
Unidentified speaker ·31:37
How to verify: Compare historical Fed rate cut cycles with corporate CapEx data from the Federal Reserve's Financial Accounts of the United States (Z.1) or BEA data.
VC startup companies account for approximately 25% of AI Lab Annual Recurring Revenue (ARR).
Unidentified speaker ·34:15
How to verify: Review 314 Research's published reports or estimates on AI Lab ARR composition and VC funding flows to AI startups.
Singapore gas oil crack spreads were above $70 per barrel at the time of recording.
Unidentified speaker ·50:50
How to verify: Check historical crack spread data for Singapore Gas Oil (3-2-1 crack or similar) from sources like Argus Media, Platts, or Bloomberg for the date of the podcast (Nov 4, 2026).
Over 60% of the global VLCC (Very Large Crude Carrier) fleet was laden (loaded with cargo) at the time of recording.
Unidentified speaker ·43:53
How to verify: Check maritime shipping data from sources like Kpler, Vortexa, or Bloomberg regarding VLCC utilization rates and laden vs. ballast ton-miles.
The US has not executed indiscriminate bombing (including energy infrastructure) or a ground invasion in the Iran conflict.
Unidentified speaker ·41:28
How to verify: Review official US Department of Defense statements and news reports regarding military operations in Iran during the period discussed.
The Fed Chair (referred to as 'freshly approved Fed share' likely meaning Chair) is a genuine believer in the AI story and tapped from the Trump administration.
Unidentified speaker ·35:30
How to verify: Verify the identity and political affiliation of the Federal Reserve Chair appointed by the Trump administration in 2026 and review their public speeches on AI.
Lee Raymond, former CEO of Exxon, stated that refining is a bad business to 'get married to'.
Unidentified speaker ·51:16
How to verify: Search for transcripts or quotes from Lee Raymond during his tenure as Exxon CEO regarding the profitability and cyclicality of the refining segment.
Do your own research
- Verify current hyperscaler credit spreads against historical data from July 2024 to confirm if they have widened beyond the 'Leopold liquidation' levels cited.
- Check GPU rental pricing trends and term structures (spot vs. forward) on major cloud provider platforms or industry reports to assess contango conditions.
- Review recent semiconductor earnings reports and guidance to see if construction pacing aligns with revenue estimates as claimed.
- Analyze Federal Reserve meeting minutes and recent speeches for explicit mentions of technology sector inflation or AI-related capital expenditure risks.
- Monitor Venture Capital funding data (e.g., from PitchBook or CB Insights) to track the health of the AI startup funding cycle and its impact on ARR.
- Examine the financials of major hyperscalers to calculate the current revenue per dollar of compute spend and compare it to year-over-year figures.
- Review geopolitical developments regarding Iran and US energy policy to assess potential impacts on oil markets and inflation expectations.
Full transcriptexpand
- 0:00
- Unidentified speaker: Nothing said on forward guidance is a recommendation to buy or sell any investments or products.
- 0:12
- Unidentified speaker: All right, everybody, welcome back to another episode of Forward Guidance and excited to be joined today by repeat guest Warren Pius, founder of 314 Research.
- 0:22
- Unidentified speaker: Warren, always great to have you.
- 0:23
- Unidentified speaker: What's going on, man?
- 0:23
- Unidentified speaker: Yeah, man.
- 0:24
- Unidentified speaker: Likewise, always enjoy our conversations.
- 0:27
- Unidentified speaker: Uh really nothing's going on in the world, right?
- 0:29
- Unidentified speaker: No, nothing to talk about.
- 0:30
- Unidentified speaker: Just another day, like nothing to talk about.
- 0:32
- Unidentified speaker: Just shoot the shit, exactly.
- 0:34
- Unidentified speaker: Yeah.
- 0:35
- Unidentified speaker: Yeah.
- 0:36
- Unidentified speaker: Yeah, I wanted to get you on the show because you you put out a couple recent reports where you've you've meaningfully changed your allocation weightings, uh, which is really interesting to me.
- 0:46
- Unidentified speaker: Um, you know, it feels like over the last few years you've you've very correctly faded many of the points of potential market noise hysteria, whatever you want to call it.
- 0:55
- Unidentified speaker: Um but I know you recently changed some of your core allocations.
- 0:58
- Unidentified speaker: So you have a really interesting way to think about it, which is just this this fundamental outlook and then this technical outlook and how those both interplay with each other to understand like where expectations lie.
- 1:08
- Unidentified speaker: And yeah, there's a lot of drivers for how to think about those two outlooks.
- 1:12
- Unidentified speaker: There's you know these these big buckets, which we'll we'll deep dive into each of them.
- 1:15
- Unidentified speaker: But you know, you have you have the Iran war going on and the energy crisis that's potentially ensuing from that.
- 1:19
- Unidentified speaker: You have the AI build-out situation and how that reflects as being like one of the core drivers of the economy right now.
- 1:25
- Unidentified speaker: You have how the Fed is reacting to that core driver and how to navigate that.
- 1:30
- Unidentified speaker: So lots to unpack.
- 1:31
- Unidentified speaker: Uh, why don't you just tell us a bit of context on these recent changes in your allocations and how you're thinking about that?
- 1:37
- Unidentified speaker: Yeah, so I think that's a good starting point.
- 1:39
- Unidentified speaker: Um, one of the things, if anyone's been listening to me over these months, they or even over my career, really going back to Ned Davis days, uh, I kind of have this uh saying that I build conviction on fundamentals but manage risk with technicals.
- 1:56
- Unidentified speaker: Uh and I it's just something I keep in my mind whenever we're going to these market events, and you can have your convictions and you can have your fundamental outlook.
- 2:04
- Unidentified speaker: But if the market starts going against you, I think you need to um have the humility to to shift gears and understand like you can silo those two things, in my opinion, you know, and and so uh, like you said, I I'd say since later 2023, early after Waller signaled that the Fed was going to start easing, we've been on like a bullish overweight equity uh tilt.
- 2:29
- Unidentified speaker: And we also saw some programmatic disinflation and things like that earlier in that this period.
- 2:35
- Unidentified speaker: And I think a lot of those, some of those tailwinds are coming to an end, but we've also been very bullish on the AI story.
- 2:42
- Unidentified speaker: And I and we remain bullish there.
- 2:44
- Unidentified speaker: And and I think that there are it's funny to me to hear the bears so confident, they're so confident.
- 2:50
- Unidentified speaker: They talk about the bubble as if it's burst already and it's like a known thing.
- 2:53
- Unidentified speaker: And I I'm just I don't know there, I'm pushing against that still.
- 2:57
- Unidentified speaker: So fundamentally, I I still believe that we have farther to go in this AI story.
- 3:02
- Unidentified speaker: But like I said, the technicals, I could be wrong.
- 3:05
- Unidentified speaker: I mean, this is a whole new uh era that we're entering into with the new technology, and the path could be very dangerous, even if I'm right about the destination.
- 3:13
- Unidentified speaker: And so um what you have to do, I think, is step back and have the humility to trust the technicals.
- 3:19
- Unidentified speaker: And so whenever you make a move, that's usually going to be where you start.
- 3:22
- Unidentified speaker: So, what did we see?
- 3:23
- Unidentified speaker: Why did we downgrade stocks at the beginning of last week?
- 3:27
- Unidentified speaker: Um, number one, the market entered what we call a consolidation, and this is like a defined period of very um tight trading range and no new highs, no new lows.
- 3:39
- Unidentified speaker: It it happens at all close to all-time highs or 52-week highs.
- 3:43
- Unidentified speaker: Um, and it occurs about 13% of the time.
- 3:46
- Unidentified speaker: We look for like a six or seven percent trading range over like a 63-day period, and we have a bunch of other um requirements.
- 3:53
- Unidentified speaker: But the bottom line is when you get into one of those consolidations, you've lost momentum in the market.
- 3:58
- Unidentified speaker: When you've lost momentum, you have to look for a trend change, and it could either be a breakout or a breakdown.
- 4:02
- Unidentified speaker: But just risk management and discipline tells you you need to um start your antenna should go up that you could you could see a major trend change.
- 4:12
- Unidentified speaker: And so we saw this consolidation, and then within the consolidation, I'd say the big uh one of the ways to foretell how you're going to resolve this consolidation is leadership trends.
- 4:24
- Unidentified speaker: And the way we've classically looked at this is through the financials versus staples ratio.
- 4:30
- Unidentified speaker: And this has been something we've used through the years.
- 4:32
- Unidentified speaker: We used it going into the Liberation Day sell-off.
- 4:36
- Unidentified speaker: We used it earlier this year because both times we entered a consolidation prior to those breakdowns.
- 4:41
- Unidentified speaker: And then in each one of those cases, we saw financials roll over relative to staples.
- 4:46
- Unidentified speaker: And this is a pattern we're seeing again right now as we enter this consolidation.
- 4:49
- Unidentified speaker: So staples, that's by now everyone knows this.
- 4:52
- Unidentified speaker: Staples up by 14% in the month of October.
- 4:55
- Unidentified speaker: Everybody's trying to rotate aggressively into these non-AI areas, and that's namely staples.
- 5:01
- Unidentified speaker: And that's not healthy.
- 5:02
- Unidentified speaker: I mean, if you study history, you know, a market led by energy and staples is a really unhealthy market.
- 5:07
- Unidentified speaker: It's just not one.
- 5:08
- Unidentified speaker: We're not going to get that big bull advance that um uh you really want to have as a tailwind if you're trying to get equity beta.
- 5:16
- Unidentified speaker: And so um that was that was the second thing we saw.
- 5:20
- Unidentified speaker: And then the the other thing that we called out uh back in coming off those July, that little mini July uh period of turbulence, the Leopold liquidation lows that we we call it.
- 5:32
- Unidentified speaker: So we saw hyperscaler credit spreads start to break back up to and through those wides that we saw back there.
- 5:40
- Unidentified speaker: And so everyone knows you're gonna need a lot of financing to build to for this AI build out.
- 5:46
- Unidentified speaker: Um and and so I I think seeing those credit spreads jump above the the July levels before equities had corrected all the way back down there tells me again, our default assumption should be changing from a sustained uptrend to some kind of breakdown.
- 6:04
- Unidentified speaker: And so those were um that was the big picture things that we saw that that got us to to downgrade equities.
- 6:12
- Unidentified speaker: Um and so yeah, that's it it again, we're we really just took risk off.
- 6:17
- Unidentified speaker: I mean, we have the we there are many clearing events coming up uh and that we're going through right now.
- 6:22
- Unidentified speaker: I mean, the the midterms are today.
- 6:25
- Unidentified speaker: Um I would say that there are a few, I don't want to act like I'm some kind of insider, but at this point in my life and career, of like there's some good contacts I have, and uh they're plugged into that, the political side, and uh, they sound very concerned about the the concern over a blue, not just a blue wave, but a blue tsunami, and even the pressure that's coming um at places like um Texas state legislature to slow the pace of data center build outs, like I think that's a real risk.
- 6:59
- Unidentified speaker: And so no one wants to put capital to work.
- 7:02
- Unidentified speaker: Uh, we had a many different um issues in the month of October.
- 7:07
- Unidentified speaker: You have end of the year, tax loss selling for mutual funds and things like that.
- 7:10
- Unidentified speaker: And that's I think all kind of conspired to create this weak market environment that we're in.
- 7:14
- Unidentified speaker: And so um, I'm not sure that if we have that blue tsunami, I'm not sure that we'll we'll we'll that we're just gonna turn around on the dime.
- 7:23
- Unidentified speaker: And so we downrated equities.
- 7:25
- Unidentified speaker: I think it's best to see that range resolve itself, and or if you want to try and buy further weakness, we need to see a true sentiment or positioning wash out.
- 7:37
- Unidentified speaker: We look at things like speculative or inverse ETF uh volume versus a percentage of speculative total, so that's levered plus inverse.
- 7:45
- Unidentified speaker: It hasn't really shown a capitulatory reading yet.
- 7:49
- Unidentified speaker: We have our own internal sentiment models that takes a whole bunch of stuff and aggregates it.
- 7:53
- Unidentified speaker: We haven't gotten to extreme pessimism there either.
- 7:57
- Unidentified speaker: Uh we do things like model out uh vault target positioning, those like vault target positioning from our estimation is still like roughly one standard deviation over the rolling mean.
- 8:09
- Unidentified speaker: And so, like, there's room for them to cut exposure.
- 8:11
- Unidentified speaker: And so the bottom line is I just don't think so.
- 8:14
- Unidentified speaker: You're either going to buy a breakout, that's how we were positioned now, or you're gonna buy extreme fear, and we don't have either of those right now.
- 8:21
- Unidentified speaker: So that's that's why we neutralize risk.
- 8:24
- Unidentified speaker: I'm curious how you you know basically stay sane and trying to get signals from market structure, because you kind of hinted at it there earlier that the way markets work has changed a lot in recent years.
- 8:37
- Unidentified speaker: Like you have so many of these weird systematic flows in into markets that I'm just curious, like contrasting and trying to find signal, because to your point, when consumer staples and energy are leading, like it's sort of the you know, the final it's not a very healthy situation to be in.
- 8:53
- Unidentified speaker: Um, but in the same vein, you have like all these different market structure flows that are just becoming so much more powerful.
- 8:58
- Unidentified speaker: You mentioned like vault targeting funds, and you have like, you know, options activities just on a you know, up-only basis.
- 9:04
- Unidentified speaker: Um, you have the passive investment thing into indices, which is interesting to compare of like, you know, all the top line indices are are still you know in a pretty good spot.
- 9:13
- Unidentified speaker: We're not that far off all-time highs.
- 9:15
- Unidentified speaker: But then you look down the long tail of asset classes, and like a lot of stuff has already been struggling for months now that doesn't really go talked about.
- 9:20
- Unidentified speaker: And I'm just curious, like, how do you find that signal in this world where the market structure is just so much more noisy?
- 9:27
- Unidentified speaker: Yeah, it's an imperfect process for sure.
- 9:30
- Unidentified speaker: I do think you have to try to avoid um, I see so much.
- 9:34
- Unidentified speaker: So this is like my little editorial commentary on market philosophies.
- 9:39
- Unidentified speaker: Like, you number one, you don't want to boil the ocean, you're not going to be able to model the markets.
- 9:44
- Unidentified speaker: That's like, you know, a fool's errand, and things are constantly changing.
- 9:47
- Unidentified speaker: I see people who struggle um consistently, like one of the things I said is build conviction with fundamentals, and a lot of people don't have that technical overlay because uh conviction is weirdly synonymous with confirmation bias, and that's the the line you have to walk there.
- 10:08
- Unidentified speaker: As far as like how do you make heads or tails out of technicals, um I wouldn't say we're no one's perfect, you know, no one's perfect.
- 10:18
- Unidentified speaker: You have to just try to find you know, test everything and then throw out the noise as best as you can tell and be very honest with yourself.
- 10:26
- Unidentified speaker: And so that's what we've done.
- 10:27
- Unidentified speaker: I think that um this year in particular, given the way we saw the market structure evolve, uh one thing we would consistently do is run an analysis of like, okay, year to date, how's the SP doing?
- 10:41
- Unidentified speaker: Uh on up and down days for every industry group.
- 10:44
- Unidentified speaker: So there's like 50 something industry groups we track in the SP 500, and um, by far the most important, the biggest spread between up days and down days was the semis.
- 10:54
- Unidentified speaker: So to make that a little more clear, what I'm saying is that when uh when semiconductors are up and you you own the SP 500 on those days, on when semiconductors are down, you're not on the SP.
- 11:07
- Unidentified speaker: So, what are those two?
- 11:08
- Unidentified speaker: The distance between those two equity lines at this point of the year is over 105%, and there's no other industry with that wide of a spread.
- 11:14
- Unidentified speaker: So, what that tells you, not only have in semis grown as a like 20% of SP 500 market cap now, they're obviously they are the lifeblood of the AI trade.
- 11:24
- Unidentified speaker: It might change, like we can talk about how that could evolve going forward, but like so far, that's been the way these uh pod shops and multi-strats, if they want to be advised to be exposed to the AI trade, they want to belong that and short something else.
- 11:39
- Unidentified speaker: And so it's been a joke of ours internally.
- 11:42
- Unidentified speaker: Like, if you want to know what Hershey's are is doing this year, just tell me what SMH is doing because it's just the moving opposite of it, you know, and there's no reason, there's no reason for that.
- 11:53
- Unidentified speaker: So this is something we've written about all year, and uh, so we've really with our macro indicators and fundamental indicators, we started really pounding the semiconductor group.
- 12:02
- Unidentified speaker: We've for years looked at GPU availability and GPU rental rates, and we do a lot of work there to try and say, hey, fundamentally is the market for compute still weak loosening or weak or uh tightening.
- 12:13
- Unidentified speaker: But um our key key point is that semis have to get into gear.
- 12:18
- Unidentified speaker: And so we we were worried coming through June that they were topping, and then our hope was that July was the bottom, you know, and that we were that the market was ready to get back into gear.
- 12:29
- Unidentified speaker: Um and the charts kind of broken down again.
- 12:32
- Unidentified speaker: This episode is brought to you by Fidelity Crypto, a platform built in-house with the same discipline.
- 12:37
- Unidentified speaker: Fidelity applies to everything, so you can invest in crypto with confidence free crypto backed by industry-leading security.
- 12:44
- Unidentified speaker: Get started at Fidelity.com slash crypto.
- 12:47
- Unidentified speaker: Crypto is offered by Fidelity Digital Assets NA, is not insured by FDIC or SIPC, and includes risk of significant loss.
- 12:57
- Unidentified speaker: Fidelity Brokerage Services LLC.
- 13:00
- Unidentified speaker: Remember NYSC SIPC.
- 13:06
- Unidentified speaker: This November 10th and 11th, crypto's best institutional event is coming back to London.
- 13:11
- Unidentified speaker: The Blockworks Digital Asset Summit at Hilton Park Lane is coming up.
- 13:15
- Unidentified speaker: 1,500 attendees, more than 700 institutions, and speakers from BlackRock, JP Morgan, Coinbase, Robinhood, Figure, and more.
- 13:23
- Unidentified speaker: Just a few years ago, a lot of these institutions thought crypto was a joke, but now JP Morgan and BlackRock are on the agenda.
- 13:29
- Unidentified speaker: So there's two days of panels, networking, inside events, all building the on-chain financial future.
- 13:35
- Unidentified speaker: Grab your ticket at Blockworks.com slash events, and I'll see you in London.
- 13:43
- Unidentified speaker: All right, so I think that leads well into like this first big bucket.
- 13:47
- Unidentified speaker: That's one impacting, like you just mentioned with SMH, how how it's really one of the biggest drivers of just correlations and beta in the broad market.
- 13:54
- Unidentified speaker: But it also feels like it's a big component of what's driving the economy right now and how to unpack that.
- 13:59
- Unidentified speaker: And yeah, I'm glad you're on.
- 14:01
- Unidentified speaker: I have a lot of conflicting signals that I'm seeing in the AI complex right now that I'm trying to figure out.
- 14:06
- Unidentified speaker: So hopefully you can help us walk through here.
- 14:08
- Unidentified speaker: But you know, it just feels like we're all trying to figure out how much of this is a leverage washout in terms of the AI buildup place versus that meaningful signal.
- 14:15
- Unidentified speaker: You mentioned SMH breaking down there from a key level.
- 14:18
- Unidentified speaker: Um, and there's a lot of conflicting signals coming from the AI complex right now.
- 14:23
- Unidentified speaker: So, you know, compute is is still quite tight.
- 14:26
- Unidentified speaker: Like you you've been putting out regularly these charts of just looking at like H100 rental indice prices and that sort of thing.
- 14:32
- Unidentified speaker: And it's just like it's clear there's a lot of demand.
- 14:34
- Unidentified speaker: There's talk of that compute market being in Contango now.
- 14:37
- Unidentified speaker: So, you know, some interesting signals there.
- 14:39
- Unidentified speaker: And on the other side of things is we've just had all the hyperscaler earnings this week, and you know, some of their cap capex estimates had missed uh earned uh analyst estimates.
- 14:48
- Unidentified speaker: Um, you also have the anthropic S1 come out and how we're trying to piece through that.
- 14:52
- Unidentified speaker: So, yeah, you know, we're all trying to figure out the question is this a leverage washout or is there more fundamental aspects at play?
- 14:58
- Unidentified speaker: So I'm curious how you're trying to tease out and and and get the signal here.
- 15:02
- Unidentified speaker: Yeah, that I mean, the the getting good data is so difficult, I think.
- 15:07
- Unidentified speaker: And you have to have the projection, there's like you hate to say it, but there's like a certain amount of faith in how this technology is going to diffuse that's all baked in there.
- 15:16
- Unidentified speaker: Um, our philosophy from day one has been to be as neutral on um, because there's so much, there's like two strong sides there's the bears and the bulls, and it's like it feels extremely emotional.
- 15:28
- Unidentified speaker: And so from day one, we've felt like we don't want to get caught in that trap.
- 15:33
- Unidentified speaker: And so the best way to do that is to have um key metrics that we're gonna monitor and adhere to.
- 15:39
- Unidentified speaker: And so the first one we built was like you said, GPU availability.
- 15:43
- Unidentified speaker: And this is something that we started tracking back in 2023, and just like Fernando, who I work with, would go out and just see how easily each hour, every day, he could um it was to obtain a uh a GPU from Neo Clouds.
- 15:58
- Unidentified speaker: So we were working with the Neo Clouds on that, and uh and what we saw early this year is like the agentic explosion happened, is like our availability metrics took a massive dip.
- 16:09
- Unidentified speaker: And so that was um, and that's basically been the state through this year, especially for black wells for B200s, B200 availability has been zero.
- 16:18
- Unidentified speaker: So that's that has had a tendency to lead rental prices, and sure enough, that's what we've seen is that rental prices continued to creep higher.
- 16:26
- Unidentified speaker: And it's important, I think, when you get into this debate to remember where we came from.
- 16:32
- Unidentified speaker: And if you would have gone back, uh everyone agreed when they modeled, we also dated modeled data center um economics.
- 16:40
- Unidentified speaker: And so this is something that whenever we started modeling uh data center economics, one of the key assumptions is that GPU rents are going to decline over time, and so we haven't seen that.
- 16:52
- Unidentified speaker: We've seen availability tighten and we've seen rents rise uh uh accordingly, and that's not in anyone's model for how these data centers are going to cash flow out.
- 17:04
- Unidentified speaker: Um, so like we looked at like our core assumption was a 25% annual decline in rents, and so we haven't seen that.
- 17:10
- Unidentified speaker: So, whenever we track, and these are all real-time daily things that you can track, whenever we do that, we see a very tight market for GPUs for compute emerging.
- 17:21
- Unidentified speaker: Now, like you said, another thing that's happened here recently, uh Silicon Data, who are we're friends with them, and Steve Howe over there is uh is someone who I've talked to for years, and he's he's moved from Bloomberg over there, and uh they do great work.
- 17:36
- Unidentified speaker: And they've started looking at term structure, and there's some other companies that show you like one year out uh commitments, and it does seem that uh farther dated uh rental prices, rents are going up, are now in Contango.
- 17:51
- Unidentified speaker: And so that offers another confusing thing because this is obviously my background is oil, and you see a lot of people take that um that Contango and they misinterpret it and they think that's a weak sign because that is a weak sign in the oil market when you have a physical commodity that has to where the curve is managing inventories.
- 18:08
- Unidentified speaker: That's not the case here.
- 18:10
- Unidentified speaker: I think that you have a convenience yield in a physical commodity world, you have a capacity premium that's emerging here.
- 18:17
- Unidentified speaker: So I think those farther out futures prices point to people locking in capacity and longer dated when longer dated uh rentals trade above spot.
- 18:29
- Unidentified speaker: I think that's a very positive sign for how I understand commodity markets and the emerging GPU market.
- 18:35
- Unidentified speaker: So everything on the GPU side looks good.
- 18:39
- Unidentified speaker: Um, and I would say, in addition to that, the the cloud revenues from the hyperscalers are now starting to ramp up.
- 18:46
- Unidentified speaker: And we saw that in the Q2 reporting season.
- 18:50
- Unidentified speaker: Um, all the hyperscalers rallied.
- 18:52
- Unidentified speaker: I know that we're in a different kind of market regime where everyone's nervous now, but I feel like the return on investment debate is kind of being ground down into nothing.
- 19:02
- Unidentified speaker: I think that the returns are there and you're starting to see them um emerge.
- 19:08
- Unidentified speaker: And then finally, I think the athletic S1, um, you know, like the the fundamentals are pretty impressive from everything in that S1, in my opinion.
- 19:17
- Unidentified speaker: I think if you look at the way revenue per dollar of compute spend has trended, gone from 57 cents in Q2 of last year up to $2.08.
- 19:27
- Unidentified speaker: It dropped a little bit with their release here in Q3, but it's still strong margins.
- 19:32
- Unidentified speaker: And the other factor is that like inference keeps dropping as a percentage of their compute spend.
- 19:36
- Unidentified speaker: So if you ran that same dollar of um revenue per inference compute spend, it has trended up every quarter, including Q3.
- 19:44
- Unidentified speaker: So I think um I think the and I'm saying all that without referencing the tech and what the tech can do in my experiences, which inform a lot of what we do as investors.
- 19:56
- Unidentified speaker: I'm just saying here's the objective facts.
- 19:58
- Unidentified speaker: The objective facts tell me that the AI trade is in intact still.
- 20:03
- Unidentified speaker: Yeah, I think it's some like you you mentioned this notion of how emotional this discussion has become.
- 20:10
- Unidentified speaker: And I feel like so much of it is because people are misinterpreting their, you know, the fundamental use cases and the value you get out of these tools, which like anybody who's tapped in, I think like, you know, nobody's nobody's debating that.
- 20:23
- Unidentified speaker: I think it's just a matter of like how dialed in you are on these tools to understanding like the potential use cases.
- 20:28
- Unidentified speaker: And it's just like I'm I'm on the same on the same boat there.
- 20:31
- Unidentified speaker: Like the the value there is just absurd.
- 20:33
- Unidentified speaker: I I don't think it's even up for debate.
- 20:35
- Unidentified speaker: Um, what is maybe a bit more often debate is just who's financing it and what does that runway look like.
- 20:41
- Unidentified speaker: Um and that's where I'm trying to figure out like what you know, if you just think about the downstream from the CapEx estimates from these hyperscalers that we got last week, like you know, it is one of the reasons that we missed on those estimates just because of how supply crunched we are on other aspects to actually make this this build-out happen.
- 20:59
- Unidentified speaker: And then obviously, you know, we had the midterms today, and there's a lot of talk of like, you know, the Texas Texas Texas governorship run, um, whoever wins there is gonna matter a lot to how much of these data centers we can build out.
- 21:11
- Unidentified speaker: So, like, how much of are are those weird aspects of just the supply crunch dynamic affecting the CapEx estimates as opposed to like an actual demand for data center and inference?
- 21:21
- Unidentified speaker: I think it's um uh really hard to disentangle.
- 21:24
- Unidentified speaker: I mean, it's there's you have capital market limitations in the face of the Fed hiking, which I think is all tied together.
- 21:29
- Unidentified speaker: I mean, you've had you've gone from going into the year or right before the war, we had another 75 basis points of cuts priced in the sofa market.
- 21:39
- Unidentified speaker: Now we had as many as six hikes priced out to 2027 for this cycle.
- 21:45
- Unidentified speaker: So things have shifted rapidly, and funding markets have also, you know, we've had so many companies hammer funding markets, and everyone's trying to get in before these IPOs, the lab IPOs.
- 21:56
- Unidentified speaker: And, you know, there's a there's a little bit of like a pull forward effect going on.
- 22:01
- Unidentified speaker: And so I acknowledge those those are real risks.
- 22:04
- Unidentified speaker: And uh this is one of the reasons we were watching credit spreads as like a key uh metric for us as far as managing risk.
- 22:11
- Unidentified speaker: And then you have politics.
- 22:13
- Unidentified speaker: I can't sit here and say I understand how all that's going to unfold, right?
- 22:18
- Unidentified speaker: Uh, I I can't, I have no idea.
- 22:20
- Unidentified speaker: The one thing we do is we just we monitor um satellite data for every every data center that's under construction, and we've created a pacing model.
- 22:29
- Unidentified speaker: This is another thing on the supply side.
- 22:31
- Unidentified speaker: So, you know, all the GPU metrics I went through are on the demand side, but on the supply side, what we we're doing is uh trying to track the pace of every project that's under construction, um, from groundbreaking to an energized shell, and then backing in, how does that align with the uh the estimates, earnings estimates, revenue estimates for the semiconductor industry?
- 22:57
- Unidentified speaker: And so far, what we're seeing is an alignment.
- 22:59
- Unidentified speaker: We're not seeing a supply shortfall yet in our satellite data.
- 23:03
- Unidentified speaker: So I think it's much more, I mean that's how markets work, it's much more anticipatory right now than reality that we can't um add.
- 23:12
- Unidentified speaker: We have our models suggest there's gonna be an additional 27 plus gigawatts coming online over 2027, um, which does, in our in our view, that supports the the top line estimates for the semiconductor group.
- 23:27
- Unidentified speaker: And so in our view, that's everything aligns.
- 23:30
- Unidentified speaker: It makes sense there.
- 23:31
- Unidentified speaker: Now, if you do hit a supply, like politics, um, if there's a supply or some kind of power shortage or material shortage or labor shortage that slows the pace down, we should see that in our model, we should see that in the data.
- 23:44
- Unidentified speaker: As far as how the market would move, like it doesn't make sense for me to see hyperscalers and semis sell off in that scenario.
- 23:51
- Unidentified speaker: Like, I uh that's where I think you can kind of see like this market is liquidating and degrossing in general, and everything's being sold.
- 24:00
- Unidentified speaker: One of the big phenomenon that was going on earlier this year was SIMI's up, hyperscalers down.
- 24:06
- Unidentified speaker: In our theory, which it was that these things are going to catch back together because it's kind of all part of one trade.
- 24:12
- Unidentified speaker: Um, if AI is like really um hitting on all cylinders, but instead now you're seeing them both sell off together.
- 24:19
- Unidentified speaker: And if that's a supply worry, it doesn't make sense.
- 24:21
- Unidentified speaker: Because if it's a supply worry, then the owners of the data center capacity that gets all the gets in front of that supply crunch, that's going the economics there is going to continue to be strong.
- 24:32
- Unidentified speaker: And uh of course, you're you can't you know restrict capex if that's what the market's cheering on, then you'll get like great returns in your your data center business, compute market stays tight, even artificially tighter, and at the same time your capital requirements go down.
- 24:50
- Unidentified speaker: So, like you need to own hyperscalers in that scenario.
- 24:53
- Unidentified speaker: I could definitely see that being bearish for semis, but not for both.
- 24:56
- Unidentified speaker: So to me, this price action is panicky and um doesn't it doesn't make sense with that supply crunch scenario.
- 25:07
- Unidentified speaker: I'd be wanting to own the hyperscalers.
- 25:10
- Unidentified speaker: I already want to own the hyperscalers when this thing's said and done, but it it would they would become even more attractive in a supply crunch scenario, in my opinion.
- 25:19
- Unidentified speaker: Awesome.
- 25:20
- Unidentified speaker: Yeah, so just to simply summarize that is like you can't have GPU rental prices increasing like they are, and hyperscalers and semis down at the same time, unless it's a liquidation varshock type of scenario.
- 25:35
- Unidentified speaker: Exactly.
- 25:36
- Unidentified speaker: Exactly.
- 25:36
- Unidentified speaker: This is and so and if it is a liquidation varshock, it goes back to our technical overlay and managing risk and maintaining your fundamental conviction as we go through this period.
- 25:47
- Unidentified speaker: So you maintain your fundamental conviction.
- 25:49
- Unidentified speaker: If my if I saw something change fundamentally, if GPU prices or availability, if availability spikes and GPU prices plummet, and it looks like we did an overbuild, there's absolutely no signs of an overbuild.
- 26:01
- Unidentified speaker: And if we hit a supply shock, it's gonna be an underbuild.
- 26:05
- Unidentified speaker: So all these things get tighter.
- 26:06
- Unidentified speaker: And so if I saw that, then my fundamental conviction would change.
- 26:10
- Unidentified speaker: But instead, everything you know, it looks it looks it looks like just a technical var var shock.
- 26:16
- Unidentified speaker: Yeah, yeah.
- 26:17
- Unidentified speaker: I'm with you.
- 26:17
- Unidentified speaker: I'm with you.
- 26:18
- Unidentified speaker: Take a disciplined approach to crypto investing with Fidelity Crypto, a platform built in-house using over a decade of crypto experience.
- 26:26
- Unidentified speaker: At Fidelity, you can trade crypto and stocks together backed by industry-leading security.
- 26:31
- Unidentified speaker: When you can see your accounts in one place, it's easier to make smarter decisions.
- 26:35
- Unidentified speaker: Get started at Fidelity.com slash crypto.
- 26:38
- Unidentified speaker: Crypto is offered by Fidelity Digital Assets.
- 26:41
- Unidentified speaker: NA is not insured by F D I C or S I P C and includes risk of significant loss.
- 26:48
- Unidentified speaker: Securities offered by Fidelity Brokerage Services LLC.
- 26:52
- Unidentified speaker: Remember NYSC, S I P C.
- 26:58
- Unidentified speaker: This November 10th and 11th, crypto's best institutional event is coming back to London.
- 27:03
- Unidentified speaker: The Blockworks Digital Asset Summit at Hilton Park Lane is coming up.
- 27:07
- Unidentified speaker: 1,500 attendees, more than 700 institutions, and speakers from BlackRock, JP Morgan, Coinbase, Robinhood, Figure, and more.
- 27:15
- Unidentified speaker: Just a few years ago, a lot of these institutions thought crypto was a joke, but now JP Morgan and BlackRock are on the agenda.
- 27:22
- Unidentified speaker: So there's two days of panels, networking, inside events, all building the on-chain financial future.
- 27:27
- Unidentified speaker: Grab your ticket at Blockworks.com slash events, and I'll see you in London.
- 27:35
- Unidentified speaker: All right.
- 27:36
- Unidentified speaker: So I want to talk about, and this starts to lead into how the Fed is reacting to all of this, but the the economic impact of this build out.
- 27:42
- Unidentified speaker: Like it's it's very clear that we're in an investment boom right now.
- 27:45
- Unidentified speaker: There's there's a lot of um financial assets that are being issued to fund this right now, and that has to be digested one way or another.
- 27:52
- Unidentified speaker: Um, but I'm curious about how you discern this distinction between what's going on in terms of what's leading so much of this economic growth, which is this build out and this investment boom with the rest of the economy.
- 28:03
- Unidentified speaker: Um, and then of course, how the Fed is trying to navigate this situation.
- 28:06
- Unidentified speaker: Like this is a very unique situation to be in where you have these two forces in polar opposites.
- 28:11
- Unidentified speaker: You know, you look at you know, stuff like you know, McDonald's stock or whatever, which has just been having like a real tough year at the same time that like so much of this is happening and like the housing market is frozen.
- 28:20
- Unidentified speaker: So yeah, just curious how you think about that whole dynamic and how how the Fed starts to navigate this.
- 28:25
- Unidentified speaker: Yeah, so that's to me, um, we get out of the fundamentals and we get out of the technicals and we go to the macro, and to me, macro is the intermediate term risk.
- 28:34
- Unidentified speaker: You know, that's that throws more um I'm it's a bearish situation.
- 28:39
- Unidentified speaker: It has evolved to become a more bearish situation.
- 28:41
- Unidentified speaker: So I just kind of say we came through the year, like I said, before the war, we were still in a cut cycle.
- 28:48
- Unidentified speaker: There are a whole different set of rules when the Fed's got your back and it's cutting side and cutting.
- 28:52
- Unidentified speaker: It's like the difference between the Fed having your back and trying to break your back is what we say.
- 28:56
- Unidentified speaker: And we've quickly transitioned with the war starting out of a cut cycle into a hike cycle, and now we actually are here in that hike cycle has realized.
- 29:08
- Unidentified speaker: The um the drivers of rates has changed.
- 29:11
- Unidentified speaker: So I I see if I can say this in the right way.
- 29:14
- Unidentified speaker: So one of the big debates that's gone on, especially in this Kevin Warsh era, is like um, what is the Fed, what is what's driving rates, right?
- 29:24
- Unidentified speaker: That's like a big thing, you know.
- 29:26
- Unidentified speaker: And so for a long time, the tenure or the the tenure and oil were related, right?
- 29:32
- Unidentified speaker: And that's like that was it gave us some hope that the uh if the war were to end suddenly and oil prices reverted to their pre-war norm, you could have uh relief in in rates.
- 29:44
- Unidentified speaker: Uh then that flipped, and this is the concerning part, it's flipped to um the correlations flipped more towards like NASDAQ.
- 29:52
- Unidentified speaker: So Nasdaq and tenure is now much more correlated than the tenure is with oil.
- 29:58
- Unidentified speaker: And so that goes to your point.
- 29:59
- Unidentified speaker: Like, there's this other thing now, the handoff of inflationary um causes has gone from the the war in oil to the build out, and that's a problem.
- 30:11
- Unidentified speaker: That's a problem.
- 30:12
- Unidentified speaker: And the the thing I would say just to be to make this this technically correct is like the one thing that hasn't changed through all this is the relationship between the sofa market and the tenure.
- 30:24
- Unidentified speaker: Its correlation has been like 90 plus percent the whole time.
- 30:27
- Unidentified speaker: So, what this reminds you is in and everyone is in the worst era where we want to step back, like the the the Fed's movements and how the market sees those is always at the center of the world.
- 30:38
- Unidentified speaker: It's just what's moving the Fed.
- 30:40
- Unidentified speaker: What is the Fed focused on?
- 30:42
- Unidentified speaker: They were focused on oil, their speech is focused on oil, and now they're focused on AI in the build out, and that's the big change.
- 30:50
- Unidentified speaker: And so with that change, the risk is that the Fed doesn't stop until they've broken something in the AI trade in the funding markets.
- 30:59
- Unidentified speaker: And uh that's a there's a less clean path out of this hike cycle with um the Fed focusing on tech and AI build out than it was back when we were focusing on just ending the war.
- 31:12
- Unidentified speaker: And you had a supply-side disruption of oil that we could we could um that could back the Fed off.
- 31:18
- Unidentified speaker: And so those are that's my thoughts, and that's how the the macro has evolved in what I would say is a disconcerting way.
- 31:25
- Unidentified speaker: The other thing, just to finish my thoughts on it, is um the the in this so it shouldn't be all that surprising when you really zoom out, is that we've never seen the Fed cutting into a capex cycle.
- 31:37
- Unidentified speaker: We've had that for the first time ever, we had the Fed cutting from 24 to um through last year as CapEx was ramping.
- 31:45
- Unidentified speaker: And so it was a little bit you could you could point to that and say, well, that was a policy mistake.
- 31:50
- Unidentified speaker: I don't, I'm not there, but uh it's not surprising entirely to see this uh this inflation pressure starting to come out in this later cycle part of the uh capex cycle.
- 32:04
- Unidentified speaker: How important is it?
- 32:05
- Unidentified speaker: Like just honing in on the specific component of the build out, which is the the debt issuance component of the build out.
- 32:11
- Unidentified speaker: Like obviously things were all well and good when we could fund this whole build out unlevered with free cash flow that was in excess, but obviously that's been tapped now.
- 32:19
- Unidentified speaker: So now we're in this phase of waiting between equity issuance and debt issuance.
- 32:23
- Unidentified speaker: Um, I imagine the Fed isn't too worried about the equity issuance side of things, but the debt market, you know, we've we've seen some some pretty big moves and things like uh the core weave uh bond market and the and the spreads there and that's and that associated component.
- 32:36
- Unidentified speaker: How yeah, how important do you think is the fact that like the stage we're in now is the one where it's so dominant by actual issuance of securities, with how that contrasts with how the Fed and the reaction function is so tied to you know rates markets.
- 32:49
- Unidentified speaker: Like credit spreads are gonna move and the Fed is hiking.
- 32:51
- Unidentified speaker: That just kind of makes sense.
- 32:52
- Unidentified speaker: Yeah.
- 32:53
- Unidentified speaker: Well, I mean, I think we're finding like, and I've I was guilty of saying this at one point.
- 32:57
- Unidentified speaker: So like I could like it's it's um I'm I'm happy, I'm it's free to say, like, I I'm rethinking how interest rate policy interacts with the AI build out.
- 33:06
- Unidentified speaker: Like earlier this year, it would be easy.
- 33:08
- Unidentified speaker: The common wisdom was that like, well, the Fed can't dent this AI build out, you know.
- 33:14
- Unidentified speaker: And I think that was true when they were funding, like you said, the CapEx build out through cash flow.
- 33:20
- Unidentified speaker: But now that they the external funding markets are there as a as a limiting factor on the supply side, like we said, then yeah, I think that this becomes a real risk.
- 33:30
- Unidentified speaker: If the Fed's gonna continue hiking, if the inflation data doesn't calm down, then I do think you're going to inevitably at some point slow down this cycle.
- 33:40
- Unidentified speaker: Again, I don't think that's like a death blow to the AI trade, but it would be a nasty path to get there, especially for SIMI's.
- 33:46
- Unidentified speaker: Like SIMIS would suffer.
- 33:47
- Unidentified speaker: And as we said, SIMI's been an important leadership group for the market.
- 33:50
- Unidentified speaker: And so the implications aren't great, but if this happens, it's it's a I would say it's a it leans bearish if we go that way.
- 33:58
- Unidentified speaker: Uh so um, yeah, I I think it's a big deal.
- 34:03
- Unidentified speaker: The other fat way the feds are can impact the AI trade, something that we looked at, and I think a lot of other people are starting to look at it, is how uh the VC market has been a huge consumer of of AI.
- 34:15
- Unidentified speaker: And we estimated that uh VC startup companies are about 25% of lab ARR.
- 34:22
- Unidentified speaker: So if we really killed um, if we really killed uh the VC funding cycle with the Fed hike, with Fed hikes and higher interest rates than uh similar to 2022, then you could see a pretty big chunk of lab ARR come out from just from that group.
- 34:39
- Unidentified speaker: So there are channels that interest rate policy is going to impact this thing.
- 34:44
- Unidentified speaker: And so that's like you need to have um, I think that just means like the macro is the risk right now on the intermediate term.
- 34:52
- Unidentified speaker: That's that's the risk.
- 34:53
- Unidentified speaker: Not this, I don't think it's the uh the fundamentals of the AI build out and to the extent I can separate those things.
- 35:01
- Unidentified speaker: I think it's the the the fact the the risk that interest rate policy chokes off um or slows down the supply uh side.
- 35:12
- Unidentified speaker: How much of a of a subscriber are you to the geopolitical game theory potentialities here around you know the AI build out and winning the AI race being a national security issue, one might argue, and that we we need to do everything we can to get to the other side of this.
- 35:30
- Unidentified speaker: And the notion that um, you know, a freshly approved Fed share that was tapped from the Trump administration is at the head here, who's a a genuine believer in this in the AI story.
- 35:42
- Unidentified speaker: Um, how much do you subscribe to the notion that there are geopolitical game theory games here to be cognizant of that we don't uh we don't tighten too much that we start to question the ability to get to the other side here of this race?
- 35:54
- Unidentified speaker: I think there's something to it.
- 35:56
- Unidentified speaker: I I mean the feds that that's ultimately when I boil it all down.
- 36:00
- Unidentified speaker: I I I don't think the feds are just gonna kill the AI trade because that because of when you get right down to it.
- 36:07
- Unidentified speaker: Number one, I do think that the the data would slow the market would we've already seen market instability with some some minor credit stress, right?
- 36:16
- Unidentified speaker: And the market instability, if we've got worse from here, would it's already in Worsh already acknowledged it.
- 36:22
- Unidentified speaker: He said, I'm unable to say financial conditions are tight right now, but he he wobbled versus like his Jackson Hull speech.
- 36:30
- Unidentified speaker: He's he's changing.
- 36:31
- Unidentified speaker: So you get another uh credit wobble, and they're gonna be done hiking, you know, and it won't and it won't be far before they start cutting, and you know, we start talking about the importance of this.
- 36:43
- Unidentified speaker: Uh, this this to a lot, maybe it won't be overt, but like to me, that's everyone knows this is a very important thing.
- 36:50
- Unidentified speaker: And to crash the AI trade or crash the AI build-out right now is too um it there's a lot of downstream consequences.
- 36:59
- Unidentified speaker: I think the Fed is in a really tough spot in general because their hiking rates uh now to respond to the political dissatisfaction that's cropped up since the co since really since COVID.
- 37:13
- Unidentified speaker: I think this is all one big story that's just evolved since we started blowing deficits out post-COVID.
- 37:19
- Unidentified speaker: And um, so now the Fed is even when they started hiking kind of in response to the Iran War, I I think you can read it in some of Waller's commentary, they're starting to become very self-conscious of the inflation story and and ultimately the political backlash.
- 37:33
- Unidentified speaker: And I think when you boil it all down, they actually risk getting more heat um if they if they damage the i AI trade um in a meaningful way.
- 37:45
- Unidentified speaker: So yeah, I think it's yeah, I I think it's a real factor.
- 37:48
- Unidentified speaker: You're probably not gonna hear about it, but I think it's it's absolutely operating in the background.
- 37:53
- Unidentified speaker: How how sensitive do you think the economy is to this potential AI wobble?
- 37:57
- Unidentified speaker: Like if the rest of the economy kind of sucks, which it which it does, um, and then this one key driver of so much economic growth starts to wobble, like how quick is the sensitivity there?
- 38:08
- Unidentified speaker: Are we talking about like quickly positioning for a gross slowdown or even a recession, or do you see it as just more a kind of chopping around here?
- 38:16
- Unidentified speaker: I think it's more um chopping because we still have that we so like we have a fiscal impulse that's sort of fading, which is another big factor, obviously, but we still have a pretty wide deficit right now.
- 38:27
- Unidentified speaker: Um, and the thing is like just like what we've seen in the markets where when AI trade sells off, to the extent rates come down when the AI trade comes off, you see a bid into things like staples and rate sensitive pockets of the market.
- 38:39
- Unidentified speaker: And I think that that's reflecting some, it's it's a little overdone in the markets, but it reflects some reality economically.
- 38:46
- Unidentified speaker: So if you were to cool the AI trade off a little bit, and right, my view is again, if the two years at 4-6, um, I think the Fed has one more hike in it, and basically they're that they're done.
- 39:01
- Unidentified speaker: They're they'll probably have a a hiking bias that they'll maintain or a hawkish bias from there.
- 39:06
- Unidentified speaker: Uh, I think that um if that hawkish bias goes away and they say, okay, we're done, we're done, done, then um I think that you have room to move the two-year down.
- 39:17
- Unidentified speaker: I think the the yield curve has room to flatten a little bit, and that opens up, that reaccelerates some of these areas of the economy that have been um kind of crowded out by the AI build out.
- 39:28
- Unidentified speaker: So I think that between the deficits and the way the AI credit wobble would interact with interest rates, I have a hard time getting to like a true uh recessionary scenario in the in the current environment.
- 39:44
- Unidentified speaker: Fair enough.
- 39:45
- Unidentified speaker: All right, last big bucket to talk about.
- 39:47
- Unidentified speaker: Um the Iran war, the energy.
- 39:51
- Unidentified speaker: I don't know if you'd believe it to be a crisis or not quite yet, but uh lots going on there in terms of how it impacts the economy.
- 39:57
- Unidentified speaker: Curious your take as it stands.
- 39:59
- Unidentified speaker: There's obviously a lot of unknown right now.
- 40:02
- Unidentified speaker: Uh, we're recording this on the day of the midterms.
- 40:04
- Unidentified speaker: There's a lot of speculation of whether a meaningful escalation comes after the midterms.
- 40:08
- Unidentified speaker: But uh yeah, obviously your background is in energy and oil.
- 40:12
- Unidentified speaker: Um, when you count the barrels out there and where it stands, and also very curious about how you how you see cracks beds and distillates and just how like elevated diesel prices have been and the impact on on inflation.
- 40:22
- Unidentified speaker: Yeah, how's how's the landscape as it stands right now?
- 40:26
- Unidentified speaker: Well, I mean, it's obviously the markets are a mess.
- 40:29
- Unidentified speaker: I think nobody has conviction to put any big trades on right now.
- 40:33
- Unidentified speaker: Um, like the little oil traders that are on my service that I talk to, I mean, it's a really difficult market to to to navigate because you have, I really do think the next big move is gonna get be dictated by the geopolitics.
- 40:47
- Unidentified speaker: And uh the the range of opinions on that are is actually wider than uh the consensus expects.
- 40:54
- Unidentified speaker: I think the consensus got really crowded into this idea that inevitable that we were gonna re uh start the conflict.
- 41:01
- Unidentified speaker: Um I don't know.
- 41:03
- Unidentified speaker: I I've I've always been a little skeptical of that, of that viewpoint that we're gonna just go right back to a bombing campaign because as we said it uh early on in this thing, like there's really like four main channels that you can prosecute this war from the US's perspective.
- 41:19
- Unidentified speaker: You have what I would call discriminate bombing, where you're hitting certain sites, but you're not blowing up oil infrastructure, you're not hitting civilian infrastructure.
- 41:28
- Unidentified speaker: We did that.
- 41:29
- Unidentified speaker: You have uh indiscriminate bombing, which is you know going to tactical nukes or mother of all bombs or whatever, and you're hitting less more a wide variety of targets, including energy infrastructure.
- 41:42
- Unidentified speaker: We haven't done that.
- 41:43
- Unidentified speaker: We've threatened it.
- 41:44
- Unidentified speaker: Trump threatens it all the time, but we haven't done it.
- 41:48
- Unidentified speaker: And I think that uh he the reason we haven't is because he knows the response would be devastating to the regional infrastructure, to the Gulf Coast countries.
- 41:57
- Unidentified speaker: And that the GCC has not been on board with that kind of a move.
- 42:01
- Unidentified speaker: You have a ground invasion, that would be the third way we can do this.
- 42:05
- Unidentified speaker: We haven't taken that route either because there would be mass casualties.
- 42:09
- Unidentified speaker: There is no political appetite for that.
- 42:12
- Unidentified speaker: Um and so when I look around, I you know, you look at what's gonna happen with the elections.
- 42:16
- Unidentified speaker: I think that the the um it would only get worse for the Republicans.
- 42:20
- Unidentified speaker: And and if we get a Democrat sweep that sweep, then you're gonna have less scope to do a ground invasion.
- 42:26
- Unidentified speaker: And then the final strategy was the blockade, and we're doing the blockade, and I think the blockade is working.
- 42:31
- Unidentified speaker: I think we are choking off Iranian exports, Iranian floating storage is down.
- 42:36
- Unidentified speaker: But um uh at the same time, a blockade, everyone knows that it requires a lot of patience to execute properly.
- 42:42
- Unidentified speaker: And there's, you know, does Trump have that kind of patience to see the strategy through?
- 42:46
- Unidentified speaker: I think that's the question.
- 42:47
- Unidentified speaker: But when I look at all those things, I just don't see anywhere.
- 42:50
- Unidentified speaker: Um there's no movement.
- 42:53
- Unidentified speaker: And then so the final thing I'd say is that you kind of look towards some kind of a deal.
- 42:57
- Unidentified speaker: And I think a deal post-election is more likely than a pre-deal pre-election because more than likely the deal's gonna look pretty uh there's some elements to it that are gonna look pretty bad politically, yeah.
- 43:08
- Unidentified speaker: Uh versus the status quo.
- 43:10
- Unidentified speaker: And so I don't know, but it's gonna be the next, the next move is going to dictate the price of oil, uh, in my view.
- 43:17
- Unidentified speaker: I think that we're we're running low on the SPR release.
- 43:21
- Unidentified speaker: The European release is of of which included a lot of distillates, the G7 release is also, you know, we have a couple months there.
- 43:30
- Unidentified speaker: So this thing's gonna need to get figured out over the I call it the next uh 45 days, in my opinion.
- 43:37
- Unidentified speaker: And you're seeing the Chinese lean on the Iranians too.
- 43:39
- Unidentified speaker: So I think that was the element that I always expected to come in at some point, and uh you're starting to hear that chatter.
- 43:47
- Unidentified speaker: I think I think that China is is sick of drawing down their own internal inventories.
- 43:53
- Unidentified speaker: Um and so yeah, and that we've had look, I mean, over 60% of the VLCCs in the global uh the global uh fleet uh were laden last I checked.
- 44:06
- Unidentified speaker: So I mean this is totally unbalanced.
- 44:08
- Unidentified speaker: The shipping is unbalanced, crack spreads, as you referenced, are totally blown out.
- 44:12
- Unidentified speaker: The diesel market is is a mess, middle distillates are a mess.
- 44:16
- Unidentified speaker: Um, and that's why you're seeing Brent WTI.
- 44:20
- Unidentified speaker: So it's blown out too.
- 44:21
- Unidentified speaker: I mean, there's this market is a mess, and then you look at the futures market, there isn't an extreme.
- 44:26
- Unidentified speaker: Like my tell navigating this whole thing has been waiting for extremes and fading those on positioning.
- 44:32
- Unidentified speaker: And what I see is positioning has been totally squared, and no one wants to take uh a strong stance ahead of the next move, and so we're all left kind of trying to guess where Trump's head's at.
- 44:44
- Unidentified speaker: Um, yeah.
- 44:45
- Unidentified speaker: Yeah.
- 44:45
- Unidentified speaker: So positioning-wise, I've always said you have to have an overweight energy within a balanced portfolio because it's just there's no other way to hedge the risk that this thing goes sideways.
- 44:56
- Unidentified speaker: Um, yeah, knowing that we can't predict what that final clearing event is going to look like, whether it's a deal or just an all-out uh attack that tries to truly end this thing, um, curious what you think about what might be some of the more durable changes to the composition of the oil landscape industry industry, especially around the US.
- 45:16
- Unidentified speaker: I mean, there's there's this notion that you know the the strait has been half closed, half open, but yeah, the blockade has been seemingly working.
- 45:23
- Unidentified speaker: Um, and during that time, you know, the the oil market and global supply chains are a dynamic system and are gonna gonna adjust.
- 45:31
- Unidentified speaker: Um and so I'm curious how you think about what might be some of the more permanent impacts from this this you know Hormuz crisis and and what that could do, especially for the US as as one of the the largest producers in the world.
- 45:46
- Unidentified speaker: Like, do you see certain changes that are started to happen and that are gonna accelerate and that might actually be like a durable permanent change to the oil landscape, knowing that we don't actually know how the final clearing event is gonna look like?
- 45:58
- Unidentified speaker: Yeah, um it's hard to say.
- 46:02
- Unidentified speaker: I I think that uh um I there's there I think that the number one thing is gonna happen is that there's probably a good chance that we uh uh it could accelerate, I think, the any transition, decarbonization transition that was already underway.
- 46:20
- Unidentified speaker: I think that you're gonna see uh less presence, U.S.
- 46:23
- Unidentified speaker: presence in the Middle East and around the Gulf Coast.
- 46:26
- Unidentified speaker: I think you're already seeing Trump trying to um take a step back there.
- 46:30
- Unidentified speaker: I mean, this is kind of like when you zoom in, this is all part of uh the uh multipolarity that's emerging, right?
- 46:38
- Unidentified speaker: I mean, like this is I I kind of I kind of hate talking like this because it's so it's like so macro um galaxy brain stuff, you know, and I don't want to go, I would like to stay, I prefer to stay very concrete with my analysis.
- 46:52
- Unidentified speaker: But in the it there is a reality here.
- 46:54
- Unidentified speaker: The reality is that this world, um, this unipolar world where we were policing the Strait of Hormuz and we are policing all of the we're policing the um South China Sea, we're policing, we're policing every region at the same time, you're seeing those that those capabilities have been stretched and exposed to a certain extent through this um through this conflict.
- 47:21
- Unidentified speaker: And so I think it fast forwards, and you're seeing like to end the war, we're having to have um we're bringing China in, we're bringing Russia in, we're having multiple, multiple major power discussions.
- 47:33
- Unidentified speaker: Um, again, I hate talking like that um on some level, but it's the truth.
- 47:38
- Unidentified speaker: And that's the that's the shift that I think is accelerating through this process.
- 47:43
- Unidentified speaker: And I do think that um it's almost objectively true.
- 47:48
- Unidentified speaker: This was a strategic miscalculation to do this um from the US's perspective.
- 47:54
- Unidentified speaker: So I think we've lost soft power.
- 47:56
- Unidentified speaker: Um, our hard power is in more question.
- 47:59
- Unidentified speaker: The multipolar world that everyone in the macro community, fourth turning types, has talked about for years.
- 48:06
- Unidentified speaker: You can see the um edges of that start to really expose themselves.
- 48:12
- Unidentified speaker: And this is an accelerant.
- 48:14
- Unidentified speaker: All right, last question for you on oil markets, and we'll, yeah, we'll come back down to Earth from the macro pie in the sky.
- 48:19
- Unidentified speaker: But it's it's kind of a random one, but just something I've been trying to figure out, you know, as a I I don't trade energy markets by any means, but I've been trying to figure out like those crack spreads in the distillates aspect of of what typically resolves and tightens those those spreads, because just on my basic understanding, which is that look with with spreads where they're at, refiners are just like exceptionally profitable right now.
- 48:42
- Unidentified speaker: Um I would imagine that for the for that spread to resolve, they they would just you know keep keep ripping these things out there until that starts to normalize versus the price of oil.
- 48:52
- Unidentified speaker: But it's surprised me that it's it just doesn't seem like that is happening.
- 48:55
- Unidentified speaker: So I'm just curious, like in in your expertise in this regard, like how does that typically resolve itself and why hasn't it yet?
- 49:02
- Unidentified speaker: Typically, it resolves itself by um refiners retooling to produce more of the product where you see these elevated crack spreads that then supply comes on and crack in and brings that down.
- 49:15
- Unidentified speaker: And you also have some level of demand destruction.
- 49:18
- Unidentified speaker: I do think there is some demand destruction going on at 200 plus dollar barrel diesel, which is where we we've been for now months.
- 49:27
- Unidentified speaker: It takes some time to work through the system.
- 49:29
- Unidentified speaker: Like if you're running middle distillates, if you're using heating oil, um, you may be able to switch.
- 49:35
- Unidentified speaker: There is some switching ability there.
- 49:38
- Unidentified speaker: And so you'll get a mix of demand destruction and some supply from refiners.
- 49:42
- Unidentified speaker: The exceptions to that rule, and when you see these crack spreads stay high like this for a sustained period, is when there's structural changes.
- 49:50
- Unidentified speaker: And there are some structural things going on right now that that make it um did this a little have a little bit of that flavor too.
- 49:58
- Unidentified speaker: So, like when we had um back in the days of the early shale revolution, when you had a ban on crude oil exports, those were the days, those were the salad days for refiners.
- 50:09
- Unidentified speaker: Those like so you had an export ban for crude but refined product you're allowed to export as a refiner.
- 50:15
- Unidentified speaker: So those types of structural periods, if you're studying history, you'll see um long periods of sustained high crack spreads, but it requires that kind of like structural oddity.
- 50:27
- Unidentified speaker: So I think we're gonna end up um a little mix of both.
- 50:31
- Unidentified speaker: And you're gonna see a reversion of these spreads.
- 50:33
- Unidentified speaker: You'll see refiners also will ultimately bid up oil.
- 50:37
- Unidentified speaker: You'll see oil prices stay sticky high.
- 50:40
- Unidentified speaker: Um, you can't have really a collapse in the price of oil when you have a uh whatever it is, $70 crack spread for uh middle distillates.
- 50:50
- Unidentified speaker: I mean, last I checked what uh Singapore gas oil cracks were above $70 a barrel.
- 50:56
- Unidentified speaker: And I don't know where they're at today.
- 50:58
- Unidentified speaker: But um, you know, I mean these these are really high levels, and like the refining business is not a business to get married to.
- 51:11
- Unidentified speaker: Here, and uh they're the most cyclical um uh businesses in the world.
- 51:16
- Unidentified speaker: You go back, um gosh, look at that uh was it Lee Raymond?
- 51:25
- Unidentified speaker: Was it Lee Raymond, the old Exxon CEO?
- 51:27
- Unidentified speaker: I think it was Lee Raymond.
- 51:29
- Unidentified speaker: You go back, he used to talk how about he was the CEO of Exxon.
- 51:32
- Unidentified speaker: He used to talk about how bad of a business refined you know, you go into his like his his quote.
- 51:38
- Unidentified speaker: So don't get married to it.
- 51:39
- Unidentified speaker: They're gonna work themselves out.
- 51:41
- Unidentified speaker: We're not gonna be stuck with a hundred dollar crack spreads indefinitely.
- 51:44
- Unidentified speaker: You're gonna see, I think you're gonna see oil prices stay sticking and go higher, you're gonna see demand destruction, and you're ultimately gonna see all these refiners point their capabilities at the the middle of the barrel um uh to solve the problem.
- 51:56
- Unidentified speaker: Um, so it's a temporary phenomenon.
- 51:58
- Unidentified speaker: Cool.
- 51:59
- Unidentified speaker: All right, great explanation.
- 52:00
- Unidentified speaker: All right, Warren.
- 52:01
- Unidentified speaker: Always great to have you on for guidance.
- 52:02
- Unidentified speaker: That was awesome.
- 52:03
- Unidentified speaker: Um, yeah, where can folks go if they want to see more of your research and and what you guys do?
- 52:07
- Unidentified speaker: Yeah, you can follow me at WarrenPies on X.
- 52:11
- Unidentified speaker: Um uh 314 Research is the number three, spell out 14research.com.
- 52:17
- Unidentified speaker: If you're an institutional investor, we we do institutional research.
- 52:20
- Unidentified speaker: If you're interested in our AI agent, we have a research agent, it's pretty cool.
- 52:25
- Unidentified speaker: And we're also developing, which I saw you guys are doing the same thing.
- 52:27
- Unidentified speaker: We're developing our own MCP.
- 52:29
- Unidentified speaker: I think this is gonna be the way people consume research going forward.
- 52:32
- Unidentified speaker: So we have our our Caliban AI research agent.
- 52:36
- Unidentified speaker: Um, that's open to anybody who wants to uh apply for that.
- 52:39
- Unidentified speaker: You can try it out, it's really neat.
- 52:42
- Unidentified speaker: Um, so that's Caliban, uh, not Taliban.
- 52:45
- Unidentified speaker: I've had people say Caliban.
- 52:48
- Unidentified speaker: I said Caliban.
- 52:49
- Unidentified speaker: I B A N.
- 52:51
- Unidentified speaker: So yeah, yeah, those are the ways to check us out.
- 52:54
- Unidentified speaker: Cool, man.
- 52:54
- Unidentified speaker: Well, Warren, appreciate you.
- 52:55
- Unidentified speaker: That was awesome.
- 52:56
- Unidentified speaker: Um yeah, great to have you as always.
- 52:58
- Unidentified speaker: Thank you for having me.
More from Forward Guidance
Long Gold And Short Semis Into Year-End? | Weekly Roundup
Episode discussion: Speakers advocate long gold and short semiconductors into year-end, citing technical breaks and debasement trade.
Speakers’ claims: The hosts analyze post-midterm market volatility, drawing parallels to late 2018 and arguing that the political 'blue sweep' was fully priced in. They identify a bottom in bonds from October and predict falling rates, while maintaining a bearish view on semiconductors due to a bubble 'coming undone.' Conversely, they recommend a long position in gold and gold miners, viewing the sector as a beneficiary of a 'debasement trade' under Democratic control. The episode also covers hedge fund blowups, crypto resilience, and specific entry strategies for home construction stocks.
The Fed Is A Sideshow While The AI Trade Implodes | Weekly Roundup
Episode discussion: Speakers argue AI CapEx cuts and energy shocks signal recession, rendering Fed policy secondary.
Speakers’ claims: The hosts argue that the Federal Reserve has become a 'sideshow' as the economy faces structural headwinds from an imploding AI trade and severe energy disparities. They cite Microsoft and Google cutting capital expenditure estimates as evidence that the AI-driven growth engine is stalling, potentially leading to a recession. The discussion highlights a liquidity crunch causing sovereigns to sell gold and criticizes European central banks for hiking rates despite an energy crisis. Additionally, the speakers warn of leverage risks in the AI sector and discuss geopolitical impacts on oil markets and crypto sentiment following a major exchange hack.