Luke Lango argues AI market faces transition period with lower CapEx growth, citing technical and fundamental headwinds
Summary
Luke Lango and Speaker_0 discuss a shift in the AI infrastructure market from a boom-bust cycle to a transition period characterized by constrained capital expenditure. Lango predicts hyperscaler CapEx growth for 2027 will fall short of Wall Street's 50% expectation, likely settling at 30-45%, driven by political, supply chain, and financial pressures. Historical semiconductor data suggests potential downside to mid-2027, with a median 29% drop following specific technical breakouts. The speakers emphasize that earnings per share (EPS) estimates must decline to signal a market bottom, viewing the current correction as a necessary reset to sustainable, cash-funded growth.
Episode sections
- 0:00 – 3:10
Intro and Thesis Adjustment
Luke Lango clarifies his long-term bullishness on AI while admitting short-term tactical caution regarding the market bottom.
- 3:10 – 9:58
Three Factors Impacting CapEx
Lango identifies Washington D.C., Wall Street pressure, and supply chain shortages as factors squeezing AI spending, predicting lower-than-expected CapEx growth.
- 9:58 – 13:23
Historical Semiconductor Analysis
Analysis of six historical SOX index rallies suggests potential downside of 29% over 34 weeks following specific technical breakouts.
- 13:23 – 17:21
Three-Stage Rebound Framework
Lango outlines technical stages for a market rebound, with Stage 3 requiring a break above October 2024 highs.
- 17:21 – 20:22
Fundamental Outlook and EPS Cuts
Lango projects lower hyperscaler spending and earnings, arguing that falling EPS estimates are a bullish signal that euphoria has ended.
- 20:22 – 22:34
NVIDIA and Market Catalysts
Lango dismisses short-term catalysts like NVIDIA earnings, viewing the real catalyst as hyperscaler CapEx guidance in early 2027.
- 22:34 – 27:33
Macroeconomic Factors
Discussion on the impact of geopolitical resolutions on treasury yields and rising credit spreads for tech companies.
- 27:33 – 32:05
Five Inputs for Hyperscaler Spending
Lango details five inputs for spending decisions: demand, ROI, regulations, Wall Street pressure, and supply chain.
- 32:05 – 34:03
Supply Constraints and Capital Spending Risks
Speaker_0 explains how supply shortages delay data center builds and identifies key variables to monitor.
- 34:03 – 35:16
Market Bottoming and Capital Preservation
Speaker_0 advises against buying dips prematurely, citing significant drawdowns in stocks like Nebius and Applied Opto.
- 35:16 – 36:20
Long-Term AI Bullishness and Political Risk
Speaker_0 asserts long-term bullishness, viewing political risk as a major headwind that could be mitigated by responsible practices.
- 36:20 – 39:29
The 'Steroid' Analogy for AI Growth
Speaker_0 compares the market transition to an athlete coming off steroids, suggesting a move to sustainable, cash-funded growth.
- 39:29 – 40:27
Cash-Flow Funding and Supply Normalization
Speaker_0 claims reduced spending can be cash-funded, eliminating reliance on debt markets and ensuring long-term durability.
- 40:27 – 44:19
Historical Comparisons and Shakeout Theory
Speaker_0 compares the current market to 1995-98, calling it a shakeout rather than a bust, and advises waiting for bottom signals.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| AlphabetGOOGL | equity | neutral | We're projecting for Alphabet, Amazon, Microsoft, and Meta to spend about five to ten percent less per year over the next several years than what the street is modeling for. |
| AmazonAMZN | equity | neutral | We're projecting for Alphabet, Amazon, Microsoft, and Meta to spend about five to ten percent less per year over the next several years than what the street is modeling for. |
| MicrosoftMSFT | equity | neutral | We're projecting for Alphabet, Amazon, Microsoft, and Meta to spend about five to ten percent less per year over the next several years than what the street is modeling for. |
| Meta PlatformsMETA | equity | neutral | We're projecting for Alphabet, Amazon, Microsoft, and Meta to spend about five to ten percent less per year over the next several years than what the street is modeling for. |
| NVIDIANVDA | equity | bullish | Okay, NVIDIA is gonna earn, I don't know, whatever their EPS is, they're gonna earn that plus 10% per year for the next 20 years. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Lango is 'more bullish, more positive, more optimistic than ever on the two to five-year five year outlook for AI stocks.'
Unidentified speaker (speaker_0) ·0:14
How to verify: Check subsequent podcast episodes or Luke Lango's public writings/posts from late 2026/2026 for consistent long-term AI thesis statements. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Primary sourceHyperscaler CapEx guidance for 2027 will be lower than Wall Street's expectation of 50% growth, likely settling at 30-40% or 40-45% growth versus 2026.
Unidentified speaker (UNKNOWN) ·4:05
How to verify: Wait for hyperscaler Q4 2026 earnings calls in Jan/Feb 2026 to compare their provided CapEx guidance against consensus estimates from Bloomberg/Refinitiv. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
The median additional drop for the SOX index after breaking below its 50-week moving average following a >70% rally above the 200-week moving average is 29%.
Unidentified speaker (speaker_0) ·10:47
How to verify: Download historical SOX data from Yahoo Finance or Bloomberg; identify the six instances (Dec 95, Oct 97, Sep 00, Oct 18, Jan 22, Sep 24) where the index rallied >70% above the 200-week MA and then broke the 50-week MA; calculate the subsequent drawdown. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
The median time for the SOX index to bottom after breaking the 50-week moving average in the six historical instances was 34 weeks.
Unidentified speaker (speaker_0) ·11:27
How to verify: Same dataset as above; calculate the number of weeks from the break of the 50-week MA to the subsequent market bottom for each of the six instances. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
SMH resistance level is at 504 and QQQ resistance level is at 681-682.
Unidentified speaker (speaker_0) ·14:43
How to verify: Check historical price charts for SMH and QQQ around the recording date (Nov 2026) to identify the 200-day moving average and recent swing highs (Leopold lows) matching these levels. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Micron trades at four times forward earnings because the market does not believe street estimates for 2028 and 2029.
Unidentified speaker (speaker_0) ·18:19
How to verify: Check Micron's current P/E ratio on financial data providers (Yahoo Finance, Bloomberg) and compare it to consensus EPS estimates for 2028 and 2029. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Hyperscalers (Alphabet, Amazon, Microsoft, Meta) are projected to spend 5-10% less per year on AI infrastructure than street models for the next several years.
Unidentified speaker (speaker_0) ·17:44
How to verify: Compare the actual CapEx guidance provided by these companies in their 2026/2026 earnings reports against the consensus estimates available prior to the guidance. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Primary sourceApplied Optoelectronics delayed a project due to a lack of optics components for a major hyperscaler customer (likely Amazon).
Unidentified speaker (speaker_0) ·31:43
How to verify: Search for press releases or SEC filings from Applied Optoelectronics (AEOI) in late 2026 regarding project delays and supply chain issues, specifically mentioning Amazon. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Primary sourceAmazon has a long-term agreement with Applied Opto but is holding back capital spending because Applied Opto lacks sufficient components, delaying data center construction.
Unidentified speaker (UNKNOWN) ·32:05
How to verify: Check Amazon's Q3/Q4 2026 earnings call transcripts for mentions of supply chain delays with optical component vendors and capital expenditure guidance adjustments. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Amazon's credit default swap spread is currently at 95 basis points, having been at 70 basis points a month prior.
Unidentified speaker (speaker_0) ·33:07
How to verify: Access Bloomberg or Refinitiv Eikon data for Amazon's CDS spreads for the relevant dates in late 2026. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Nebius is down 50% from its highs, and Applied Opto is down 80% from its highs.
Unidentified speaker (UNKNOWN) ·34:06
How to verify: Calculate the percentage decline from the 52-week high to the current price for Nebius and Applied Opto using financial data providers like Yahoo Finance or Bloomberg. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
The speaker is more bullish on the 2-5 year outlook for AI stocks than ever before.
Unidentified speaker (speaker_0) ·35:35
How to verify: This is a subjective sentiment claim; verify by reviewing the speaker's prior public statements or podcast episodes from 2026-2026 to compare stated outlooks. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Primary sourceThe speaker compares the current AI market conditions to 1995-96 and 1997-98, describing it as a shakeout rather than a dot-com bust.
Unidentified speaker (UNKNOWN) ·40:27
How to verify: This is a historical analogy claim; verify by comparing the P/E ratios, earnings growth rates, and market breadth of the current period (2026) to the specified historical periods using S&P 500 data. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Data suggests a potential 30% drawdown over the next six months.
Unidentified speaker (UNKNOWN) ·40:27
How to verify: The speaker references 'our charts' and 'our data'; verify by requesting the specific proprietary dataset or model from the podcast host, Luke Lango, or checking his published research on DYORfunds. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
If capital spending is reduced to current levels, it can be cash-funded, eliminating reliance on debt or equity markets.
Unidentified speaker (speaker_0) ·39:29
How to verify: Analyze the free cash flow of major AI infrastructure companies (e.g., Amazon, Microsoft, Google) in their 2026 financial statements to determine if their projected capex can be covered by operating cash flow. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
NVIDIA will earn its EPS plus 10% per year for the next 20 years.
Unidentified speaker (speaker_0) ·43:09
How to verify: This is a forward-looking forecast; verify by comparing the stated 10% CAGR to long-term consensus analyst estimates for NVIDIA's EPS growth from major investment banks. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Primary sourceMicron is trading at four times forward earnings.
Unidentified speaker (speaker_0) ·43:28
How to verify: Calculate Micron's current price-to-forward-earnings ratio using the consensus analyst EPS estimates for the next 12 months and the current stock price. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Anti-AI populist backlash could influence the 2028, 2030, and 2032 elections, potentially leading to harsh crackdowns.
Unidentified speaker (UNKNOWN) ·38:46
How to verify: Monitor political polling data, legislative proposals regarding AI regulation, and campaign platforms of candidates running in the 2026 and 2026 US elections. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.
Do your own research
- Verify hyperscaler CapEx guidance for 2027 against current Wall Street consensus estimates in equity research reports.
- Check the SOX index price relative to its 200-week and 50-week moving averages to assess technical status.
- Monitor credit default swap (CDS) spreads for major tech companies like Amazon, Alphabet, and Oracle for signs of financing stress.
- Review Micron's forward P/E ratio and compare it to consensus EPS estimates for 2028-2029 to validate valuation assumptions.
- Analyze historical SOX index performance during similar technical breakouts to assess downside risk probabilities.
- Track political developments and regulatory pressures in key states like Texas that may impact AI data center approvals.
- Assess supply chain status for critical components like optical interconnects to evaluate potential delays in data center construction.
- Read earnings call transcripts for NVIDIA and other major chipmakers to gauge management sentiment on demand and supply constraints.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): And that's why I just want to be abundantly clear. I'm not like coming up here as an AI bull and saying, like, oh my God, I'm freaked out now, sell everything. Like that's not at all what I'm saying.
- 0:10
- Unidentified speaker (speaker_0): I am actually
- 0:13
- Unidentified speaker (speaker_0): more
- 0:14
- Unidentified speaker (speaker_0): bullish, more positive, more optimistic than ever on the two
- 0:21
- Unidentified speaker (speaker_0): to five-year five year outlook for AI stocks.
- 0:25
- Unidentified speaker (UNKNOWN): Hello,
- 0:25
- Unidentified speaker (speaker_1): and welcome to being exponential. So the AI trade has taken a beating, and after months of selling pressure, investors are asking one question: When is it finally time to start buying again? So today, we're breaking down Luke's three-stage roadmap for getting back on the offensive, the key technical levels that could signal a market reversal, and whether the next wave of AI spending could
- 0:45
- Unidentified speaker (speaker_1): set us up for a major rally.
- 0:47
- Unidentified speaker (UNKNOWN): All
- 0:47
- Unidentified speaker (speaker_1): Alright, right, Luke, let's start with the big picture. Are we approaching the bottom of the something? sell-off or is there is there more to go?
- 0:54
- Unidentified speaker (speaker_0): Um, I I think there's more to go. And I think I have to level set and just say
- 1:01
- Unidentified speaker (speaker_0): we were wrong. Right? Like we thought that
- 1:06
- Unidentified speaker (UNKNOWN): the
- 1:06
- Unidentified speaker (speaker_0): big question I've been telling everybody to ask themselves is
- 1:11
- Unidentified speaker (speaker_0): well, the amount of money that
- 1:14
- Unidentified speaker (UNKNOWN): the
- 1:14
- Unidentified speaker (speaker_0): hype. hyperscalers spend on
- 1:18
- Unidentified speaker (speaker_0): expanding AI infrastructure, well that number go up
- 1:26
- Unidentified speaker (UNKNOWN): or
- 1:26
- Unidentified speaker (speaker_0): go down. If it goes up,
- 1:29
- Unidentified speaker (UNKNOWN): you
- 1:29
- Unidentified speaker (speaker_0): buy the dip in AI stocks. If it goes down,
- 1:32
- Unidentified speaker (UNKNOWN): you
- 1:32
- Unidentified speaker (speaker_0): don't buy the dip in AI stocks.
- 1:35
- Unidentified speaker (speaker_0): That is still the question that that is still the North Star question.
- 1:40
- Unidentified speaker (speaker_0): And for months, because the AI trade really has been stuck in neutral since May. So for months, I'd ask that question, and I would always come back to the demand signals are really, really strong. You have accelerating cloud growth at the hyperscalers. You have really good ARR growth numbers
- 2:00
- Unidentified speaker (speaker_0): out of the frontier labs. You have reports. of massive productivity increases in corporate America. You have massive profit margin growth across the entire tech ecosystem. So the demand signals were always there. Muse gets to 10 million users. It's in all these things happen, right?
- 2:17
- Unidentified speaker (UNKNOWN): The
- 2:17
- Unidentified speaker (speaker_0): coding, AI coding tools go crazy. And so for months, I asked that question and I kept going back to the demand signals and said demand is not slowing. therefore, the amount of money these guys spend is not going to slow. And that is still true. Like even after last week's brutal sell
- 2:37
- Unidentified speaker (speaker_0): off, that is still true. Demand is not slowing. Everybody and their best friend in the AI supply chain is reporting earnings, and nobody is reporting slowing demand. Nobody. But what I got wrong, and what I missed, and what I will own here and now, is that
- 2:57
- Unidentified speaker (speaker_0): demand is the most important factor in determining capital spending plans, but not the only factor in determining capital spending plans. And we completely
- 3:10
- Unidentified speaker (speaker_0): underestimated the other factors that came. can impact capital spending plans. And those three factors, those three additional factors, are Washington, D.C., Wall Street itself, and supply chain shortages.
- 3:28
- Unidentified speaker (speaker_0): The AI trade has been squeezed by all three factors. And because of developments in the last week, I believe that all three of those factors are now. hitting hard enough to offset still robust demand and result in lower than expected capital
- 3:48
- Unidentified speaker (speaker_0): spending plans out of the hyperscalers over the next several years. That doesn't mean they stop spending. But what it does mean is that they do not spend as much as Wall Street had hoped for. 2027 numbers,
- 4:05
- Unidentified speaker (UNKNOWN): I
- 4:05
- Unidentified speaker (speaker_0): think, are gonna be a when they all issue their guidance in January or February of next year. That's when they're gonna give their 2027 CapEx guidance. I think their CapEx guides are going to be lower than what Wall Street expects. I think they're still gonna be up year over year from 2026. In fact, they definitely will be up year over year
- 4:25
- Unidentified speaker (speaker_0): from 2026. But right now, Wall Street Street's expecting like 50% growth in 2027 capex versus 2026 capex. It probably comes in more like 30 to 40% or 40 to 45%, not 50%. So you're gonna get a downward cut there. And then you got to cut 28, 29, and 30 because those are based off of
- 4:45
- Unidentified speaker (speaker_0): 27. And so I think that the capital spending, the actual capital spending plans of the hyperscale hyperscalers are going to come in below plan, uh below street estimates for the next three to four years. And what that means is the amount of money flowing into the funnel is still going to remain robust, but it's going to be less than what Wall Street had planned for, which means
- 5:05
- Unidentified speaker (speaker_0): EPS estimates all throughout the supply chain have to move lower as well. And that is the pain that you're seeing right
- 5:11
- Unidentified speaker (UNKNOWN): now. I
- 5:12
- Unidentified speaker (speaker_0): think this is just a transition period because I think on the other side of this, we actually get to a more durable environment, right? I mean, where we were the track we were heading on was an unsustainable one. We've always framed it as like boom bus, right? Even though we've been AI bold, we've always framed it as this boom will eventually end in a bus. That's just how it
- 5:32
- Unidentified speaker (speaker_0): is, right? We're going really fast, we're spending a lot of money, we're going into debt markets, we're issuing equity. There's just a lot. of things going on to where you know we're going so fast that eventually we are going to drive ourselves off the cliff.
- 5:44
- Unidentified speaker (UNKNOWN): But
- 5:44
- Unidentified speaker (speaker_0): instead, what we're doing is we're kind of pulling back and saying, let's slow her down. And by doing that, we are actually improving the long-term durability of the CapEx cycle. Right? If all of a sudden capital expenditures do not grow beyond a trillion dollars per year on AI. stuff. So before it was
- 6:04
- Unidentified speaker (speaker_0): supposed to be like, you know, 1 trillion, 1.2, 1.3, 1.5, PWC was at 1.8 per year by the 2040s, right? Let's say that doesn't happen. And let's say we just stabilize at 1 trillion. Well, guess what? At 1 trillion, the hyperscalers have enough cash to fund it. They don't have to go to the debt markets anymore. They don't have to issue equity anymore. So you get rid of the dilution risk, you get rid of the leverage
- 6:24
- Unidentified speaker (speaker_0): risk, you get rid of the debt blow up risks, and all of a sudden that risk becomes that that gets sidelined if we slow down the building and we start, you know, these companies start working with legislators in Texas and legislators in South Carolina and legislators in Georgia and Ohio to, let's say, more responsibly build these things, or at least give the optics a
- 6:44
- Unidentified speaker (speaker_0): more responsible build, then you satisfy this anti-AI populist backlash and you sideline that risk. All of a sudden, maybe in 28, AI is not on the ballot. AI was on the ballot in 26, but depending on how we act over these next two years, maybe AI is not a hot topic by 28, and all of a sudden that's the election that really
- 7:04
- Unidentified speaker (speaker_0): matters, and AI is off the ballot. So I actually think what we're doing right now is ensuring more durability in the long term.
- 7:13
- Unidentified speaker (UNKNOWN): And
- 7:13
- Unidentified speaker (speaker_0): that's why I think AI is going to. through what we're calling a transition period, not an ending, not a burst, not a uh, you know, the dot-com bus scenario. We're going through a transition period on the other side of which will be a much healthier, more durable boom. And this transition period will offer us a fantastic buying
- 7:33
- Unidentified speaker (speaker_0): opportunity. But having said that, the trade is a mess right now. The fundamentals are confused. confusing. There's a lot of uncertainty out there. And key technical levels are breaking everywhere. With all that in mind, I just think that it's too early to back the truck up
- 7:53
- Unidentified speaker (speaker_0): and say, let's go buy all these data center stocks down 50 50-60% 60% with both hands, lever up and and hope that you there's know a rebound coming. soon. There is a rebound, but I'm not sure how soon it happens. And I think that we probably go lower before that big rebound does arrive. So that's me
- 8:13
- Unidentified speaker (speaker_0): sort of just level setting and basically saying
- 8:17
- Unidentified speaker (speaker_0): we got it wrong. But it's okay in markets to be wrong. It's not okay to stay wrong. And that's why we're adjusting to the f the the new facts and evidence.
- 8:29
- Unidentified speaker (UNKNOWN): Yeah,
- 8:29
- Unidentified speaker (speaker_2): of course.
- 8:30
- Unidentified speaker (speaker_1): I mean the the frequency of events that occurred over you know the past two weeks, it
- 8:36
- Unidentified speaker (speaker_1): it's unsustainable, uh you know, to say the least. Um but uh just on that on that topic, so what what are your angles? What are you looking for? What is the cue to get back in, to start thinking about by buying into the rebound? Is there a specific vantage point that you're you're on the lookout, a
- 8:56
- Unidentified speaker (speaker_1): specific technical level?
- 8:58
- Unidentified speaker (UNKNOWN): Uh
- 8:59
- Unidentified speaker (speaker_0): okay, so if you want to talk technicals, I mean I I sent over a graphic to you that we can look at. It's that the um
- 9:08
- Unidentified speaker (speaker_0): the the semis could drop before they pop graphic. What we really looked at or we're looking at is is soft. SOX, the semiconductor index. We've talked about it a million times on this podcast. And what we have with SOX is a very unique scenario where semiconductor stocks had this massive euphoric rally
- 9:28
- Unidentified speaker (speaker_0): where they got super stretched above their long-term moving averages. And then last week they just plunged below them without any concern for that technology. technical support level. If this were a healthy market, buyers would have stepped in at those technical support levels. They didn't. The big one that we analyzed is the
- 9:47
- Unidentified speaker (speaker_0): 50-week moving average, which roughly approximates to a 250-day moving average or like a one-year moving average. So we have our 20-week or 50-week moving average. We studied all times in history. Stocks SOX started back in the early 1990s, all times in recorded history that SOX
- 10:08
- Unidentified speaker (speaker_0): popped like crazy and then dropped below its 50-week moving average. That happened six times before: December of 95, October 97, September 2000, October 2018, January 2022, and September of 2024. Those are the six
- 10:28
- Unidentified speaker (speaker_0): previous instances where SOX rallied at least 70% or more above its 200-week moving average, and then dropped sharply below its 50-week moving average. Exact scenario we have today. And you know what happened after all six of those instances? We
- 10:47
- Unidentified speaker (speaker_0): dropped more. Looking at the data in December 95, we dropped another 24%. In October 97, we dropped another 34%. In September of 2000, the dot-com bus, we dropped another 59%. October 2018, we dropped another 16%. January 2022, we dropped another 35%. September 24, we
- 11:07
- Unidentified speaker (speaker_0): dropped another 21%. So the median across those six historical instances for the exact price dynamic we have today is a drop of 29% more, not 29% peak to trough, 29% more after the 50-week break in SOX. And it took a median of
- 11:27
- Unidentified speaker (speaker_0): 34 weeks for the index to bottom, which approximates to about six months. So that puts us out to July. So what the data tells us is the price dynamic we have right now. is massive rally in stocks, big break below a major support level. That has happened six times
- 11:46
- Unidentified speaker (speaker_0): before. And the historical behavior is another roughly 30% drop over the next six months, which means if you kind of apply that to today's standards, a 30% drop from here into the summer of 2027. History does not repeat, it rhymes.
- 12:03
- Unidentified speaker (speaker_0): So don't take that as like, you know, that's the that's the Bible. We're gonna etch that in stone and we're gonna follow that playbook. But the this is the technical information that has me worried. This is what happened last week, technically, that really got me worried. I always thought we would hold these big levels, we would hold the Leopold Pold lows, we would hold the 200-day moving
- 12:23
- Unidentified speaker (speaker_0): averages, we would hold the 50-week. 50 week. We didn't hold any of that stuff last week. SMH lost it, SOX lost it, AIS lost it, Triple Q's lost it, AIQ lost it. Everything lost all the levels last week. And so that's worrisome from a technical perspective. That's pretty
- 12:43
- Unidentified speaker (speaker_0): worrisome. And even with Monday's rebat, rebound, or we got that little tepid rally on Friday, and then we had a bigger, much more powerful. rebound rally on Monday, as of this recording of NASDAQ's up 2.7% today. Even with those rallies, we are still below all these levels. We are still below the 50-week. We are still below the 200 day.
- 13:03
- Unidentified speaker (speaker_0): We are still below the Leopold lows on pretty much everything. Uh, you know, QQQ, SMH, SOX, Socks, AIS, on all those core AI benchmarks. We're still below all those levels. So none of the technical damage. has been repaired. That's what has me worried. When does it get repaired? Well, that's what I sent you was this, you know, we went defensive. What
- 13:23
- Unidentified speaker (speaker_0): gets us back on offense graphic? You throw that up for the for the viewers right now. For us, we're kind of segmenting this rebound into three stages. The stage one bounce is what we're seeing right now. Of course, you plunge a crap ton. you're gonna rebound a little. That just is what happens. That happened in 2000
- 13:43
- Unidentified speaker (speaker_0): so many different times. It doesn't mean the rebound is here. It also doesn't mean it's a dead cat bounce. It's just a bounce. And whether or not it's a dead cat or it's actually the beginning of a more prolonged, stronger, big time rebound rally is what happens next, right? That determines it. So stage one, we kind of have this dead cat cap bounce during this time. or maybe
- 14:03
- Unidentified speaker (speaker_0): dead cat bounce, you don't really know what it is. Um, it's an oversold bounce. During this time, you don't you don't really want to buy, you don't want to get aggressive, you want to kind of stay on, you don't want to panic sell anything, you you don't want to buy anything, you just kind of want to stay on hold. You want to be like, hey, I have my stocks, I'm just waiting, I'm waiting to see what happens. Then you get to the decision point,
- 14:23
- Unidentified speaker (speaker_0): which is whether or not this stage one bounce turns into a stage two. constructive rebound rally or gets faded at critical techno levels and gets confirmed as a dead cat bounce. Those levels to us are going to be more or less the 200-day moving averages on a lot of these uh AI benchmarks and the Leopold
- 14:43
- Unidentified speaker (speaker_0): lows on a lot of these AI benchmarks. So That's SMH at 504. That's triple Q's at uh 681, 680. 682. And with this bounce that we are seeing right now in the AI complex, we are rallying up towards those levels. We are starting to creep up to 680 on QQQ. We're starting to
- 15:02
- Unidentified speaker (speaker_0): creep up to uh 504 on SMH. We're starting to creep up to the Leopold Lowe's on SOX Thoughts on AIS. We're starting to get up to those levels. So pretty soon we're gonna going get to to that decision. point of okay, we're at the big first resistance level. Are we going to retake that?
- 15:20
- Unidentified speaker (UNKNOWN): And
- 15:21
- Unidentified speaker (speaker_0): then are we going to hold that or are we going to get faded at those levels? And if we get faded at those levels, that probably means that this precedent that we have, 30% drop, another 30% drop over the next six months, that looks more likely. So you want to sell into that. fade. If we break above
- 15:41
- Unidentified speaker (speaker_0): these major resistance levels and we start to really now we retake the Leo Pold lows, they turn into support, we retake the 200 moving averages, they turn into support. This starts to look like a big time rebound rally. You want to start to get constructive there and you want to start buying back into these names. That's that's how I feel about it. And then you kind of have stage three, which is this
- 16:00
- Unidentified speaker (speaker_0): broad confirmation that the AI trade is in. indeed back. That's the AI trade resuming. That happens when you have all of the benchmarks above their 200 day, turning it into support. Socks gets above the number is 11684. That's the October 28th high. So that would end the trend of lower
- 16:20
- Unidentified speaker (speaker_0): highs and lower lows. That's like the full resumption point. So that's how I'm viewing it technically. and how we're playing it technically, which means right now we're in technical no man's land. And I'm not saying buy the dip, and I'm not saying run for the hills. What I'm saying is all right, sit on the sidelines, let the market
- 16:40
- Unidentified speaker (speaker_0): digest what's going on, and then respond to we're gonna get to some pretty important technical levels pretty soon, respond to how the market reacts to those technical levels.
- 16:50
- Unidentified speaker (UNKNOWN): Gotcha.
- 16:51
- Unidentified speaker (speaker_1): Yeah. So today I mean, today's market, uh, you know, relative to especially the two thousands moves a lot more. Uh it swings a lot more than it used to. We see ten percent dips. We just saw the biggest loss in capital in the market in one day. Um now, do you think there are any events that could uh
- 17:11
- Unidentified speaker (speaker_1): escalate the the bounce back? So we have anthropics IPO coming up. Is there anything potentially that could uh you know put some some rocket fuel into the rebound?
- 17:20
- Unidentified speaker (UNKNOWN): Yeah,
- 17:21
- Unidentified speaker (speaker_0): so now now we're kind of talking about the fundamentals of it. So uh
- 17:24
- Unidentified speaker (UNKNOWN): for
- 17:24
- Unidentified speaker (speaker_0): the viewers, throw up the the AI capex and chip earnings outlook uh graphic. This is kind of talks about the fundamentals of what's going on. We we kind of talked about this a little bit earlier. The amount of money coming into the funnel, hyperscaler spending, I think. is going to be lower than planned. We're projecting for Alphabet, Amazon, Microsoft, and Meta
- 17:44
- Unidentified speaker (speaker_0): to spend about five to ten percent less per year over the next several years than what the street is modeling for. And because of leverage and supply dynamics, that flows through into
- 17:58
- Unidentified speaker (UNKNOWN): um
- 17:59
- Unidentified speaker (speaker_0): 15% pretty much uh lower. earnings across the sector, the semiconductor sector, in 2028, 2029, and in 2030, roughly 15% below plan. So these are the cuts that are coming. And these are the cuts that the market is now priced for. This is why Micron trades at four times forward earnings because nobody
- 18:19
- Unidentified speaker (speaker_0): believes those earnings. If the market believed that Micron was going to earn what the street thinks is it's going to earn in 2028 and 2029, then it wouldn't trade at four times earnings. It's trading at four times forward earnings because no one believes the forward earnings. The market is priced for these cuts.
- 18:35
- Unidentified speaker (UNKNOWN): So
- 18:35
- Unidentified speaker (speaker_0): what I'm watching is when they actually start to freaking cut. Because the market always moves before the fundamentals. It's why you cannot wait for the fundamentals. That's why we talk so much about technicals here. Because the market is always sniffing stuff out. before the fundamentals move. You know, I saw this chart, it was like how much
- 18:55
- Unidentified speaker (speaker_0): a stock price falls before its EPS estimates fall. Right. And it was like during the dot-com bust, it was like stocks were already down 60 to 70 percent, a lot of stocks before their EPS estimates even started to fall a little. And that's what we have right now. We have a bunch of AI supply chain stocks that are 30, 40, 50, 60% or more off. their highs,
- 19:15
- Unidentified speaker (speaker_0): and their EPS estimates are still rising. Those estimates are going to start falling. And when they do start falling, I think that's actually a bullish indicator because now the buy side is going to be like,
- 19:29
- Unidentified speaker (UNKNOWN): oh,
- 19:29
- Unidentified speaker (speaker_0): finally, sell side's not just euphoric hyping everything up. And now we can kind of get back to you know a reality situation. EPS estimates start falling, stocks fall. find a floor, and then they rebound. So I'm actually looking for weirdly enough, as a bull, I'm looking for EPS estimates to fall. And so I want to see that happen.
- 19:49
- Unidentified speaker (UNKNOWN): Um the
- 19:50
- Unidentified speaker (speaker_0): anthropic rogue show is starting. We're gonna get the IPO. I think that could provide a little bit of a lift, but there's also a risk that people are just really scared at that point, and it doesn't provide a lift at all. So I'm not like counting on the anthropic IPO to save anything here. I love the company long term, but in the short term, I mean they're raising in November, which is a
- 20:10
- Unidentified speaker (speaker_0): sleepy time for markets, anyways. Uh, there's gonna be a lot of questions about that timing.
- 20:16
- Unidentified speaker (UNKNOWN): Uh
- 20:17
- Unidentified speaker (speaker_0): I'm not counting on that to like bring us home to the end of the year.
- 20:21
- Unidentified speaker (UNKNOWN): Uh,
- 20:22
- Unidentified speaker (speaker_0): you have NVIDIA's earnings report, which is gonna be really weird now. with the whole sex scandal stuff.
- 20:30
- Unidentified speaker (UNKNOWN): Um,
- 20:30
- Unidentified speaker (speaker_0): so I don't
- 20:31
- Unidentified speaker (UNKNOWN): uh
- 20:32
- Unidentified speaker (speaker_0): I'm not counting on that to save the day either. I think a lot of people maybe were counting on that to save the day. I don't think it can save the day anymore. I as it relates to the Jensen saga, I think it's an NVIDIA problem. I don't think it's an AI problem. I if anything, I think it actually creates tailwinds for the AI complex over the next several uh quarters and and years, even because.
- 20:51
- Unidentified speaker (UNKNOWN): um
- 20:52
- Unidentified speaker (UNKNOWN): it
- 20:53
- Unidentified speaker (speaker_0): doesn't impact demand but it impacts people's view of NVIDIA and so maybe Microsoft accelerates their own chipbuilding efforts, maybe Meta goes to AMD, maybe Amazon builds more training, Alpha builds more TPU. And so what you get is you get a demonopolization of the industry. And when you demonopolize, you
- 21:13
- Unidentified speaker (speaker_0): broaden out demand. And so the demand that was just kind of all flowing into NVIDIA maybe flows outward to other suppliers. Because if you go to Marvell or you go to Broadcom or you go to AMD, there's a whole different list of, you know, that's a tailwind for the networking companies and the optic companies and all the other companies out there because NVIDIA does a lot of stuff in house. So I think it's actually, you know,
- 21:33
- Unidentified speaker (speaker_0): if anything a tailwind for the AI trade and a headwind for NVIDIA. But I also think it means that the NVIDIA report in November is not gonna like save us. Uh I don't really see our savior, if you will, coming until late January, early February with the hyper the next time the hyperscalers report. Because the only thing that matters is how much money
- 21:52
- Unidentified speaker (speaker_0): they're spending.
- 21:54
- Unidentified speaker (UNKNOWN): Um
- 21:54
- Unidentified speaker (speaker_0): and I think that when they come in in late January and early February, it's gonna be bad news equals good reaction. They're gonna issue lower than expected CapEx numbers for 2027, and we're gonna rally off that because we sold off into it. That's what I think is going to happen. And so I kind of see that's why that three month window to me is kind of the window I'm looking at. Uh we don't get another report from
- 22:14
- Unidentified speaker (speaker_0): those guys from for three months. I think we we probably have a lot of uncertainty and panic selling into that, and then we rally like hell out of it. So that's how I'm looking at kind of the the fundamental outlook. for for what's going on. Another piece of the equation here, which is I think a big part of the rally today, is the Iran war. Um,
- 22:34
- Unidentified speaker (speaker_0): you know, Wall Street is one of the big pressure forces impacting the AI spending trade. And that can
- 22:44
- Unidentified speaker (speaker_0): improve meaningfully with actual Iran war resolution, because long long-term term treasury yields will go lower. the Fed won't have to hike interest rates. And so the debt side of the the pressure here uh gets meaningfully relieved. So I think that could help us in in the short term for sure. But that's been like a you know a
- 23:03
- Unidentified speaker (speaker_0): wild card situation for a long time, and I'm not hanging my hat on it uh right now. But um that that is something to to definitely watch. Um
- 23:13
- Unidentified speaker (speaker_0): credit spreads
- 23:14
- Unidentified speaker (UNKNOWN): are
- 23:14
- Unidentified speaker (speaker_0): something to watch. Those are not getting better, even despite today's rally. A lot of credit spreads, Alphabet's credit spreads are going up uh today. So I think you gotta Oracle's is is going up today. Uh so I think you gotta be aware of that as well. So again, I I think you have to watch Washington, you have to watch Wall Street, you have to watch uh supply chain shortages.
- 23:33
- Unidentified speaker (UNKNOWN): Yeah,
- 23:33
- Unidentified speaker (speaker_2): definitely. So uh
- 23:34
- Unidentified speaker (speaker_1): two bearish questions on uh two of those points. So you mentioned it was an NVIDIA play. problem, but I think we can almost consider NVIDIA like the central bank of AI financing. Wouldn't that make it a a headwind for the broader IA market if NVIDIA gets severely impacted by the the honeypot situation? And then
- 23:54
- Unidentified speaker (UNKNOWN): I don't
- 23:54
- Unidentified speaker (speaker_0): think it gets severely impacted. Uh yes. If NVIDIA were to get severely impacted, then yes, because they are a financier for a lot of these projects. that would really hurt the AI trade. Yeah, I don't think the impact's gonna be um severe. I don't think Jensen's gonna
- 24:09
- Unidentified speaker (UNKNOWN): get in
- 24:10
- Unidentified speaker (speaker_0): trouble, if you will, like you know, legal trouble or governance trouble. I think he's gonna remain CEO. I don't think he I mean I I don't think he shared any secrets. I think he just he I mean he did an immoral thing. Um
- 24:24
- Unidentified speaker (speaker_0): and that sucks.
- 24:27
- Unidentified speaker (speaker_0): You feel terrible for the family,
- 24:30
- Unidentified speaker (speaker_0): but financial impact I think remains pretty small. So
- 24:36
- Unidentified speaker (UNKNOWN): no,
- 24:36
- Unidentified speaker (speaker_0): I'm I'm not really concerned about the NVIDIA situation blowing up into like this severe crisis, identity crisis for for NVIDIA.
- 24:45
- Unidentified speaker (speaker_1): And then um in in terms of the rate, rates. so we just had a a surprisingly good jobs report. So there is a concern of you know rates being hiked in that regard. Do you think that is also a potential issue that we we're seeing down the pike?
- 24:59
- Unidentified speaker (speaker_0): Um no. No. I think the thing that you have to watch for are the credit spreads. Uh rates matter, but the reason rates matter is for the cost of debt of these companies so because they're tapping tapping debt markets, uh these companies being the hyperscalers. and the the other neo clouds and stuff like that. And that
- 25:19
- Unidentified speaker (speaker_0): number, the cost of debt for them is impacted by treasury rates plus their spreads. And their spreads are kind of the real driving force of of the change in those numbers. And so I think yes some relief in rates would help. I don't think rates are going higher. The jobs report is is is good, but it's not good enough to be like this is this economy economy's is running running
- 25:39
- Unidentified speaker (speaker_0): hot. Uh I do think that, you know, if AI spending slows or gets a little bit more responsible, let's call it, then you're going to have some inflation relief and the Fed's not going to hike into that. So I don't think rates are going higher. I think rates are going lower. And I think that that's going to be beneficial for for um
- 25:59
- Unidentified speaker (UNKNOWN): for
- 25:59
- Unidentified speaker (speaker_0): AI stocks in the long run. But again, in the short term, we're in this transition. period where
- 26:04
- Unidentified speaker (UNKNOWN): we
- 26:04
- Unidentified speaker (speaker_0): just got to figure out how much spending is in the pipeline. Because everyone wants to talk about backlogs and everyone wants to talk about purchase agreements and long-term this and long-term that. None of that matters. Like those were words all used in 99, too. Like go through the I wasn't, you know, I was four years old in 99. ninety nine. So I I wasn't actively doing it myself. But I I am a student
- 26:24
- Unidentified speaker (speaker_0): of history and I've been through the earnings cross. calls of the telco companies and all those companies in ninety nine and two thousand. And guess what? It was all the same damn buzzwords. Our backlogs are huge. We got long term agreements and all this crap and price fixes and all this stuff. And you know what? None of it saved them because when the spending slows, none of those contracts hold up. So I
- 26:44
- Unidentified speaker (speaker_0): just think that all that is a bunch of bull crap. And what you have to really focus on is the spending, the actual spending of the hypers hyperscalers, which is why I think we're just kind of in no man's land until we get a big update from them in in January, February.
- 26:58
- Unidentified speaker (speaker_1): Yeah, definitely. So that that is the essentially the key to the all clear signal is the hyperscaler spending. There's is there any other, you know, factor that you're on the lookout out for that could uh you know be a potential boon or an additional boon to uh
- 27:13
- Unidentified speaker (UNKNOWN): the spending of the future?
- 27:16
- Unidentified speaker (speaker_0): look at the inputs to that right the hyperscaler spending is itself an output that is a decision these companies and boards are making based on a variety of inputs and those inputs are going to be uh demand so th I think there's four inputs for their decision to or
- 27:33
- Unidentified speaker (UNKNOWN): let's
- 27:33
- Unidentified speaker (speaker_0): call it five. There are five inputs to their decision to continue spending or what level they spend. Number one is demand.
- 27:42
- Unidentified speaker (UNKNOWN): There
- 27:42
- Unidentified speaker (speaker_0): is no problem with demand. We've been over this. Every demand signal out there for AI is very robust and very strong and likely to grow because of agentic AI, because of the coding tools, because of physical AI, et cetera, et cetera. The demand signal is there. Check. Green check. Number two is going to be ROI. So not just is the
- 28:03
- Unidentified speaker (speaker_0): demand there, are you turning it into revenues, but are you turning it into profits? Are you getting ROI on the spending? That's like a half check right now. Because for a lot of these hyperscalers, we talked about this, that their cloud businesses are absolutely on fire, but their core businesses are still adjusting to new AI paradigms. Google search
- 28:23
- Unidentified speaker (speaker_0): is adjusting to AI modes and is unsure exactly how to monitor. monetize it most effectively. And so they're not really seeing a huge return on the spending there. Meta is still trying to figure out how to monetize Muse and get a return on spending there. Also, how to create their neo cloud
- 28:42
- Unidentified speaker (speaker_0): business and get a return on spending there. Amazon is still trying to figure out how to maximize the efficiency of its warehouses and logistics with all. automation and get a return on there, not to mention their ad business as well. So a lot of these companies, that that second one, the ROI factor, is
- 29:01
- Unidentified speaker (speaker_0): a half check, let's call
- 29:03
- Unidentified speaker (UNKNOWN): it.
- 29:03
- Unidentified speaker (speaker_0): The other three factors are the ones where we're not getting any checks. The third factor is Washington. What are the policies and the regulations around building? data centers? They're about to get a lot more strict. We know that. They're not going to get strict enough to like just completely throttle the industry. Democrats would be absolutely
- 29:24
- Unidentified speaker (speaker_0): silly, stupid, dumb to do that, because then that would choke the economy going into the mid going into 2028, and they would just shoot themselves in that election. So I don't I don't think they're going to do that, but you are going to get more restrictions because they have to simultaneously keep the build-out happening to keep the economy turning,
- 29:44
- Unidentified speaker (speaker_0): yet also appease this anti-AI populist backlash, which for a lot of them is the reason why they got married. I think the girl down in Texas won because, in large part, of her anti-AI policy. So she's gotta she's gotta come through on that. So they gotta balance this line. And what that means is costlier and slower. builds going
- 30:04
- Unidentified speaker (speaker_0): forward. If it costs more to build one data center and it takes longer to build one data center, then that means my spending is going to get stretched out from 2027 to 28, 29, 30, and beyond. So that's a negative impact on spending plans. And that's a big change in the last week. Not that Democrats won the midterms, but that
- 30:23
- Unidentified speaker (speaker_0): Democrats won governorships in certain states. and also commandingly won Senate and House seats in certain states that are very important for the data center build out. So that's one of the things that changed in the last week. The fourth factor is gonna be Wall Street pressure. Because these guys are financing now
- 30:43
- Unidentified speaker (speaker_0): with equity and debt, they are reliant on the market. Mr. Market now has a seat at the table. It was, you know, before when it was just cash funded. it was just boards deciding to spend money. But now that they're relying on external capital from the market itself, Mr. Market has a seat at the table in deciding how much uh spending happens. And Mr. Market is almost
- 31:03
- Unidentified speaker (speaker_0): pressuring them to stop spending by blowing out their credit spreads on debt and by driving their stocks lower on equity, right? Their cost of equity goes up if their stock keeps going down and their cost of debt goes up. if their credit spreads keep blowing out. So the Wall Street pressure on slowing the build out has intensified dramatically. That's picked up
- 31:23
- Unidentified speaker (speaker_0): in the last week. So that's another factor that's changed. No check mark there. And number five is going to be supply chain shortages. If I'm deciding what to spend, if I don't have enough supply of X, Y, or Z to build 50 data centers in 2027, then I'm only gonna spend what the supply can guarantee me, right? And
- 31:43
- Unidentified speaker (speaker_0): so everything is supply constrained, and capital gets supply constrained there too. We saw Applied Optoelectronics, one of the optics names we talked about on this podcast a lot. They had to delay a project because there wasn't enough, um, they didn't have enough optics components to fulfill for a major hyperscaler customer. So if let's say that hyperscaler hyperscalar customer is Amazon, and
- 32:03
- Unidentified speaker (speaker_0): I think it is Amazon,
- 32:05
- Unidentified speaker (UNKNOWN): then
- 32:05
- Unidentified speaker (speaker_0): Amazon is like, hey, applied opto, we want to, we want to build this data center with these optics. And they're like, we don't have enough components for it, but I'm in a long term agreement with applied opto. I'm like, damn, well I can't build it now. I guess I gotta wait. So now I'm holding back the money until they get their supply and then I can go and build the a data center when they get their supply. So supply constraints can
- 32:25
- Unidentified speaker (speaker_0): also negatively impact or delay or stretch out capital spending plans. So those are the five factors that are driving capital spending plans themselves. And those are the five things you have to continue to monitor over the next several weeks and months. Are we getting deals with Washington? Are we seeing Core Weave and Nebius
- 32:45
- Unidentified speaker (speaker_0): and In SpaceX and X Nebius Nebbius doesn't operate in Texas, but the ones that do operate in Texas?
- 32:51
- Unidentified speaker (speaker_0): are you negotiating with the new governor? Are are there deals happening? You gotta watch that. Posit development there will really help. Wall Street pressure, are you seeing that you'll get a check mark there? You'll get a green check if all of a sudden you get some good deals there.
- 33:06
- Unidentified speaker (UNKNOWN): Um
- 33:07
- Unidentified speaker (speaker_0): are these credit default swaps, are those spreads, are they are they coming down? You know, we're at 95 basis points for Amazon right now. Is that gonna going come to back down? to 70? We're at 70 a month ago. Watch that. That could meaningfully help. That'll boost spending plans if that happens. These supply shortages, listen to the earnings reports. Are we getting delays and cancellations and stretch outs because
- 33:27
- Unidentified speaker (speaker_0): of supply shortages? As long as that persists, it's going to be downward pressure on capital spending plans. Um, and then demand. You guys still watch demand on those ARR numbers. I think that's all gonna be fine and dandy. And then ROI, you're gonna want to see these guys. get better ROI numbers, but again, you're not gonna get an update on that until January, February with the next wave of earnings report. So when I look at it from all these perspectives and all these
- 33:47
- Unidentified speaker (speaker_0): angles and try to cut this pie as many ways as I can, I come to the conclusion that technically and fundamentally there is no reason to rush in and buy this dip right now. There just isn't. I think this will become a fantastic buying opportunity.
- 34:03
- Unidentified speaker (speaker_0): but why bleed out?
- 34:06
- Unidentified speaker (UNKNOWN): We
- 34:06
- Unidentified speaker (speaker_0): uh we we have stocks that are down across the AI complex, right? I mean, what Nebbius is down 50% from its highs, applied opto is down 80% from its highs. Like these are stocks that have been absolutely clobbered, and people have been telling you to buy the dip the whole way down. You don't have any money left for the real bottom, right? So that's what we're kind of doing is trying to say, okay, let's let's at least step to
- 34:26
- Unidentified speaker (speaker_0): the side. sidelines, try and preserve capital, and try and have cash ready for when the real bottom forms, we can go ahead and buy that one. I'm not seeing signs of a real bottom yet. And so for that reason, we're saying sideline
- 34:40
- Unidentified speaker (UNKNOWN): it.
- 34:42
- Unidentified speaker (speaker_1): Excellent. Yeah, I mean, I I think we've exhausted the uh what what should be be looking for conversation. And I just wanted to for for funsies. kind of uh hypothetical questions on compute specifically so in terms of you know politics and the the lawmakers uh what do you think of kind of uh the
- 35:01
- Unidentified speaker (speaker_1): US adopting that French model of you know uh taxing the internet so if if lawmakers started to tax compute what outcomes do you see coming from that? Is there is there a potential for a tailwind there or would that be a headwind?
- 35:16
- Unidentified speaker (speaker_0): All of these things
- 35:18
- Unidentified speaker (UNKNOWN): are
- 35:19
- Unidentified speaker (speaker_0): short term headwinds and long term tailwinds. And that's why I just want to be abundantly clear. I'm not like coming up here as an AI bull and saying, like, oh my God, I'm freaked out now, sell everything. Like that's not at all what I'm saying.
- 35:33
- Unidentified speaker (speaker_0): I am actually
- 35:35
- Unidentified speaker (speaker_0): more
- 35:37
- Unidentified speaker (speaker_0): bullish, more. positive, more optimistic than ever on the two to five year outlook for AI stocks. I've never been more positive on that outlook because
- 35:54
- Unidentified speaker (speaker_0): I think the changes that are coming, there will be a painful transition adjusting to these changes, but the changes that are coming, if we are successful in implementing them and executing on them,
- 36:09
- Unidentified speaker (speaker_0): are changes that create a much more durable growth backdrop on the other side of this transition. It's like
- 36:20
- Unidentified speaker (speaker_0): if you are
- 36:23
- Unidentified speaker (UNKNOWN): I
- 36:23
- Unidentified speaker (speaker_0): think this is a good analogy. If you're
- 36:26
- Unidentified speaker (UNKNOWN): on the
- 36:26
- Unidentified speaker (speaker_0): on the workout train and you're working out every day, right? And you want to get as big and as strong as possible. Well, the quickest way to do that is to
- 36:36
- Unidentified speaker (UNKNOWN): do
- 36:36
- Unidentified speaker (speaker_0): creatine and steroids and all these supplements and just jack your body up with legal and illegal stuff, and you're gonna get massive, right? Like you will get massive. You work out every day and you take all these supplements, legal and illegal, you will get massive. No doubt about it. But you do that forever, and you are going to
- 36:55
- Unidentified speaker (speaker_0): absolutely destroy your body, if not kill yourself, right?
- 37:01
- Unidentified speaker (speaker_0): What a trainer would tell you, what a physician would tell you, what an expert would tell you is: don't do that stuff. But let's say you've already done all this stuff. Transition off of it.
- 37:11
- Unidentified speaker (UNKNOWN): It's
- 37:12
- Unidentified speaker (speaker_0): gonna going to be a rough transition off of it. You're Your gaining. gains will be slower, but with just some basic protein powder and a little bit of creatine and a healthy diet, you'll be able to have a great physique, great health, great condition for a very long time. Right?
- 37:30
- Unidentified speaker (UNKNOWN): It's
- 37:30
- Unidentified speaker (speaker_0): going to be a really tough transition going from, you know, I'm you know steroided out and I can bench press 9,222 pounds down to you know a more reasonable human level.
- 37:41
- Unidentified speaker (UNKNOWN): But
- 37:41
- Unidentified speaker (speaker_0): on the other side of that transition, you're going to be a much healthier person with much more longevity and durability and less health risk, long-term health risks, right? You're giving up short-term gains for long-term durability, for longevity. That is exactly what's happening to the AI trade right
- 37:57
- Unidentified speaker (UNKNOWN): now.
- 37:58
- Unidentified speaker (speaker_0): We got steroided up. And now we are the market is trying to say,
- 38:04
- Unidentified speaker (UNKNOWN): all right,
- 38:05
- Unidentified speaker (speaker_0): I guess we're going to come off the steroids. And that's going to be painful for a bit to the earnings estimates of a lot of the companies that relied on the steroids. But we're still going to have protein powder and we're still going to have a little bit of creatine. We're still a little bit of all this other stuff on the other side. And it's going to be a very healthy long-term term trade. And so that's why I think, you know, the ant I always
- 38:25
- Unidentified speaker (speaker_0): said politics was the big risk here.
- 38:27
- Unidentified speaker (UNKNOWN): If
- 38:27
- Unidentified speaker (speaker_0): we can get rid of that political risk, because now we responsibly build data centers and we advertise the jobs we're creating and we create more, I don't know, not moral AI, but you know, AI that thinks about the world. Like the
- 38:43
- Unidentified speaker (speaker_0): anti AI populist backlash stops.
- 38:46
- Unidentified speaker (UNKNOWN): And
- 38:46
- Unidentified speaker (speaker_0): all of a sudden the political risk is gone. And it's not on the twenty 2028 twenty eight ballot. It's not on the twenty thirty 2030 ballot or the 2032 ballot. It's gone. That'd be great. That'd be great for the longevity of the trade, right? Because the worst case outcome would have been you get massive and building anti-AI backlash into 28, and you get a true and total blue
- 39:06
- Unidentified speaker (speaker_0): wave built exclusively on anti-AI policies, and you get a super harsh crackdown in 29. That would have shot the trade out dead, right? That would have been just killed. it entirely. That I don't think that's going to happen now.
- 39:18
- Unidentified speaker (UNKNOWN): So
- 39:18
- Unidentified speaker (speaker_0): that's great. Another great thing is that when again, when you reduce the capital spending to current levels, you maintain it current levels. Guess what? You can cash fund it all. So
- 39:29
- Unidentified speaker (speaker_0): it doesn't matter what the treasury rate is, it doesn't matter what the credit spread is. It doesn't matter Uh, whether the stocks go up or down because we're not relying on debt debtor or equity markets. So you get rid of that external cash. capital risk, that's a fantastic place to
- 39:41
- Unidentified speaker (UNKNOWN): be. So
- 39:42
- Unidentified speaker (speaker_0): I did, and then you get rid of the supply shortages, right? If you build more slowly and more responsibly, you wait for power to come online. You wait for there to be enough optics, you wait for there to be enough chips, and all of a sudden the supply shortage risks get thrown out the window. So by slowing down now and by going through a painful transition period, we ensure
- 40:03
- Unidentified speaker (speaker_0): durability on the other side of the transition, which is why I think you're going to get a very fantastic buying opportunity in all these stocks. Just maybe not here, not now. It feels like there's probably some more
- 40:16
- Unidentified speaker (speaker_0): chaos on the horizon, if you will. Um, but for now, I feel short-term pain, long-term gain. I want to just emphasize I am very, very bullish on. the
- 40:27
- Unidentified speaker (UNKNOWN): I
- 40:27
- Unidentified speaker (speaker_0): mean anything really out to 12 months. Look, look at our charts, look at our data. It says six months, 30% drawdown. That's what the data says. I'm not bearish beyond six months. This is not the dot-com bus. This is 1997-98. This is 95-96. This is a shakeout.
- 40:44
- Unidentified speaker (UNKNOWN): Does
- 40:45
- Unidentified speaker (speaker_0): that mean the shakeout is bottomed? Probably not, but it is still just a shakeout. So get ready, have the cash. ready, and buy the dip when bottom signals do actually form.
- 40:54
- Unidentified speaker (speaker_1): Definitely. Yeah. I I I want to wrap up the uh kind of episode. I just want to have one last question here. I want to borrow your uh your steroid analogy there. So one of the conversations that comes up a a lot is um if we do have a slowdown and data center build build-out, out, there is this potential for compute crunch in the future, you know, vis
- 41:14
- Unidentified speaker (speaker_1): vis-a-vis vis we stop taking the steroids, then we have to go to the gym to lift the actual weight of the compute, but we just don't have the juice to do it. Do you think there is a future where that happens? Or is that uh kind of a hypothetical uh nothing burger?
- 41:28
- Unidentified speaker (speaker_0): Yeah, I think that's exactly what happens.
- 41:31
- Unidentified speaker (UNKNOWN): And
- 41:32
- Unidentified speaker (speaker_0): you know what that's super bullish for? The entire freaking AI trade. If all of a sudden you really are compute constrained because you're limiting the amount of stuff that can be built every year, or you're capping your spending at some certain level because the markets are pressuring to, and Washington's just introducing red tape and all this stuff.
- 41:51
- Unidentified speaker (UNKNOWN): You
- 41:51
- Unidentified speaker (speaker_0): know what that does to the existing real estate out there? You know what that does to the existing infrastructure? It makes it so freaking valuable. And you know what it does to the economic. economics of Frontier Labs? It meaningfully improves them because they're going to have way more pricing power over people like
- 42:11
- Unidentified speaker (speaker_0): us, over enterprises, over all that stuff. And so, yeah, I think that's exactly what's going to happen. And that is a very long term bullish thing. Because if you if you really are supply-demand supply, demand, and imbalanced, balanced, and you now cap yourself to only bring on a certain amount of supply. every single year, then the supply, demand, imbalance is
- 42:31
- Unidentified speaker (speaker_0): set to persist. The supply glut will not happen. And so the pricing power of the suppliers, the margins of the suppliers,
- 42:40
- Unidentified speaker (UNKNOWN): the
- 42:41
- Unidentified speaker (speaker_0): revenues of the suppliers stays higher for longer. So what we have to do right now again is just have a little bit of a reset.
- 42:48
- Unidentified speaker (UNKNOWN): You know,
- 42:48
- Unidentified speaker (speaker_0): everybody thought the spending would grow to the trees and in 50%. infinity forever. It's not. Washington's changing that. Wall Street's changing that. Supply shortages are changing that. Near-term weakening ROI economics are changing that. So now we gotta reset
- 43:04
- Unidentified speaker (UNKNOWN): those
- 43:04
- Unidentified speaker (speaker_0): I those estimates. They come down a little bit. Okay, stocks get hit.
- 43:08
- Unidentified speaker (UNKNOWN): They
- 43:09
- Unidentified speaker (speaker_0): bottom bought out. them And now you look at the landscape, and it's like, wait, we're stable. We're good. It's all cash funded. Trillion. bucks per year. Okay, NVIDIA is gonna earn, I don't know, whatever their EPS is, they're gonna earn that plus 10% per year for the next 20 years. There will be confidence in the out-year estimates at that point. Micron will not trade at four times forward earnings
- 43:28
- Unidentified speaker (speaker_0): because people will believe the forward earnings at that point, which is why one of the things I'm watching for is those EPS estimates to come down. When they start coming down, we're gonna going to start bottoming in a big way. Watch
- 43:39
- Unidentified speaker (UNKNOWN): it.
- 43:40
- Unidentified speaker (speaker_1): Excellent. Excellent. All right. So I think that's a great place to wrap it, Luke. Um, we are in the rough, growing pains of the AI market, but the future outlook is bright. It should be beautiful. Um, that is being exponential. Signing off. Take care.