I don't like trying to predict market peaks, but this certainly feels like an inflection point. But: “Markets can remain irrational longer than you can remain solvent.” - John Maynard Keynes
So I'll just continue to dollar cost average into my index funds and durable companies and take the long view.
DCA into "durable companies" works until you check what's inside the index. The S&P just hit all-time highs while behaving like a five-stock portfolio, concentration exceeding the dot-com peak. Your index fund is making a concentration bet whether you chose one or not.
That is somewhat true, but the S&P automaticity rebalances itself through technology cycles and the top 5 companies right now are themselves very diverse.
Amazon is a global retailer, a large entertainment company, a logistics operator and the largest cloud operator.
Google is the same story, its search, its AI, its entertainment in YouTube, its hardware and software.
Apple is also similar, its hardware, its services, its music and video streaming, etc…
Microsoft might be the most diverse, from gaming to personal computers to enterprise software to cloud services to now AI. Shoot they even own LinkedIn, Xbox, Bing, and Activision Blizzard. Even all of those together are ultimately a small part of the company compared to Azure and enterprise solutions.
Meta is more concentrated in ad revenue but still is a very large and diverse business, from three of the most used apps world wide to a growing hardware business and AI.
The point is that none of these companies are businesses I’d want to bet against or not be invested in, at least within the index.
The March 9, 2000 data point is the most specific and testable historical parallel Ive seen in any tech bubble analysis this year. PHLX SOX surging 50% in 25 trading days has happened exactly twice in the indexs history. The first time was the literal day before the dot-com peak. We're living through the second time right now.
The question that determines wether this is a genuine echo or a misleading pattern is what drove the semiconductor demand in each case. In 2000, chip demand was being pulled forward by enterprise IT spending on Y2K remediation followed by the telecom buildout. When the pull-forward exhausted itself the demand cliff arrived overnight because the spending had been borrowed from the future rather than generated organicaly.
The AI semiconductor demand has a different structure but its not immune to the same dynamic. $725 billion in hyperscaler capex is being committed for 2026 based on revenue growth projections that assume AI adoption continues accelerating. If those projections prove even 20% optimistic the capex gets revised downward and the chip orders that drove the 50% surge in SOX get cancelled or deferred. The chips are real. The demand is real. The question is wether the demand has been pulled forward from 2027-2028 into 2026 the way telecom demand was pulled forward into 1999-2000.
Burry taking short positions with 2027 expiry is the timing signal that deserves attention. Hes not betting the crash happens tomorrow. Hes betting the demand pull-forward becomes visible in earnings revisions within 18 months, which is consistent with the historical lag between peak semiconductor ordering and the moment the market realises the orders represented a cycle peak rather than a new permanent baseline.
The "rising tide lifts all boats" observation is the part that should concern index investors most. When every stock in a semiconductor index rises at least 14% in a month regardless of individual fundamentals, the market is pricing the sector as a single trade rather than as individual companies. Thats exactly how corrections propagate fastest because when the sector trades as one position, it sells as one position.
Worth pairing with the Micron print from March: revenue nearly tripled year over year, 75% gross margins on what was a commodity 18 months ago, $33B guide for next quarter. Stock fell anyway. The demand question got answered. What spooked the tape is the next one, which is where your capex math lives.
good stuff as always, thx. On the Semi's, there was a “HF Prime Book” graphic going around the street late last week on “Info Tech”. The dispersion between the “HF Prime Book” vs my overall "Nasdaq Speculator Positioning" oscillator (which remains pinned at Extreme Lows and showing the first true net-short position going back 18+ months) can be explained by the dramatic increase in Semi’s (and trash AI companies) YTD. The Prime Chart is clearly negative for "Info Tech” vs “Nasdaq” pair trade. Importantly, the correlation into year end ’25 between SMH and QQQ was 0.8. The YTD correlation has disconnected to near 0.6. No SMH position here, not a permabear, just an observation.
I don't like trying to predict market peaks, but this certainly feels like an inflection point. But: “Markets can remain irrational longer than you can remain solvent.” - John Maynard Keynes
So I'll just continue to dollar cost average into my index funds and durable companies and take the long view.
I think that using the Dollar Cost Average strategy is a long term winner!
DCA into "durable companies" works until you check what's inside the index. The S&P just hit all-time highs while behaving like a five-stock portfolio, concentration exceeding the dot-com peak. Your index fund is making a concentration bet whether you chose one or not.
That is somewhat true, but the S&P automaticity rebalances itself through technology cycles and the top 5 companies right now are themselves very diverse.
Amazon is a global retailer, a large entertainment company, a logistics operator and the largest cloud operator.
Google is the same story, its search, its AI, its entertainment in YouTube, its hardware and software.
Apple is also similar, its hardware, its services, its music and video streaming, etc…
Microsoft might be the most diverse, from gaming to personal computers to enterprise software to cloud services to now AI. Shoot they even own LinkedIn, Xbox, Bing, and Activision Blizzard. Even all of those together are ultimately a small part of the company compared to Azure and enterprise solutions.
Meta is more concentrated in ad revenue but still is a very large and diverse business, from three of the most used apps world wide to a growing hardware business and AI.
The point is that none of these companies are businesses I’d want to bet against or not be invested in, at least within the index.
The March 9, 2000 data point is the most specific and testable historical parallel Ive seen in any tech bubble analysis this year. PHLX SOX surging 50% in 25 trading days has happened exactly twice in the indexs history. The first time was the literal day before the dot-com peak. We're living through the second time right now.
The question that determines wether this is a genuine echo or a misleading pattern is what drove the semiconductor demand in each case. In 2000, chip demand was being pulled forward by enterprise IT spending on Y2K remediation followed by the telecom buildout. When the pull-forward exhausted itself the demand cliff arrived overnight because the spending had been borrowed from the future rather than generated organicaly.
The AI semiconductor demand has a different structure but its not immune to the same dynamic. $725 billion in hyperscaler capex is being committed for 2026 based on revenue growth projections that assume AI adoption continues accelerating. If those projections prove even 20% optimistic the capex gets revised downward and the chip orders that drove the 50% surge in SOX get cancelled or deferred. The chips are real. The demand is real. The question is wether the demand has been pulled forward from 2027-2028 into 2026 the way telecom demand was pulled forward into 1999-2000.
Burry taking short positions with 2027 expiry is the timing signal that deserves attention. Hes not betting the crash happens tomorrow. Hes betting the demand pull-forward becomes visible in earnings revisions within 18 months, which is consistent with the historical lag between peak semiconductor ordering and the moment the market realises the orders represented a cycle peak rather than a new permanent baseline.
The "rising tide lifts all boats" observation is the part that should concern index investors most. When every stock in a semiconductor index rises at least 14% in a month regardless of individual fundamentals, the market is pricing the sector as a single trade rather than as individual companies. Thats exactly how corrections propagate fastest because when the sector trades as one position, it sells as one position.
It is very interesting to see the similarities between these two scenarios
Worth pairing with the Micron print from March: revenue nearly tripled year over year, 75% gross margins on what was a commodity 18 months ago, $33B guide for next quarter. Stock fell anyway. The demand question got answered. What spooked the tape is the next one, which is where your capex math lives.
Covers the economic side of the current worldly affairs! Must read ✅️
good stuff as always, thx. On the Semi's, there was a “HF Prime Book” graphic going around the street late last week on “Info Tech”. The dispersion between the “HF Prime Book” vs my overall "Nasdaq Speculator Positioning" oscillator (which remains pinned at Extreme Lows and showing the first true net-short position going back 18+ months) can be explained by the dramatic increase in Semi’s (and trash AI companies) YTD. The Prime Chart is clearly negative for "Info Tech” vs “Nasdaq” pair trade. Importantly, the correlation into year end ’25 between SMH and QQQ was 0.8. The YTD correlation has disconnected to near 0.6. No SMH position here, not a permabear, just an observation.