In short: The one short he will name, precisely because it is an index and not a company. "I'm short for example in terms of index. I'm short the SOXX index, S-O-X-X. Companies I don't want to name, but it's an index and it moves, and if it moved up from here I probably short more. If it went down 10, 15% here I'll probably cover half of it." Also "some specific semiconductors" alongside it, sized "on and off — nothing goes straight down, nothing goes straight up. We're a very active manager." He is explicitly not waiting for a crash: asked whether he needs a 50% implosion, he says a correction is fine — but the structural case is the capex peak. "Let's say in the coming years 1.7 trillion of spending, like Nvidia said 70% growth, let's say the year after it's going down 20%. So these stocks are going to go down."
SOXX is a fund that owns the big semiconductor companies as a basket. Tardif is short it — he has sold borrowed units expecting to buy them back cheaper — and he explains why he uses the basket rather than naming individual chip stocks: he does not want to disclose single-name shorts, and an index removes the risk that one company surprises him with good news while the sector rolls over.
The argument is arithmetic, not sentiment. Nvidia has guided to 70% growth next year. If the whole AI supply chain grows at that rate, total spending on this build-out reaches roughly $1.7 trillion a year. He simply does not believe an economy the size of the world's can sustain spending at that level indefinitely — so somewhere ahead is a peak, and after a peak, spending does not plateau, it declines. Chip company earnings today are the mirror image of that spending, so when it turns, so do they.
He is careful about two things. First, timing: "it could be in 27, maybe it's in 28. That's a real debate. In my mind it's not if." Second, sizing: he adds to the short when the index rallies and covers half of it after a 10-15% fall, so the position is managed against the tape rather than held to a target price. He does not need a crash — a correction is a perfectly acceptable outcome.
26:27I'm short the SOXX index, S-O-X-X. Companies I don't want to name, but it's an index and it moves, and if it moved up from here I probably short more. If it went down 10, 15% here I'll probably cover half of it. So we're — so you're not looking for 50% implosion. You're happy if we just get a correction, that's okay — Well, ultimately I do think we're going to have — we have a lot of issues in the economy and I do think that when we go, as we talk about the AI, over the mountain, when we're going to be shrinking the spending. So if we go
In short: Q2 delivered a 100% gain in the SOX with heavy retail involvement, so chopping around afterwards is "to be expected" — but he doesn't think late June was the end of the mega trend. Realized vol ~55 on a one-month look-back (peak ~185): "how much leverage do you need for sectors that are realizing that level of volatility?"
4:03As much as the big deleveraging event has happened, degrossing is still happening across the whole hedge fund complex. — It's been entertaining to see socks and IGV basically have a negative one correlation. That does suggest that there's clearly a technical element to this that some bad news for AI leads into it and ServiceNow and Salesforce stocks go up several percent.
In short: The other half of the rotation, and the week's most jarring statistic. "Despite Nvidia's guide of 70% revenue growth versus consensus of only 47%, the semiconductor index ended down 2.3% while software was up 5.9% on solid earnings." Year to date the SOX is +62%; since the June 22 momentum unwind it is −22% against IGV's +25%. Goldman's Delta One desk supplies the de-rating mechanism: credit "is starting to ask questions that equity largely ignored," Street dialogue has shifted toward inventory builds across parts of the semi chain, and power bottlenecks plus policy pushback on data centres widen the range of outcomes — "none of these kill the AI story, but the range of outcomes is clearly getting wider = lower multiples (SOX on 24-month forward P/E has gone from 21-22 to 15ish)."
The semiconductor index is the cleanest gauge of the AI infrastructure trade, and this week it did something genuinely strange: Nvidia guided to 70% growth against an expectation of 47%, and the index still fell 2.3% on the week.
Goldman's derivatives desk gives three reasons, and they are worth separating. First, credit investors have started asking questions equity investors have ignored — the cost of insuring Nvidia's and Broadcom's debt against default has widened sharply as attention turns to who is guaranteeing the AI build-out. Second, the conversation has begun to include inventory: are chips being bought and stockpiled rather than installed? Third, power shortages and political pushback against new data centres make it harder to know how fast, and how far, this actually gets built.
None of those is fatal, and the desk says so: "none of these kill the AI story, but the range of outcomes is clearly getting wider." A wider range of outcomes is exactly what compresses a multiple, and it has: the index's forward price-to-earnings ratio has fallen from 21-22 times to about 15.
That is the whole state of the trade in one number. Earnings are being revised up; the multiple paid for them is falling faster.
Full passage: premium transcript (PDF).
In short: His primary signpost: the semiconductor complex "peaked in June of this year… the stocks peaked in June 26th," sold off, and the attempted rally "seems to be rolling over." Semis are "the classic canary in the coal mine" for a second-derivative slowdown; if they go to new lows, "the bubble may have already burst."
SOXX is a basket of chip stocks, and Dowd watches it as his single most important early-warning gauge. Chipmakers sit at the front of the AI supply chain, so their share prices turn before the spending does — "the classic canary in the coal mine."
The key point is that the build-out doesn't have to stop to hurt them; it only has to slow. Fast-growing stocks are priced on how quickly growth is accelerating, so when the rate of increase merely eases, the valuations fall hard. The tape is already showing it: the group peaked on 26 June, sold off, and the comeback rally "seems to be rolling over." If it makes new lows, "the bubble may have already burst" — people just won't know for a couple of months.
4:02On tremendous orders and good fundamentals, but the stocks peaked in June 26th of this year, had a sell-off. Now we're trying to attempt a rally and that rally seems to be rolling over. So, we're kind of at the beginning phases of the market starting to realize that. And then let's also not forget there's not enough power to plug all these data center build-outs that have been hyped into the grid.
In short: The damage gauge for the semis leg of the AI trade: "the SOXX Index, which is the iShares semiconductor ETF, is down 23% from its peak on June 22nd and is down 4% this week" — against NASDAQ −7% from its Jan 2 all-time high. "Investors have displayed their nervousness by selling tech stocks and all AI related plays."
SOXX is a fund that holds the major semiconductor stocks, so its price is a single gauge for the whole chip complex. He uses it to size the damage: down 23% from its June 22 peak and down 4% in the week, against a NASDAQ that is down 7% from its January 2 high.
The gap between those two numbers is the point. Semis have fallen more than three times as far as the broad index, which is what "investors have displayed their nervousness by selling tech stocks and all AI related plays" looks like in practice — the selling is concentrated in the AI supply chain, not spread evenly.
9:28That is why NASDAQ is down 7% from its all-time high on January 2nd. Also, the Socks Index, which is the EyesShares semiconductor ETF, is down 23% from its peak on June 22nd and is down 4% this week. The change in the AI debate is also impacting fixed income markets. Coreweave, the AI data center company, is in the process of raising debt of 2.
In short: "Taking a big knife to the semis" — reduce into the bounce. The US Semiconductor Index "began its descent" in the same five weeks Samsung and Hynix fell 40%, "one that is becoming increasingly sickening for its formerly jubilant holders" (illustrated with a five-year price chart). "No slice of the S&P 500/NASDAQ was more adored than the semis were this year." Tactical instruction: "the sell-off in semis has been so intense that a bounce before long is possible, even probable. If so, that will likely be an opportunity to once again reduce exposure for disciplined investors… History is alpine lake clear that the bigger the bubble, the bigger the bust."
SOXX is a fund that owns the big US-listed semiconductor companies — the chip designers and manufacturers that were the single most popular thing to own during the AI boom. Hay's argument here is about plumbing: the Korean chip stocks and the American chip stocks are the same trade held by the same kind of investor, so when Samsung and SK Hynix fell 40% in five weeks, SOXX started sliding in exactly the same five weeks. Korea was the leading indicator, not a separate story.
What makes it alarming rather than routine is that the Korean names were falling despite excellent profits and low price-to-earnings ratios. If good earnings and a cheap multiple didn't stop a 40% fall there, then "the semis are cheap" is not a reason to hold them here. He describes the SOXX decline as "increasingly sickening for its formerly jubilant holders" — that is, the people who most enjoyed the ride up are the ones still fully exposed.
His tactical advice is the part worth writing down. After a fall this violent, a sharp bounce is likely — and that bounce, he says, is an exit, not an entry: "an opportunity to once again reduce exposure." The reasoning is that busts unwind in stages, with rallies along the way that tempt people back in, and that the size of the bust is proportional to the size of the bubble that preceded it. "History is alpine lake clear that the bigger the bubble, the bigger the bust."
In short: The leadership group — semis are "in charge of the market," now 18-19% of the S&P (up from 2% a decade ago), on a massive run since March 30th. Super overbought / consolidating after a big IPO + earnings, but "the market lives and breathes by this index" — if it collapsed, by definition the market collapses.
SOXX is a basket of semiconductor stocks (chip makers) — Sohn uses it because it's the most "even-footed" semi ETF, not dominated by Nvidia, so it reads the whole group rather than one stock. His point: semis are the market right now. They've gone from 2% of the S&P 500 a decade ago to roughly 18-19% today, and they've been on a tear since late March.
The chart is "super overbought" (it has run too far, too fast and may pause), but that's a leadership group consolidating, not topping. His blunt framing: the market "lives and breathes by this index" — if semis roll over, the whole index rolls over with them, and anyone who owns an S&P fund feels it whether they realize they own that much chip exposure or not.
2:55What is the semiconductor chart telling you? This is a market all about semis. We've seen semis on a massive run, especially since March 30th — going on almost 3 months. They're 18% of the S&P 500 now, going on 19%. Ten years ago they were 2%. From 2% to 18-19% in 10 years. Using SOXX (the most even-footed semi ETF, not too weighted to Nvidia): super overbought, perhaps consolidating, a little crescendo as we just had a big IPO and earnings — but they are in charge of the market right now. The market lives and breathes by this index. If this index collapsed, by definition the market would collapse — on the surface, index-based investors will feel that pain.
In short: The gauge of the AI-infrastructure trade — +107% total return since January (vs the Roundhill Mag7 ETF, MAGS, −8%); the visible mirror of capex flowing to the chip cartel.
Full passage: premium transcript (PDF).
In short: "Semis are the muscle of the AI trade — I prefer to own the skeleton underneath." Owns the index, not picking a winner, and trimming. Warns the SOXX has become "a gambler's paradise" of 2–4× levered derivatives — "weapons of mass destruction" — and sees more relative value in the energy "halo."
SOXX is the main semiconductor-stocks ETF. Larson owns it but is trimming, and his framing is memorable: "semis are the muscle of the AI trade — I prefer to own the skeleton underneath," meaning the power, copper and cooling that chips literally can't run without. He also warns that SOXX has become "a gambler's paradise": traders pile into 2x–4x leveraged versions of it (funds that move two-to-four times as much as the index, up or down), which he calls "weapons of mass destruction." He'd rather rotate that money into cheaper, harder-to-replace energy and metals assets — the "halo."
20:28So my favorite way to play it, and I have played it, as an allocator was by the index, the ETF that supports that entire, the SOXX, and the SOXX has been on tilt. It's the best chart ever that you want to be on and god forbid you be against it. And it's kind of funny, you look at the derivatives that fall off of even something like SOXX.
In short: Semiconductor stocks +80% in nine weeks — a dot-com-style "last gasp" blowoff top; "shortages lead to gluts." Semis are ~17% of the S&P 500 and ~30% of the Nasdaq.
SOXX is a basket of semiconductor (chip) stocks. Dowd notes the group jumped ~80% in nine weeks — the kind of vertical spike that marked the peak of the dot-com bubble, a "blow-off top."
His warning is simple supply-and-demand: today's chip shortages encourage everyone to build, which leads to gluts (too much supply) and falling prices. With chips now ~17% of the S&P 500 and ~30% of the Nasdaq, a chip reversal would drag the whole market down.
3:31And there was uh some uh some signs that it was bursting. But we've got a recent last gas blowoff top we believe uh due to the IPOs uh and uh semiconductor shortages which if you've seen the semiconductor stock move since the bottom put in in the market in at at the end of March. Uh semiconductor stocks are up 80% in nine weeks.
In short: +280% in 56 weeks (vs the Nasdaq's +130% into the 2000 top — "double"), ~54% above its 200-DMA; "the semis are starting to crash."
SOXX tracks semiconductor (computer-chip) stocks. McDonald lays out the numbers for why he thinks it's a "hyper-bubble": it's up about 280% in 56 weeks, and it sits roughly 54% above its 200-day average price (a gauge of how stretched it is above its long-run trend).
For context, the Nasdaq rose "only" 130% in the equivalent stretch before the 2000 dot-com crash — so today's chip run is about double the most famous bubble in history. His blunt conclusion: "the semis are starting to crash."
16:31Also, don't forget the Let me give you some stats. The SOXX ETF, right, which is the semiconductors, it's over 54% above its 200-day moving average because of these a lot of the Microns of the world. And then if you look at the EWY, which is the South Korea ETF, which has a lot of these semiconductor memory players in there, it's 156% above its 200-week moving average.
In short: Semis "can drop 30% in the next month" — up 36% in 14 days and ~106% above the 200-week MA, a COVID-style blowoff (his "crack smokers"). [He says "SOXS"; means the long semi ETF.]
SOXX is a basket of chip stocks. He thinks it "can drop 30% in the next month." His reasoning is that it has gone parabolic — up 36% in just 14 trading days and roughly 106% above its long-run average price line — a blow-off top he compares to the Nasdaq melting up right before the COVID crash. He calls the chasers "crack smokers." (On the page he says "SOXS"; he means the regular long semiconductor ETF.)
12:36And um yes, so I think the semiconductors um the semiconductor ETF or socks can drop 30% like in the next month. It's It reminds me of COVID where you're up on a relief rally. Never forget this. What I call the crack smokers. Never forget the crack smokers. From late January of 2020 to February 19th, the Nasdaq was up every single day almost 10% total.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.