1The semiconductor cycle is being priced as if it stopped being cyclical
Warren Pies, of 3Fourteen Research, on Excess Returns, puts the current semiconductor run 31 months in and measures it against the 1990s cycle, which lasted roughly 63 months and returned about 11x. This one is up roughly 4x so far. On that arithmetic the recent weakness is one of the 20%-plus pullbacks he expects about once a year inside a megatrend, rather than the end of one.
The uncomfortable part of his own case is the valuation. Margins have exploded and price-to-sales has expanded alongside them — the reverse of the usual cyclical pattern, where the multiple compresses as earnings peak. He calls that the one genuine air pocket of exuberance he can find, and declines to extend it to the rest of the market.
His demand evidence is the checkable part. The token panic — the fear that cheap open-source models collapse the price of inference — runs into pricing for frontier tokens rising and total inference volume up 40% month on month in OpenRouter data. Spare capacity on his firm's Neocloud GPU-availability index sits at roughly 0-3%. Supply that tight is hard to square with demand having rolled over.
Prices in the same news window went the other way. fell 5.25% and 1.4% on semis weakness, and slipped about 2% in pre-market trade on memory-pricing concerns, in the same week it joined the trillion-dollar club (Yahoo Finance, Google News). went the other way, rising on a ByteDance deal. None of that settles the argument either way. A cycle priced as though it is not a cycle only survives if the margins hold, which puts the next few quarters of margin data in the deciding seat.
Two channels published inside the day's window, so nothing on this page is corroborated by an independent second voice.
2A soft June jobs report did not make a rate cut the Fed's next move
June payrolls came in weak and the dollar tumbled — the pairing that normally pulls the next cut forward. Pies reads it as a true hold, and says that if the Fed moves at all, a hike is likelier than a cut, while not expecting one inside 2026 because the committee's core still treats current rates as restrictive. The consequence he draws is not about the funds rate: with no cuts arriving, multiple contraction becomes his base case, so index returns have to come out of earnings rather than out of a re-rating.
He puts a policy mistake at the top of his 2026-27 risk list — his framing is that bull markets get killed by central banks rather than dying of old age. The ranking does not change his position. He holds an S&P 500 target of 7850, about 15% higher on the year, and stays long equities until the Fed actually tightens or the economy visibly weakens. The trigger is the tightening, not the forecast of it.
Underneath the aggregate labour number sits a composition problem he calls a K-shaped economy. Fixed investment in IT and equipment, at about $1.5trn, has overtaken residential investment at about $1.1trn — data-centre capex crowding out housing. Residential-construction payrolls have fallen 2% two months running, against the 8% mark he treats as pre-recessionary. So the same labour market can read as a muddle-through in aggregate and as a contraction in the interest-rate-sensitive part. The winning side of that split shows up in the tape too: was bid on gas turbines taking demand from the AI and data-centre build-out (Finnhub). With the dollar falling, precious-metals miners led the pre-market gainers, at the front (Google News).
3The crude story flipped from shortage to glut, and the margin moved downstream
Crude sits at four-month lows. Pies's mechanism is added supply meeting a missing buyer: strategic-reserve releases putting barrels into the market while China is absent as the swing consumer. Positioning has followed — managed money is roughly 40% net short, which he reads as extreme pessimism rather than confirmation, and his path is a short-covering spike toward $85-90 Brent before crude settles into a $75-80 clearing range. UBS cut its Brent forecast in the same week, and Hormuz flows are recovering. His condition for oil mattering to the wider macro is narrow: it takes a re-closure of Hormuz.
That tail has not shut. Reuters coverage carried through Finnhub reported Iranian de-escalation holding, alongside the death of Ayatollah Khamenei and a week of mourning and funerals — a succession sitting on top of the same strait.
Downstream is where he says the money is being made. Refiners (, , ) are "minting money" on cheap crude against US gasoline and distillate inventories at multi-decade lows. He frames it as a decent energy trade rather than anything systemic, and pitches it at the sector rather than at a single name. The asymmetry is the point: the glut that squeezes the producer is the input cost that pays the refiner, so the two halves of the same energy story move in opposite directions.
4Index reconstitution became a twice-yearly forced-buying event, and the next date is already set
Ross Givens built a video on one mechanism: about $12trn tracks the Russell indexes, and the funds tracking them are mechanical, price-insensitive buyers of additions and sellers of deletions. The 2026 rebuild ranked constituents on 30 April and the new indexes took effect on Monday 29 June. What turns it into a recurring calendar item is the change he leads with — reconstitution now runs twice a year instead of annually, with the next effective date 11 December 2026 and a preliminary add-and-delete list published around November.
His flagship name is (Jack in the Box), up 20% on that Monday, its biggest single-day move in six years, on a Russell 2000 addition with short interest of 30-35% and a valuation of 0.2 times sales. He says he is long and will not sell unless it falls back into the 13s.
The supporting case studies come with the move already in them: (Hut 8, the largest Russell 2000 Growth addition by size, roughly doubled), (JFrog, the largest addition by weight, 40 to 92), (Kymera, dropped from the Russell 1000 into the 2000 at a large new weight, 70 to 115), (Life Time, the same move down), and (Bloom Energy, promoted from the 2000 into the 1000, where the forced buying and selling offset). Three of the five come with the run already banked, one is described as a wash, and none is a loser — a curated sample rather than a track record. The video also breaks three times to sell a subscription trading service. The mechanism is checkable; the evidence shown for it is not.
His own caveat is the familiar one: desks front-run the event, so "buy the rumor, sell the news" applies, and he suggests trimming a quarter to a half of a position into post-addition strength. The flow has a sector shape as well — small-cap index weight tilting out of industrials (−7%) and into healthcare (+4%) and energy (+2.5%), with half of this year's Russell IPOs in healthcare. The mechanical part of the move therefore happens between November's preliminary list and December's effective date, which makes the next one observable before it lands rather than after.
What the sources recommended
Two videos produced exactly one single-name call; everything else was pitched at sector or theme level.
The buy. (Jack in the Box), from Ross Givens, on a weeks-long horizon — a Russell 2000 addition with 30-35% short interest at 0.2 times sales, held unless it falls back into the 13s. Two things travel with it. His own hedge is that post-addition strength is where the desks sell the news, so he trims into it rather than adds. And the video carrying the call breaks three times to sell a subscription service and shows only case studies that worked, so the call is not disinterested.
Everything from Warren Pies sat above the single-name level: semis buyable on dips, refiners favoured on cheap crude and multi-decade-low product inventories, equities held long until the Fed tightens or growth cracks. The only caution offered on either video was to trim partial profits into post-addition strength, which is a de-risking step rather than an exit.
The forced-buying mechanic gets retested on a bigger index next week
US equity markets were closed on Friday 3 July for the Independence Day observance, 4 July falling on a Saturday, so none of this gets priced until Monday 6 July. The dated event inside that week is an expected Nasdaq-100 inclusion around 7 July, reported by Yahoo Finance and Google News alongside the SpaceX listing frenzy that has bid up space-exposed stocks and ETFs — the same mechanical inclusion flow as the Russell claim, on a larger index. Adjacent to it, rose on a Starlink Wi-Fi deal (Yahoo Finance). The semiconductor claim gets its next read from memory pricing: if it keeps pressing after the trillion-dollar week, the margin structure holding up the multi-year case is where the strain shows first.
