1May's 4.2% inflation print was an oil price, and that oil price has fallen
The number that unsettled markets in May — US consumer prices back at 4.2% year on year — came almost entirely from one input. Joseph Wang, on Fed Guy, takes it apart: core CPI rose 0.2% month on month, which is essentially where the Fed wants it, and the whole gap between core and headline is the crude spike the Iran war produced. Crude has since fallen back from above $100, with the Strait of Hormuz transiting again under US naval escort, and oil dropped more than 4% on reports that a US–Iran agreement is close. University of Michigan medium-term inflation expectations fell in the same week, which is the measure the Fed watches for second-round effects.
Wang's sequence from there is mechanical: a durable peace keeps crude down, the statement moves from a cutting bias to neutral, and cuts follow eventually. Rates still price roughly two more hikes ahead, so almost none of that is in the price yet.
What breaks it is inventory. The cushion that got crude down was not new production. It was a Strategic Petroleum Reserve drawn to near its minimum operational level, plus a Chinese reserve draw — and that cushion is now spent. If no deal is signed by month-end there is nothing left to absorb the next disruption, and the disinflation reverses from exactly where it came from. The deal is not signed yet: reports over the weekend disagreed on whether the signing is Friday or Sunday, though the UK, France, Germany and Italy are described as ready to lift sanctions, with an oil-sanctions waiver and an asset release in the draft.
2Big tech has stopped buying its own stock, and the new supply is arriving all at once
The most structural claim in the day's material has nothing to do with valuation. For a decade the most price-insensitive buyer of US equities was the issuers themselves. Joseph Wang's argument is that this bid has reversed, and that the same firms are now on the other side: has signalled stock sales, is running an at-the-market offering, announced a raise of about $7bn and fell roughly 30% on the week, and the buyback support big tech used to provide is being redirected into AI capital spending. Behind that sits the primary market — SpaceX's Friday listing raised about $75bn, the largest IPO on record, with Quantinuum, OpenAI and Anthropic queued behind it. Rising net issuance is the unusual state here, not the normal one.
Three channels noted the same funding mechanism: mega-caps sold to raise the cash to buy the new paper, with cited each time and on the same near-term-weak list. That is a rotation inside the index rather than money arriving from outside it.
The three readings of SpaceX itself do not agree, and the disagreement is the useful part. Wang reads it as supply: multi-year losses, and staggered lockups that will release insider and venture selling for months. The Compound and Friends reads it as idiosyncratic — the oversubscription figures were called fake on air, and the float-adjusted index impact judged manageable rather than a liquidity drain; the same episode read Berkshire financing roughly 10% of Google's equity issuance as evidence the supply is finding committed buyers. Brent of SpotGamma, on Excess Returns, reads it as pure flows: 550m shares at $135 for a valuation near $1.7trn — roughly 94 to 100 times sales, which would rank about eighth in the S&P 500 — up 20 to 30% intraday on its debut, with a tight float meeting index demand from the Russell rebalance and some twenty ETFs. He would rather be long than short into the options listing, citing Avis and its 750% low-float melt-up as the template. Even the valuation is contested: the three estimates ran from about $1.7trn to $2trn.
The horizons reconcile better than the views do — a squeeze and an overhang can both be right — but only one of them has a date on it, and it is the lockup schedule. The last cohort is not encouraging on that score: (CoreWeave) and Cerebras faded about 50% from their listings.
3The memory shortage is real, and the trade on it is already crowded
The most detailed single-name work in the day's videos was a supply-chain map. Business with Brian argues that AI is consuming high-bandwidth memory faster than it can be manufactured — HBM prices roughly doubled in a quarter and the makers are sold out into next year — and that the way to own that is the chokepoints rather than the maker. (Micron), which has just crossed a $1trn market value, is the obvious expression and, in his framing, the crowded one. He maps five layers instead: fab tools and , inspection and , test (Advantest) and , foundation , (BE Semiconductor) and , and interface . The claim underneath is that these get paid whichever maker wins, across a five-year buildout in which HBM4 and hybrid-bonding orders are being placed now for 2027. He names two ETFs on the theme without choosing between them: , which holds makers only, and , which holds makers plus the supplier stack.
Two other channels supply the positioning half of the picture, and it points the other way. On The Compound and Friends, Friday's selloff was read as healthy precisely because of what it hit: had run about 150% above its 200-day moving average, and momentum crowding reached the 90th percentile before unwinding sharply. Brent of SpotGamma measured the same thing in options — call buying in , , and briefly carried more premium than and did.
Hold the caveats with the basket. It is one person's framework with financials paraphrased rather than verified; and trade over the counter with thin liquidity; is a loss-making 2027 story; carries an open antitrust inquiry and is the smallest weight in his own basket for that reason; and the video making the case also carried a paid sponsorship for a junior gold miner. What survives all of that is a physical-shortage story with a real price attached, arriving at a moment when the crowd is already positioned for it.
4The rotation out of AI has failed twice now, and both times it was oil
The S&P 500 is up about 11% so far this year, and roughly 40% of its members are down over the same stretch, with the equal-weighted Magnificent Seven negative. The index gain is the AI-capex story and not much else — though not uniformly, since (Broadcom) missed on revenue.
Brian Levitt of Invesco, on The Compound and Friends, argues the earnings are broadening even where the prices are not: seven of eleven sectors grew earnings at double-digit rates in the first quarter. What stops price following earnings, on his reading, is that an oil shock arrived and broke the broadening trade for the second year running. His conditions for it to work are four — oil peaking, which he thinks it has, rates peaking, inflation expectations peaking, and the Fed easing. That is the same list as the first section approached from the other end, and the reason a Gulf headline is a breadth story as much as an inflation one. Levitt watches five-year breakevens (250bp comfortable, above 3% not), credit spreads, bank lending, the dollar and system leverage, and explicitly not price-to-earnings multiples. His central line is that the knockout punch only lands when you are already in a bear market, and there has been no genuine recession in eighteen years and no visible credit problem now.
The counter-case reached the same episode second-hand: BofA's Subramanian argued for taking profits, with seven of ten bear signposts triggered and dispersion echoing February 2000. The hosts pushed back and framed it as a rotation call rather than a move to cash. Worth noting the composition of the room — an Invesco strategist and a bullish-leaning house discussing the week's one bear note, with no bear present to argue it.
Two structural facts sit alongside. has outrun the S&P by about 12% over two months, and SpaceX listing on the NASDAQ rather than the S&P could entrench that gap. And the expressions offered for a broadening are deliberately not AI names: (Travelers), with underwriting profit up 21% year on year and the improvement credited to AI, as the flagship for AI showing up in a customer's numbers rather than a supplier's; financials and healthcare as the laggards that would benefit; (Casey's) and (Hilton) as evidence the non-AI economy is still delivering. If the oil call in the first section holds, that is where it shows up. If it does not, this is the third year the same trade fails for the same reason.
What the sources recommended
Twelve buy calls came out of six videos, and they are not twelve independent ideas: ten are one person's single thesis, and the other two come from a video that is openly a promotion. No source issued a sell or an avoid.
The memory basket. , , , , , , , , and are the ten names in the Business with Brian supplier map above, presented as a hypothetical $100 basket weighted toward the front of the stack — $25 to tools, $25 to inspection, $22 to test, $20 to the foundation layer and $8 to , the lightest weight because of that antitrust inquiry. Read it as one call on the HBM supply chain rather than ten.
The miners. (Idaho Strategic Resources, pitched in a $28–30 buy zone as a profitable producer that also holds the largest US rare-earth land package) and (Integra Resources, around $2.50, a roughly 40%-margin Nevada producer funding its own growth) come from Ross Givens, in a video that promises 200 to 300% upside in its title and funnels to a paid subscription service. Both are thinly traded small-caps. and were named only as large-cap alternatives, with no call attached.
The macro thesis under those miners arrived twice in one day, both times attached to a sales funnel. Ross Givens and Felix, on Felix & Friends, each argued that the government inflates away $39trn of debt by holding rates below inflation, replaying 1946–74, and each landed in hard assets — gold miners in one case, physical silver in the other. Felix's version reached no ticker at all; the stocks he named were illustrations, and the actionable item was an email-gated workbook. He does argue against the popular version of his own trade, saying COMEX is durable rather than about to break, and he advises sizing and averaging in rather than going all in.
The bearish material stayed observational. The SpaceX lockup overhang, down about 30% on its raise, and near-term weakness in the mega-caps were all discussed as risks, and none of them was put as a sell.
The week stacks a Fed meeting, a record IPO's first options and a $2.5trn expiry into four days
The Federal Reserve meets 16–17 June, Kevin Warsh's first meeting as chair. Three of the day's videos flagged it, and the expectation reported is that the statement moves from a cutting bias to neutral, with the dot plot possibly dropped altogether; Joseph Wang argues the Iran de-escalation raises the odds of a dovish surprise. SpaceX options list on the 16th, the trigger Brent of SpotGamma named for a squeeze. VIX expiry falls on the 17th, and quarterly options expiration on Thursday the 18th carries roughly $2.5trn of S&P delta, call-weighted, pulled forward a day because US markets close for Juneteenth on the 19th. The levels he named around it: the S&P at 7,420–7,450 with mildly positive gamma, 7,575–7,600 as the cap he expects into expiry, and 7,400 as the line below which dealers sell weakness rather than buy it. His correlation gauge, which fired before the 5 June spasm that traded 108m contracts, has dropped back under 8 again.
After that the calendar keeps testing the same claims. SpaceX joins the Russell around 20–21 June and the NASDAQ in early July, which is when the index demand in the second section either materialises or does not. SK Hynix gives an investor update on the 22nd — the cleanest outside read on the memory shortage in the third. Quantinuum's IPO tests whether the appetite that absorbed $75bn extends to the next one. Earnings are second order this week: on the 16th, (CarMax) and (Jabil) on the 17th, (Accenture) and (Kroger) on the 18th.
