1The AI build-out is being paid for out of buybacks, and that flips net equity supply positive for the first time since 2021
Andy Constan of Damped Spring, speaking on Excess Returns, put a number on where the money for AI capex is coming from: about $700bn. That is the size of the swing he expects in 2026, as the aggregate flow flips from companies retiring their own shares to companies selling new ones — enough, on his arithmetic, to turn net equity supply positive for the first time since late 2021. Roughly a trillion dollars a year of capex, still growing, has to be funded from somewhere, and the buyback line is the first place it comes out of.
The company-level evidence is already on the tape. cancelled its buyback and issued $80bn of common stock. raised its buyback but offset it with restricted-stock grants, leaving the net roughly flat, and added $28bn of bonds to fund capex. Constan expects issuance from and next — Amazon is already in the corporate bond market with no buyback running — while is the one still repurchasing, which makes it a relative-supply story rather than a judgement on the business.
The second half of the supply wall is the new-issue calendar. came to market as the largest IPO on record, an $85bn float priced at $135 and trading in the $170s, with Anthropic and OpenAI queued behind it; Constan's estimate is that this year's IPO paper could reach $3trn.
Two things keep this from being a trade, and Constan names both himself. It is a long-term headwind rather than a crash catalyst — his expectation is that while it is still unclear whether the capex pays off, the heaviest issuers trade down first, and he puts a three-to-six-month lag on the signal. And the whole thesis is gated on a question he leaves open: whether AI delivers disinflationary growth, more output at the same cost with no job losses, or simply cost-cutting, in which case it is not obvious who buys the output. He leans towards the first. Nothing else in the day's material corroborates the supply framework — it is one macro shop's view, arriving on its own, against a backdrop he describes as benign: index highs, a solid bond market, consumer saving with room to fall, no bubble-pop signal. The consequence, if he is right, is not a top but a change of direction in the flow that has been absorbing equity supply since 2021 — and the heaviest issuers feel it first.
2The same capex bill that is a supply problem for the spenders is a revenue line for the suppliers
Ross Givens spent his eight minutes at the other end of the same invoice. His framing is "backdoor AI" — not the companies doing the spending, but the ones paid for the physical build: for data-centre site grading and foundations, for the electrical switchgear that ties a site to the grid, for the copper running through it. All three are breakout trades rather than valuation cases — buy through $900, above $312, above $72 — and no fundamentals were shown for any of them. The same video pushed a $5 trading membership and a newsletter sign-up twice inside those eight minutes, which is worth knowing before weighing the conviction attached to the calls.
The supplier side also showed up in the day's news, priced rather than pitched. crossed into the trillion-dollar club. traded about 9% higher pre-market after Morgan Stanley put the hard-disk shortage running to 2028. Both are components rather than platforms, and both were repriced on the same build-out.
The two videos never address each other, and on the surface they point opposite ways — one treats a trillion dollars of AI capex as an overhang, the other as a customer. They are not really in conflict. It is one bill read from both ends, and the sequencing is the useful part: on Constan's own mechanism the money leaves the issuers before it arrives at the suppliers.
3The newest tanker trade is priced on a chokepoint the US-Iran deal would reopen
Givens' fourth pick was , and that thesis is entirely geographic: if the Strait of Hormuz stays fragile, crude goes the long way round Africa, voyages lengthen, and tanker owners get paid more to move the same barrels. His entry is above $40 with a stop near $36.50, and he attaches his own escape hatch — "trim fast if it doesn't run."
The day's dominant news runs straight at it. A US-Iran peace deal was the most-reported thread across Reuters and CNBC, with crude falling back toward pre-war levels and Tehran cleared to sell oil the moment a deal is signed. Reuters gives the trade partial cover — shipping angst still puts a floor under rates — but a floor is not the same input as an acute disruption premium, and the premium is what the tanker case is priced on. The test is narrow and close: whether Iranian barrels actually move, and whether routes normalise with them.
One counter-current sits inside the same story. doubled its buyback on the cash the war generated — the crude price is handing the premium back, the balance sheet is keeping it. That is a buyback going up in the same week as the cancellations in the first section, and it is a reminder that a lower oil price and a weaker oil sector are not the same claim.
What the sources recommended
Five buys, four of them out of a single eight-minute video, and each of those four conditional on a price trigger the stock had not yet crossed.
The build-out three. (Sterling Infrastructure) through $900, (Powell Industries) on a breakout above $312 with about 8% of downside risk, (Freeport-McMoRan) above $72 through four-month resistance with an 8% stop — all from Ross Givens, all chart-triggered, all from the video carrying the paid-membership funnel noted above.
The tanker. (Frontline) above $40 with a stop near $36.50, carrying Givens' own "trim fast if it doesn't run" and the peace-deal problem above unresolved.
The lightning round. (Cava Group), a buy from Jim Cramer on CNBC with no horizon and no reasoning given, which makes it the thinnest call on the page. The same segment flagged an insider purchase at without putting a direction on it.
Constan's supply work is not a sell list, and he is careful about that. He names the heavy issuers as the ones that trade down first but stops short of an exit on any of them; the thesis is thematic, and on his own three-to-six-month lag it is not built to be acted on name by name.
The oil half of the day gets tested this week; the supply half has a three-to-six-month fuse
The US-Iran story moves immediately: a memo to Congress, a $300bn fund already more than half committed, and Iranian oil cleared to sell on signature. That is the direct read on the tanker case — not whether a ceasefire holds, but whether barrels and routes normalise. Alongside it, Reuters had both stocks and gold steady ahead of new Fed chair Warsh's first verdict, and his guidance plus his handling of the balance sheet is the variable Constan calls the pivotal one for 2026. Earnings are light by comparison: and before the open, and after the close, with tomorrow (consensus $3.75) the closest thing to an AI-consulting read-through and ($1.64) the consumer one. None of them tests the supply thesis, which is the awkward thing about it — by construction it never has a bad day, only a bad year.
