1Thirty-year yields went vertical in four countries, and the recession signal underneath them is contested
The day's macro argument arrived second-hand and was sharper for being specific. John Arnold, speaking on Patrick O'Shaughnessy's podcast and relayed on The Compound's Animal Spirits, calls the long end of the global bond market a fire alarm: 30-year yields in the United States, the United Kingdom, France and Japan all rising steeply at the same time, on a fiscal and debt problem he argues equity prices are not discounting. Yahoo Finance's markets coverage carried the same fact in plainer form, as yields spiking. Read the chain before the claim, though: this is one manager's view, reported by a show that heard it somewhere else. Four countries in the evidence, one voice behind it.
The second half of that segment is where it stops running one way. The yield curve has un-inverted, and on the classic reading — the Harvey indicator — recessions arrive after the un-inversion rather than before it. The hosts put the obvious question to their own chart: whether this cycle is the first exception. They left it open, and nothing else in the day's material closes it.
The two halves point at different things, which is the useful part. A curve that un-inverts is a statement about the business cycle. A long end rising across four sovereign issuers at once is a statement about their balance sheets. If it is the second, then the reflex that weaker growth brings lower yields and a relieved equity market is the wrong map to be holding.
2Software is being priced as if disruption takes the revenue, when the record says it mostly takes the multiple
The most transferable idea of the day was a Sparkline Capital paper by Kai Wu, discussed on The Compound. Companies that were genuinely disrupted — Blockbuster, Borders, RadioShack, the newspaper groups — held their revenue and their market share for years after their share prices had collapsed. The market marked the terminal value to zero while the income statement was still working. came up as the other side of that trade, the disruptor rather than the disrupted. It matters now because the same mechanic is the cleanest description of what is happening to enterprise software under the fear that AI removes the seats it is sold by.
The tape offers a partial version of it. slipped 0.9% on an outlook that disappointed despite an earnings beat — the forward story moving the price where the quarter did not. Whether , the session's largest decline at 31.5%, belongs to the same story is not something the day's reporting establishes: the move is in the data, the reason is not.
The counter-case ran in the other video. is down about 56% from roughly $230 on exactly that fear, which takes it from about 60 times trailing earnings to something like 14–20 times forward while revenue still grows 22%. The mechanism offered against the bear case is pricing: around half of new contracts are consumption-based rather than per-seat, which breaks the link between the revenue line and the customer's headcount — and that link is the whole premise of the seats argument. Now Assist annual contract value has gone from zero to $750m against a $1.5bn target. served as the valuation comparison.
So the day held both readings of a single de-rating: research saying prices habitually write the future of a disrupted business to zero, and a pitch saying the assumption doing the writing is already breaking. The share of new contracts sold on consumption rather than seats is the number that decides between them.
3A trillion-dollar listing can enter the index in fifteen days, against a float too small to absorb the buying
SpaceX has filed its S1: $18.7bn of revenue, a $4.9bn net loss, and Starlink average revenue per user down from $99 to $66 as the subscriber mix shifts toward consumers. The Animal Spirits segment spent its time on the plumbing rather than the profit and loss, and the plumbing is where the claim sits. The NASDAQ-100 has a fast-entry rule that admits a new listing after 15 days, against the three to six months a seasoned listing normally waits. Weighting is modified-capitalisation and capped at three times free float, which pins the position at roughly 4–5% of the index. A small float in front of a large pool of buyers who have to own it is a squeeze by construction, and retail is expected to take a disproportionate share of the allocation.
The hosts were explicit that they are not worried about where the money comes from, with OpenAI and Anthropic each plausibly worth $1trn-plus and queued behind it. 's 100-times-sales listing came up as the cautionary comparison. appeared alongside as a musing about consolidating Musk's companies rather than as a call.
Hold the two facts together without resolving them: a business losing $4.9bn with falling revenue per user, and a mechanism that would set its price regardless of that. Index-driven demand does not read the S1. It does, however, stop — and the S1 is what is left when it does.
4Oil rebounded on a strike in the Gulf, and the equity bid stayed in semiconductors
A US strike on an Iranian military site, with tensions around the Strait of Hormuz alongside it, was the most heavily syndicated story of the day: Reuters carried it seven times through the Finnhub feed, with Yahoo Finance and Google News both on it as well. Oil rebounded. The equity coverage read the event as cautiously optimistic rather than defensive.
Where money actually moved was semiconductors. crossed a $1trn market value, up about 8% pre-market, with an endorsement from President Trump attached to the run. rose 5–8% on AI data-centre demand, with an analyst price target of $300 published against it. One CNBC item in the same feed had bears adding short exposure to small-caps ahead of the coming economic data — the opposite end of the risk curve from the semiconductor bid.
Crude rising on a supply risk while semiconductors lead the tape is a market treating the event as contained. The Strait is the line that would change that reading, and unlike the rest of the day's open questions it does not wait for a data release.
What the sources recommended
One buy call came out of the day's two videos, and it arrives inside a promotion for a paid course.
The buy is (ServiceNow), pitched on Felix & Friends on the de-rating described above, with a suggested entry around $110–120 and the presenter's own framing that this is early and higher-risk. The catalyst stack offered with it is specific: chief executive Bill McDermott buying $3m of stock and cancelling his scheduled selling, a $5bn buyback, a GSA federal agreement struck at a 70% discount, six members of Congress and President Trump (over $1m) among the disclosed holders, 85% of the Fortune 500 as customers and 98% retention.
The caveats belong with the call rather than in a footnote. The video is a promotion for a paid course; its title bills the company as "Trump's $100 billion defense stock," which ServiceNow is not — it sells enterprise-workflow software; and the pitch leans on a past call up 3,000% as its credential. The thesis is assembled from the bullish side of the evidence, with no bear case put against it. The underlying figures are checkable and the framing is not neutral; both are true at the same time.
The day's arguments run on very different clocks
The software question gets marked at every guidance print, and the session showed the reaction function plainly: up 13% on an earnings beat, down 12.5% on weak guidance, down on an outlook rather than a quarter. Guidance is setting prices; results are not.
The listing question has a date attached to it. The S1 is filed, and the fifteen-session window that decides the index question starts the day SpaceX trades, not the day it prices.
The yield question is the slowest of the three. An un-inverted curve is measured against recessions that followed it by quarters, so the "first exception" argument cannot be settled by anything in the next few sessions — which is exactly why the long end is worth watching in the meantime. One positioning note sits under all three: the day's reporting had bears building against small-caps ahead of the coming economic data.
