CoreWeave's Flagged Tension Resolved the Other Way, This Round

What the dossier said to watch

CoreWeave’s dossier named a specific tension after the Q1 2026 print: a $99.4 billion backlog sitting next to a light Q2 forecast, revenue up 112% year over year in the same quarter shares still fell roughly 10% on. The dossier called it “the tension to track quarter over quarter” — whether a growing backlog of contracted-but-unbilled revenue would eventually translate into guidance the market believed, or whether investors would keep discounting it every time forward guidance came in soft.

The Q2 print, reported August 11, answered that question, and it answered the other way from Q1. CoreWeave reported revenue up 112% year over year to $2.58 billion — the identical growth rate as the quarter before — against a $104 billion revenue backlog, up from $99.4 billion, and active contracted power approaching 500 additional megawatts to reach 1.5 gigawatts. Shares jumped more than 9% after hours, then more than 20% the next session.

Same growth rate, opposite market reaction

The reversal is the finding here, not the revenue number, because the revenue growth rate did not change between quarters. What changed was the forward signal attached to it: Q1’s light guidance told the market the backlog might not convert on schedule; Q2’s backlog growth and power buildout told the market the opposite, that contracted capacity is expanding faster than it’s being questioned. A backlog that grows from $99.4 billion to $104 billion in one quarter, alongside a near-500-megawatt jump in active power, is the kind of forward evidence that makes a 112% revenue print read as durable rather than borrowed against a shakier future quarter.

This doesn’t resolve the dossier’s underlying question so much as it answers one round of it. The backlog-to-guidance gap the dossier flagged is a recurring test, not a one-time one — CoreWeave will report another quarter with another guidance number, and the same tension between contracted power and delivered revenue applies every time. What Q2 shows is that the market’s read on that gap can swing 30 points of stock reaction between quarters on the same underlying growth rate, which says more about how thin the market’s confidence in the backlog actually is than about the backlog itself.

That thinness has a history behind it, and the history is why a guidance signal moves this stock as hard as it does. CoreWeave’s dossier tracks a run of volatility over the prior year: a $9 billion bid for data-center operator Core Scientific that shareholders voted down in October 2025, a 16.3% single-session drop in November on a disclosed construction delay at a data-center provider, a $33 billion market-value loss over six weeks that December compounded by short-seller Jim Chanos’s public bet against the stock, and founders who have sold $2.3 billion of stock since the post-IPO lockup lifted. None of those four events is the backlog-to-guidance question directly — but each one taught the market to treat CoreWeave’s forward signals as unreliable, which is exactly the discount Q1’s light guidance triggered and Q2’s backlog growth partially reversed. A stock the market already doubts swings harder on the same news than one it trusts by default; that discount, more than the backlog number itself, is the likeliest reason the same 112% growth rate produced a 10% drop in one quarter and a 20% pop in the next. CoreWeave’s own dossier names a second, unresolved watch item that this print doesn’t touch: the S-1’s disclosed 62%/77% Microsoft-led customer concentration, the number the dossier says to hold against future filings. Neither of the two wire reports on this print discloses an updated concentration figure. A bigger backlog from a customer base this concentrated is not the same evidence as a bigger backlog spread across a diversified one, and this print doesn’t distinguish between them.

Sources