CoreWeave does not need to beat the Street on Tuesday, because CoreWeave already told the Street it would not. The company guided second-quarter revenue to $2.45bn–$2.6bn, a midpoint of $2.53bn that sits roughly $160m below the $2.69bn LSEG consensus that was standing when the guide was issued. That is not a company sandbagging into a beat; that is a company telling you the top line is capacity-constrained. So the number that matters on 11 August is not revenue. It is the contracted revenue backlog, which finished the first quarter at $99.4bn, and whether the second quarter added enough new commitments to keep the compounding story intact. CoreWeave shares traded at $92.19 on the morning of 10 August, and the options market is pricing a move of roughly 15% in either direction.
Here is the part almost nobody is writing about: the derivatives market and the sell-side are positioned in opposite directions, and you can measure the gap precisely. Pull the 14 August option chain and the at-the-money $92 strike shows a call mid-price of $6.15 against a put mid-price of $7.33. Same strike, same expiry, and the puts are bid roughly 19% over the calls. That is a market paying a premium for downside protection into the print. Now look at the sell-side: of 14 analysts tracked by Visible Alpha, 11 rate CRWV a Buy with a mean target near $147, more than 60% above the current quote. One of these two groups is going to be badly wrong, and the options market is the one with money at risk over a four-day horizon. Having tracked the AI-infrastructure complex through the Situational Awareness unwind in early August, I have learned to weight the four-day money over the twelve-month opinion.
Key facts before the print
- $92.19 — CRWV share price, 10 August 2026, 05:27 ET (Nasdaq)
- ±14.6% — options-implied move, from the $13.48 at-the-money straddle on the 14 August expiry (Nasdaq option chain, 10 August 2026)
- $2.45bn–$2.6bn — company guidance for Q2 revenue, midpoint $2.53bn, versus $2.69bn LSEG consensus (CNBC, 7 May 2026)
- $99.4bn — contracted revenue backlog at the end of Q1 2026, up nearly 50% sequentially and close to 4x year over year (CoreWeave Q1 2026 earnings call, 7 May 2026)
- –$1.22 — consensus Q2 loss per share, against a $0.27 loss a year earlier (Yahoo Finance / Zacks)
- ~$147 — mean analyst price target across 14 analysts, 11 of whom rate the stock Buy (Visible Alpha data)
- $60.82–$137.98 — CRWV closing range over the past six months (Nasdaq historical data, 10 February to 7 August 2026)
What the options market is actually saying
Start with the arithmetic, because the headline numbers in this article are derived rather than borrowed. On the 14 August expiry, the first expiry that captures Tuesday evening’s release, the $92 call was quoted $6.05 bid and $6.25 offered, a $6.15 mid. The $92 put was $7.15 bid and $7.50 offered, a $7.33 mid. Add them and the at-the-money straddle costs $13.48 against a $92.19 share price. That is a 14.6% implied move by Friday’s close.
Apply it symmetrically and you get the two numbers in the headline. The bull case resolves near $106. The bear case resolves near $79. Those are not targets in the analyst sense; they are the boundaries the option market is charging you to cross, and roughly a third of the time the realised move lands outside them.
The asymmetry inside that straddle is more informative than its width. At the identical strike and expiry, puts cost $1.18 more than calls. In a market with no directional view, those two prices would sit far closer together. The skew tells you that hedgers are paying up, and it is consistent with what the equity has already done: CRWV has fallen from a $137.98 close in May to $92.19, a drawdown of roughly a third, while the sell-side target has stayed near $147.
There is a second reading worth holding. An implied move of 14.6% on a company with a $99.4bn backlog is the market admitting it cannot forecast the quarter from the outside. Backlog conversion, depreciation schedules on GPU fleets and the timing of when contracted power actually energises are not modellable from public data. That uncertainty is precisely what you are paying for in the straddle.
What CoreWeave has actually built, and what it costs
CoreWeave rents GPU compute at hyperscale. The financial shape of that business is closer to a data-centre REIT than to a software company: enormous up-front capital, long contracted revenue, and a depreciation line that determines whether growth ever becomes profit. The market keeps pricing it on software-style growth multiples while it is funded like infrastructure, and that mismatch is the source of most of the volatility in the shares.
The first quarter was, on the company’s own telling, transformational. “We delivered our strongest quarter for customer bookings, signing more than $40 billion of new commitments and growing contracted revenue backlog to nearly $100 billion,” chief executive Mike Intrator said on the 7 May call. Chief financial officer Nitin Agrawal put the backlog at $99.4bn, up “nearly 50% sequentially” and “close to 4x year-over-year.” Intrator’s framing was blunter still: “We have reached hyperscale.”
That is the bull case stated by the people who built it, and it is not empty. A backlog approaching $100bn against roughly $2.5bn of quarterly revenue implies close to a decade of contracted work, assuming the counterparties pay and the capacity arrives. Estimates circulating ahead of the print put remaining performance obligations near $107bn as of 30 June, which would imply roughly $10bn of net new bookings in the second quarter.
Hold that $10bn figure, because it is the single cleanest test of the thesis. Q1 added more than $40bn. If Q2 adds $10bn, bookings decelerated by roughly three quarters in a single quarter. A bull will correctly say $40bn was inflated by one enormous contract and that $10bn on top of $99bn is still extraordinary. A bear will correctly say the second derivative just turned hard negative, and that a company valued on compounding backlog cannot afford a decelerating one. Both readings are defensible from the same number, which is why the straddle is priced where it is. FinanceFeeds has tracked how quickly this complex re-rates in the Situational Awareness fire sale and its rebound.
The customer concentration problem nobody wants to price
Every contracted-backlog business carries the same hidden risk, and it is not demand. It is counterparty concentration. CoreWeave’s backlog is dominated by a small number of very large customers, principally Microsoft and OpenAI. That is a strength while the AI capital cycle runs hot, because it means the revenue is investment-grade in character. It becomes the entire risk the moment one of those counterparties slows its own build.
The cross-industry parallel that clarifies this is not from technology at all. It is from commercial property. A landlord with a single anchor tenant on a fifteen-year lease looks extraordinarily safe on paper, right up to the point the anchor renegotiates. The lease is an asset; the concentration is a liability; and the market never prices both at once. It prices the lease during the expansion and the concentration during the contraction. CoreWeave has spent 2026 being repriced from the first frame to the second, which is most of the explanation for a stock that fell from $138 to $92 while its backlog grew.
This is also where the industry response matters. The AI-infrastructure trade has visibly broadened out beyond the pure GPU renters: FinanceFeeds has covered how the AI trade is leaking into unglamorous industrial names, and the power-and-cooling suppliers underneath the buildout, such as Vertiv, now trade as a distinct expression of the same demand. Investors who want AI capital-expenditure exposure without single-tenant risk have alternatives they did not have eighteen months ago, and that competition for the same dollar is part of why CRWV has de-rated.
The physical constraint deserves a mention too, because it is the one thing capital cannot solve quickly. Contracted power is not delivered power. Local resistance to new sites is real and rising, as FinanceFeeds documented when Nashville chose to pay $37m rather than let a data centre open. A backlog denominated in gigawatts is only worth what the grid and the planning process actually permit.
Bull case versus bear case
| Bull case — resolves toward $106 | Bear case — resolves toward $79 | |
|---|---|---|
| Backlog | RPO confirmed near or above $107bn, implying ~$10bn net new bookings on top of an already vast base | Any print that shows backlog flat or down exposes the deceleration from Q1’s $40bn of commitments |
| Revenue | Delivery above the $2.6bn top of guidance signals capacity energising faster than modelled | Landing at the $2.45bn low end confirms the constraint is real and pushes revenue right |
| Margins | Depreciation and interest scale slower than revenue; loss per share better than the –$1.22 consensus | Loss lands nearer the –$1.40 tail of the range as financing and depreciation compound |
| Concentration | New named counterparties dilute Microsoft and OpenAI exposure | Backlog growth comes from the same two names, deepening the single-tenant problem |
| Guidance | Q3 guided at or above consensus, ending the pattern of guiding below the Street | A second consecutive below-consensus guide reframes the story as structurally supply-capped |
Financing, disclosure and the regulatory edge
The regulatory dimension here is less about a rulebook and more about disclosure quality and the cost of money, and both cut against a highly levered builder.
Start with disclosure. Remaining performance obligations are an accounting construct with real latitude in them. The revenue-recognition standard governs when a contract enters the backlog and how much of it counts, and it permits meaningful judgement over cancellable terms, ramp schedules and options to extend. A $99.4bn number is not a bank balance. Investors are entitled to ask what proportion is non-cancellable, how it is weighted across years, and how much depends on capacity that is contracted but not yet energised. The companies that answer those questions in detail tend to earn a durable premium; the ones that answer them thinly get repriced on the first disappointment.
Then the cost of money, which is where the macro backdrop bites. This is a business that borrows to build and repays out of contracted cash flows, so the discount rate applied to a decade of backlog is not an abstraction. It is the valuation. That makes CoreWeave unusually sensitive to the front end of the curve, and the front end has been moving: the July payrolls print came in negative, which reopened the debate about how quickly the Federal Reserve moves in September. Falling rates would be a straightforward tailwind for the equity. A rate path that stays higher because inflation proves sticky compresses exactly the long-duration cash flows this stock is priced on.
Export controls sit behind both. CoreWeave’s fleet is Nvidia silicon, so the rules governing where advanced accelerators may be sold and deployed are an input to its supply, its expansion geography and ultimately its cost base. Nothing in the second-quarter release will resolve that, but any commentary on chip allocation and delivery schedules is a live variable rather than boilerplate.
What happens next
Three concrete expectations, with the reasoning attached.
First, revenue will matter less than the market thinks and backlog will matter more. CoreWeave has already guided the top line below consensus, so a revenue miss against the Street is close to pre-announced and largely discounted. The variable with the power to move the stock 15% is the RPO line and whatever management says about the pace of new commitments. Watch the bookings number before the revenue number.
Second, the put skew is a better guide to the near term than the $147 mean target. Options positioned into a four-day event are a cleaner read on immediate risk than twelve-month targets that have not been marked down as the stock halved. If the print is merely adequate, the most likely outcome is a move smaller than the 14.6% implied, and the premium sellers win. That is the base case precisely because the market has already de-rated the shares by a third into it.
Third, the guidance for the third quarter is the real event. A second consecutive guide below consensus would tell investors the constraint is structural rather than a timing artefact, and would justify a materially lower multiple on the same backlog. A guide at or above consensus would be the first evidence in two quarters that capacity is catching up with demand, and it is the single most plausible route back toward the sell-side’s target range.
The wider question this print helps answer is whether the AI capital cycle is decelerating or merely digesting. CoreWeave, reporting a day before Nebius, is the first of two clean reads on that question in the same week. Two decelerating prints would be a signal. One would be a data point.
Frequently asked questions
When does CoreWeave report second-quarter results?
CoreWeave reports after the US close on Tuesday 11 August 2026, with the conference call scheduled for 5:00pm Eastern Time. The first options expiry that captures the release is Friday 14 August, which is the contract used to derive the implied move in this article.
What is the options-implied move for CRWV?
Roughly 14.6% in either direction. The at-the-money $92 straddle on the 14 August expiry cost about $13.48 against a $92.19 share price on the morning of 10 August. That frames a bull resolution near $106 and a bear resolution near $79 by Friday’s close.
Why is the consensus price target so much higher than the implied range?
They measure different things over different horizons. The roughly $147 mean target is a twelve-month view from 14 analysts, 11 of whom rate the stock Buy. The straddle is a four-day view priced by traders with capital at risk over the event. The two can diverge for a long time, and at present they diverge sharply.
What is CoreWeave’s revenue backlog?
Contracted revenue backlog finished the first quarter at $99.4bn, up nearly 50% sequentially and close to four times the prior year, after more than $40bn of new commitments were signed in the quarter. Estimates ahead of the second-quarter print put remaining performance obligations near $107bn as of 30 June.
What is the single biggest risk in the CRWV story?
Customer concentration. The backlog is dominated by a small number of very large counterparties, principally Microsoft and OpenAI. That makes the contracted revenue high quality while the AI capital cycle runs hot, and makes it the dominant risk if either counterparty slows its own build.
Is the guidance shortfall against consensus a red flag?
Not by itself. Guiding revenue to a $2.53bn midpoint against a $2.69bn consensus signals a supply constraint rather than a demand problem, which is a materially different diagnosis. It becomes a red flag only if the third-quarter guide repeats the pattern, because that would indicate the constraint is structural.
This article is informational analysis and is not investment advice. Prices, option quotes and implied moves were captured on 10 August 2026 and move continuously. Figures attributed to consensus are third-party estimates and are not company guidance. Always verify current market data before making any investment decision.
