CoreWeave Just Booked $104 Billion of Future Revenue — and Burned $5.7 Billion in a Single Quarter

Revenue more than doubled and the contracted backlog reached $104 billion — but the same quarter consumed $5.7 billion of cash. $CRWV rose almost 18% pre-market anyway, and both halves of that sentence are the story.

There is a particular kind of company the market has never quite known how to price: one growing fast enough to justify almost any multiple, funded by debt large enough to threaten almost any outcome. CoreWeave is the purest example in the market today. On the evening of August 11, 2026, the AI cloud provider reported second-quarter results that beat on both lines and lifted full-year guidance — and $CRWV jumped roughly 18% in pre-market trading the following morning, from an August 11 close of $90.32 toward $106 a share. The results were covered in detail by CNBC and Bloomberg, with the full figures in the company’s own second-quarter release.

This profile walks through what CoreWeave reported, how a company that rents out graphics processors became a $49 billion business in under a decade, why the market rewarded a quarter that lost $626 million, and the quantifiable reasons for caution underneath a very good headline.

Company snapshot

• Company: CoreWeave, Inc.  ·  Ticker: CRWV (NASDAQ)
• Sector: Technology · AI cloud infrastructure
• Headquarters: Livingston, New Jersey · Founded 2017
• CEO: Michael Intrator (co-founder and chief executive)
• Co-founders: Michael Intrator, Brannin McBee, Brian Venturo — former commodities traders
• Origin: launched as Atlantic Crypto, an Ethereum mining operation; renamed CoreWeave in 2019
• Public since: March 2025, on the Nasdaq
• Approx. market cap: ~$49B  ·  Prior close: $90.32 (Aug 11, 2026)

Snapshot figures are widely reported company facts; the Q2 results and market data are sourced below.

What actually happened

CoreWeave reported second-quarter 2026 results after the close on August 11 and cleared the bar on the lines investors were watching. Revenue came in at $2.58 billion against roughly $2.56 billion expected, up 112% from $1.21 billion in the same quarter of 2025. On an adjusted basis the company lost $1.03 per share, materially better than the $1.20 loss the Street had modeled. Shares rose about 14% in after-hours trading on the print, then extended the move to roughly 18% by the following pre-market session.

But the number that did the real work was not on the income statement at all. CoreWeave’s revenue backlog — contracted future revenue it has signed but not yet delivered — stood at approximately $104 billion as of June 30, up 246% year over year. And that figure is already stale in the company’s favor: it excludes more than $25 billion of net new customer commitments signed early in the third quarter. For a business whose central question is “will the AI demand still be there in three years,” a backlog of that size is the most direct answer management can give.

Management also raised the full-year outlook. Full-year 2026 revenue guidance went to $12.4–$13.2 billion from a prior $12–$13 billion, and adjusted operating income guidance to $960 million–$1.15 billion from $900 million–$1.1 billion. Third-quarter revenue was guided to $3.4–$3.6 billion. The company also lifted its year-end active power target to more than 1.85 gigawatts.

The numbers that moved it

Sources: CoreWeave Q2 2026 results release, CNBC, Bloomberg, Seeking Alpha, StockTitan, 24/7 Wall St. Figures adjusted / non-GAAP where labeled.

Read that table twice, because it contains two opposite companies. The top half describes a business compounding at triple-digit rates with a 59% adjusted EBITDA margin. The bottom half describes a business that consumed $5.7 billion of free cash flow in ninety days and intends to spend up to $39 billion this year against roughly $13 billion of revenue. Both are accurate. Neither cancels the other out.

Six months of tape

The chart supplies the context the one-day pop hides. CRWV has traded between $60.55 and $153.20 over the past 52 weeks — a range in which the low is less than half the high. Even after an 18% gap, the stock sits roughly 30% below that high. This was not a breakout into new territory; it was a sharp recovery inside a very wide band. Buyers here are not buying a stock that has been quietly working, but one that has already round-tripped a great deal of value in both directions.

Inside the business: renting the scarcest thing in tech

CoreWeave’s origin story is one of the more improbable in the current AI cycle. It was founded in 2017 by three commodities traders — Michael Intrator, Brannin McBee and Brian Venturo — under the name Atlantic Crypto, mining Ethereum with racks of graphics processors. When the crypto market broke in 2018, the founders were left holding an enormous inventory of GPUs and no profitable use for them. In 2019 they renamed the company CoreWeave and began renting that hardware out as cloud infrastructure. The chips bought to mine coins turned out to be the exact chips the machine-learning boom would require.

What CoreWeave sells today is not a general-purpose cloud in the mold of AWS or Azure. It is a platform purpose-built for AI workloads: large, contiguous clusters of high-end accelerators, wired with the networking and storage that training and inference demand, sold to customers who need enormous capacity quickly and cannot wait in line at a hyperscaler. It competes on time-to-capacity and on scale of contiguous compute — both harder to supply than they sound, because both are ultimately constrained not by chips but by electricity.

That is why power, not revenue, is the operating metric management leads with. During the quarter CoreWeave expanded active power by nearly 500 megawatts to reach 1.5 gigawatts, against total contracted power of roughly 3.7 gigawatts. Active power is what can bill today; contracted power is the runway. The gap between them is the growth plan expressed in electricity.

The customer roster has broadened considerably. CoreWeave announced business with Anthropic and Meta during the quarter, and named new customers including Bentley Systems, Caterpillar, Grammarly and Isomorphic Labs, alongside expanded relationships with Cognition, Databricks and Hudson River Trading. That breadth matters for a reason covered below: CoreWeave has historically been extraordinarily dependent on a very small number of customers.

The quarter also brought capital and index milestones. The company raised more than $10 billion in unsecured debt and convertible bonds, closed a $3.1 billion delayed-draw term loan, received a $1 billion strategic investment from Jane Street, and was selected for inclusion in the Nasdaq-100 Index.

Why the market rewarded it

Three things stacked up. First, the beat was clean where the prior quarter had been messy. CoreWeave was punished in May on softer revenue guidance and a raised spending forecast, so the bar going into this print was set by fear, not enthusiasm. A double beat plus a guidance raise directly reversed that indictment.

Second, the backlog reframed the risk. The central bear argument against every AI infrastructure name is that today’s demand is a spending bubble that evaporates on contact with reality. A $104 billion contracted backlog, growing 246% and supplemented by $25 billion of fresh commitments weeks into the new quarter, is the most concrete available rebuttal. It does not prove customers will pay — contracts can be renegotiated and counterparties can weaken — but it converts a question of faith into a question of credit.

Third, operating leverage finally showed. Adjusted operating income of $128 million and adjusted EBITDA of $1.51 billion at a 59% margin suggest the fixed-cost base is starting to be absorbed by volume — the pivot the entire investment case depends on.

The case for caution

The most important number in this report is not the backlog. It is the capital expenditure guidance: $35–$39 billion for 2026, raised from $31–$35 billion, against full-year revenue guidance of $12.4–$13.2 billion. CoreWeave intends to spend roughly three dollars building for every dollar it collects this year. That is not a criticism in itself — it is the correct strategy if demand persists — but it defines the risk precisely. This company’s equity is a leveraged bet that AI compute demand holds long enough to fill capacity that is being financed today.

Because that spending is funded with borrowed money, the cost of the bet compounds. Free cash flow was negative $5.7 billion in the quarter. The net loss widened to $626 million from $290 million, and the widening was driven in large part by a surge in interest costs — the direct, mechanical consequence of financing a buildout at this speed. A business can survive negative free cash flow indefinitely so long as capital markets stay open to it. The risk is not that CoreWeave is unprofitable; it is that the terms on which it borrows could change faster than the revenue arrives.

Customer concentration is the second real exposure. Microsoft accounted for roughly two-thirds of CoreWeave’s 2025 revenue, and while the Anthropic, Meta and enterprise wins are genuinely diversifying that base, a book this concentrated means the health of a handful of counterparties is effectively the health of the company. There is a further wrinkle unique to this market: CoreWeave’s largest customers are also, increasingly, building their own AI infrastructure. Today’s anchor tenant can become tomorrow’s competitor.

Finally, positioning. Short interest sits near 12% of shares outstanding, which cuts both ways: it can accelerate a rally like this one, and it reflects a substantial cohort of investors who have studied the same balance sheet and reached the opposite conclusion. A move partly powered by short covering is not the same as a move powered by durable buying, and the two are difficult to distinguish in real time.

What to watch next

The immediate tell is whether CRWV holds the gap. Beyond the tape, three items matter. First, active power against that raised 1.85-gigawatt year-end target — power delivered on schedule is the leading indicator for revenue delivered on schedule. Second, the pace at which backlog converts into billed revenue, which is where the $104 billion either becomes real or stays theoretical. Third, financing terms on the next capital raise: with capex guided to $39 billion at the top end, CoreWeave will return to the capital markets, and the cost at which it does is the cleanest read available on how lenders assess this risk.

The bottom line

CoreWeave delivered the quarter the bulls needed: revenue more than doubled, the loss came in narrower than feared, operating leverage appeared meaningfully for the first time, and management raised the full-year outlook while pointing to a $104 billion backlog that grew 246%. An 18% pre-market move on a $49 billion company is the market taking that seriously. But the risk here does not sit quietly in a footnote. The same company is spending three times its revenue on construction, burned $5.7 billion of cash in ninety days, funds itself with debt whose interest costs are already widening the loss, and derives a majority of its revenue from a small group of customers who are also potential rivals. That is not a reason to dismiss it — the demand appears real and the contracts are signed — but it is a reason to size it honestly. CRWV is a bet on the durability of the AI buildout, expressed with leverage. Anyone buying it after an 18% gap should be clear that they are underwriting that bet, not avoiding it.

This report is for information purposes only, and is neither a solicitation or recommendation to buy nor an offer to sell securities. TopStockPicks.co is not a registered investment advisor and is not a broker-dealer. TopStockPicks.co has NOT BEEN COMPENSATED for coverage of CRWV. TopStockPicks.co and its affiliates or officers may buy and sell shares of CRWV in the open market at any time without notice. TopStockPicks.co does not set price targets on securities. Always do your own due diligence and consult your financial advisor. Never invest into a stock discussed by TopStockPicks.co unless you can afford to lose your entire investment.