CoreWeave's Backlog Lawsuit Failure: A Comprehensive Explanation of the Reasons for Stock Revaluation

Brett Knoblauch of Cantor Fitzgerald expected CoreWeave to have approximately $40 billion in new backlog for the second quarter, anticipating a pace similar to the first quarter. What was actually announced was a revenue backlog of $104.2 billion as of June 30, up from $99.4 billion as of March 31 in the first quarter. This represents a net increase of $4.8 billion, and when considering the $2.58 billion recognized as revenue during the quarter, the gross increase amounts to about $7.4 billion, indicating that the growth driven by backlog conversion, which bullish investors had cited, fell short in numerical terms. Subsequently, CRWV shares closed at $107.73 on August 12, 2026 (Wednesday), rising 19.28% on that day alone.

This combination is quite rare and merits detailed analysis. Typically, when stock prices are re-evaluated by about 20% on the day of a major missed expectation, it does not occur. Understanding why this movement happened is crucial for deciphering how AI infrastructure-related stocks are currently being evaluated.

Backlog is the most critical indicator supporting the AI data center investment story because it is the only line that can convert future demand into concrete numbers. CoreWeave's backlog increased by 246% year-on-year, from $30.1 billion to $104.2 billion. At first glance, this figure appears overwhelmingly positive, but when looking at the quarterly growth, the picture changes. In the first quarter earnings announcement, it was officially stated that the backlog as of March 31 was $99.4 billion, marking the largest in the company's history.

Knoblauch's model estimated that the second quarter backlog would reach approximately $131 billion (the street's overall estimate was about $104.4 billion). The street's estimate was nearly accurate. However, Cantor's estimate of $131 billion deviated by about $27 billion from the actual results, leading to the analyst who emphasized backlog acceleration being the most embarrassed.

Calculating the continuous backlog growth, with a net increase of $4.8 billion and including the $2.58 billion recognized as revenue, the gross increase is about $7.4 billion. Although approximately $7.4 billion in new contracts were accumulated over three months, it fell far short of the anticipated $40 billion. In our preview released in early August, we pointed out the risks of backlog progress, and this actual result confirmed that risk.

What the actual quarterly results showed

Other items aside from the backlog were generally good, with some very strong content. Here, the differences in accounting standards become important. CoreWeave reports two types of losses, and combining different figures often leads to a mistaken designation of "missed expectations."

|----------------|-------------------|------------------------------------| | Metric | Q2 2026 Actual | Prior Estimate | | Revenue | $2.58 billion (up 112% YoY) | About $2.56 billion | | Adjusted Loss Per Share | $1.03 | $1.20 (LSEG Consensus) | | $128 million | About $66 million | | Adjusted EBITDA | $1.51 billion (59% margin) | Not individually aggregated | | Backlog as of June 30 | $104.2 billion | About $104.4 billion (Street), about $131 billion (Cantor) |

On an adjusted basis, the loss per share of $1.03 exceeded the adjusted consensus of $1.20 by 17 cents, which is a good result. On the other hand, the GAAP net loss of $626 million was mainly due to interest on debt associated with business expansion, and confusing GAAP values with adjusted values to argue "missed expectations" misrepresents the reality.

$25 billion not reflected in the backlog

And here the truth becomes clear. CoreWeave's second-quarter disclosure clearly states that "the $104.2 billion does not include over $25 billion in new customer commitments added in early Q3."

In just six weeks after June 30, contracts exceeding $25 billion were established, more than three times the announced figure (gross increase of $7.4 billion for the quarter). The backlog figure on the balance sheet is no longer keeping up, and the market has shifted its evaluation criteria from the "snapshot" at the time of the report to the "video" represented by the sign rate.

Physical capital investments also support this. At the end of the period, there were 51 data centers with 1.5 gigawatts of operational power, and contracted power increased from 3.7 gigawatts to 4.2 gigawatts as of August 11. The year-end active power target has also been revised upward to over 1.85 gigawatts. The ratio of contracts over 48 months has improved qualitatively from 10% to 21%.

Guidance has also been revised upward. The revenue outlook for Q3 is $3.45 to $3.6 billion, for the full year 2026 is $12.4 to $13.2 billion, and the end-of-period ARR (Annual Recurring Revenue) forecast is $18.5 to $19.5 billion.

Options market undervalued, stock price remains high

Before the earnings report, the ATM straddle expiring on August 14 priced in a 15.5% volatility (with an $85 strike: call $6.85 / put $6.35). This pricing was slightly lower than the average absolute volatility of the last four quarters (16.76%), indicating that a "normal" CRWV earnings reaction was expected, but it opened with a 19.28% rise. It is common for straddle buyers to be rewarded after earnings announcements for large AI stocks.

However, the true value lies in the subsequent movements. Checking the history on stockanalysis.com, on August 10 (Monday) it was $88.19 (-2.74%), on August 11 (Tuesday) it was $90.32 (+2.42%), on August 12 (Wednesday) it was $107.73, on August 13 (Thursday) it was $106.29 (-1.34%), and on August 14 (Friday) it was $105.26 (-0.97%). Although there was a profit-taking phase over two days, only 2.3% of the 19.28% rise was retraced.

A rapid rise that fills the gap is merely a liquidity event. If it maintains after two sessions post-surge, that indicates a fundamental re-evaluation (re-rating). The sustainability of that is what will determine its authenticity.

Moreover, there is unexpected backing from another side. On August 13 (Thursday), Knoblauch raised his target price from $167 to $176, maintaining an "overweight" rating. He continues to be bullish despite having his own predictions overturned. SanDisk also conducted a similar re-rating during its earnings briefing that week.

How the reading of backlog indicators has changed

Backlog is merely a snapshot at a certain point in time. In a market where billions of dollars can move with a single deal, there are often much larger economic news items that slip between the deadline and the actual disclosure date. Many traders considered the quarterly disclosure as "all information," but those who shorted based on backlog deceleration theory ended up losing by reading the correct numbers of a "window that has already closed."

However, this logic works both ways. Future commitments are merely "commitments" and not revenue. Furthermore, it is premised on continued funding from customers over multiple years and CoreWeave's quarterly expenditures of about $9 billion. There is also a high concentration risk among customers, and the debt incurred to cover massive capital investments turns adjusted profits into GAAP losses. Traders who benefited from the re-rating are not guaranteed safety in the following years.

This conflict runs throughout the entire AI infrastructure. The reverse pattern of Tesla's record sales earnings report followed by a sharp drop. The structure of NVIDIA's 2026 earnings, Micron's earnings preview, and comparisons of Marvell/Broadcom AI semiconductors also exhibit the same phenomenon.

Frequently Asked Questions

Did CoreWeave beat or miss earnings?
In terms of financial performance, it was a beat, but in the key metric (backlog) that bullish investors focused on, it missed. Revenue was $2.58 billion, exceeding the estimate of $2.56 billion, and the adjusted loss per share was $1.03, better than the consensus of $1.20. However, the gross increase in quarterly backlog was about $7.4 billion, far from Cantor's model of $40 billion.

Why did CRWV stock rise 19.28% despite missing backlog?
Because the quarterly earnings announcement disclosed that "new customer commitments exceeding $25 billion were contracted in early Q3," which were not included in the backlog calculation. The market responded to the recent sign rate rather than the static numbers as of June 30.

Is CoreWeave profitable?
Not profitable on a GAAP basis, with a net loss of $626 million in the second quarter. On the other hand, adjusted EBITDA was $1.51 billion with a 59% margin. This discrepancy is mainly due to interest burdens from financing data center construction.

What does the "implied volatility" of options indicate?
It indicates the "cost of earnings risk" and does not predict the price movements themselves. In CoreWeave's case, the implied volatility was 15.5%, while the actual movement was 19.28%, resulting in losses for option sellers. The important point is that it maintained its value over the next two days.

Who is the CEO of CoreWeave?
Michael Intrator. He is a co-founder and has led the transition from cryptocurrency mining infrastructure to AI cloud.

Summary
The main concern for future trading is whether the "high sign rate at the beginning of Q3 = stock price re-rating" will be reflected in the new backlog announced in September, or whether the $25 billion is merely a front-loading that will leave the next figures once again unsatisfactory. The answer will be revealed in the Q3 earnings report.