CoreWeave Q2 Earnings: $99B Backlog on the Line
CoreWeave reports Q2 Tuesday after close. With shares down 44% from highs and a $99.4B backlog at stake, this is the week's binary trade.
Key Points
- CoreWeave is expected to post Q2 revenue of $2.56 billion — up 110.7% year-over-year — while widening its net loss to $1.42 per share from $0.60 a year ago.
- A $99.4 billion contracted backlog and a CapEx program ballooning to $31–$35 billion for 2026 have turned Tuesday's print into a direct referendum on whether AI infrastructure spending is generating returns fast enough to justify the losses.
- Watch the backlog update number above everything else: Cantor Fitzgerald's Brett Knoblauch sees nearly $40 billion in new Q2 contracts as the line between a relief rally and another leg down toward $70.
CoreWeave enters Tuesday's Q2 earnings report trading near $85 — 44% below its 52-week high of $153.20 — and more than 50% off peak, making this the highest-stakes binary print in the AI infrastructure sector this week. Street consensus calls for $2.56 billion in revenue, up 110.7% year-over-year, alongside a net loss of $1.42 per share that is more than double the $0.60 loss posted in the year-ago quarter. The stock's collapse into the print has eliminated any cushion; either the backlog number buys management credibility or the selloff accelerates.
The CapEx Problem Inside the Growth Story
The surface numbers at CoreWeave look like a high-growth machine. Revenue doubling year-over-year, adjusted EBITDA margins running at 56% in Q1 2026, and a contracted backlog of $99.4 billion that most SaaS companies would need a decade to accumulate — on paper, this is what AI infrastructure monetization is supposed to look like. The problem lives two lines below: CapEx guidance for 2026 was raised to $31–$35 billion due to component pricing pressure, a figure that dwarfs the company's revenue base and is growing faster than the top line.
In Q1 2026, the EPS of negative $1.40 missed the consensus estimate of negative $0.91 by 53.85%, and the net loss widened from $315 million to $740 million in a single quarter. Those are not the kind of misses that get absorbed quietly when a stock is already trading at a fraction of its IPO-era enthusiasm. The Q2 loss estimate of $1.42 per share represents a further deterioration, and if the actual print lands meaningfully worse — as it did in Q1 — the $85 floor becomes a ceiling.
The input cost story is not isolated to CoreWeave's balance sheet. Tesla CEO Elon Musk called memory pricing "insane," and Amazon CEO Andy Jassy explicitly cited "inflated" memory chip costs as a driver of his company's own elevated capex guidance. That structural headwind — a small set of memory vendors supplying insatiable AI processor demand — flows directly into CoreWeave's cost structure, making the CapEx blowout less a one-time shock and more a permanent feature of the business model for the next several quarters. Traders pricing a mean-reversion trade in CRWV need to account for the fact that the cost environment has not improved materially since Q1.
The Bull Case Rests Entirely on Backlog Velocity
Cantor Fitzgerald's Brett Knoblauch is the most visible bull on the street, carrying a Buy rating and a $167 price target — implying 95% upside from current levels. His conviction is not sentimental. The thesis rests on a specific, measurable catalyst: he expects CoreWeave to beat Q2 backlog estimates by a wide margin, and he sees the company on track to add approximately $40 billion in new contracts during the quarter, consistent with the pace established in Q1. If that number lands at or above $40 billion, the total contracted backlog grows toward $135–$140 billion, giving the revenue model a multi-year visibility argument that justifies tolerating near-term losses.
The logic is straightforward. A company with $135 billion in committed future revenue, even at a significant discount rate in a 4.69% 10-year Treasury environment, is not priced like a speculative bet at an $85 stock price — it is priced like a distressed asset. Bulls argue the market is systematically undervaluing the visibility embedded in long-term AI cloud contracts, treating CRWV as if backlog were a fiction rather than a hard commitment from enterprises that have bet their own AI roadmaps on CoreWeave's infrastructure. The counterargument is execution: getting new chips online at scale, managing power constraints across data centers, and translating contracted revenue into actual delivered compute without slippage.
The power problem is real and specific. Data center capacity expansion has been constrained by delays in both chip delivery timelines and utility-grade power connectivity at new sites. CoreWeave acknowledged in Q1 that power limits were a binding constraint on its ability to convert capital spending into revenue-generating capacity. If Tuesday's call surfaces evidence that those constraints are easing — new sites going live ahead of schedule, power agreements signed, Blackwell-era GPU deployments accelerating — the stock has a legitimate case for a violent short-squeeze. Short interest in CRWV has built substantially during the 50% drawdown, and a backlog beat plus operational progress is exactly the kind of data combination that triggers forced covering.
What the Print Actually Needs to Deliver
Strip away the analyst targets and the binary framing reduces to four specific data points management must address on Tuesday's call. First, the Q2 backlog number: anything at or above $135 billion total is a beat, anything below $130 billion is a miss with implications for the demand narrative. Second, the CapEx trajectory: if management narrows or holds the $31–$35 billion full-year range, that is neutral to positive; an upward revision to that range confirms that the cost structure is still not under control and removes the margin expansion story for 2026. Third, the power and capacity update: specific megawatt figures, new data center go-live dates, and GPU rack utilization rates are the operational metrics that turn abstract backlog into deliverable revenue. Fourth, Q3 revenue guidance: consensus is modeling continued triple-digit growth, and any guidance that implies a deceleration below 90% year-over-year growth will be read as the beginning of a normalization that the valuation cannot yet support.
There is a reasonable scenario in which CoreWeave beats on revenue, reports a wider-than-expected loss, updates backlog above $135 billion, and holds CapEx guidance — and the stock still sells off, because the market resets the EPS path lower. That is the trap in buying into the print: the revenue upside is priced in by the bull community, but the loss widening risk is not fully discounted. Conversely, a clean beat on all four metrics in a stock with this much short interest could produce a 20–30% single-session move, which is precisely why this is the most asymmetric setup of the week.
Watch the $75 level as the next technical support if the print disappoints — that corresponds to the pre-earnings range from late May and represents approximately a 12% additional drawdown from current levels. On the upside, a backlog-driven relief rally has room to $105–$110 before encountering meaningful resistance from the distribution pattern built during the May–June decline. The AMAT print Thursday will then confirm or deny whether the semiconductor equipment cycle is still in a build phase that can sustain CoreWeave's infrastructure ambitions through year-end.
The Weekly Investor
Daily market analysis for active traders. Free.
