CoreWeave Offers So Much at $85

CoreWeave (CRWV) screens near $84.89 after nearly $70 off its May peak, despite delivering $2.58B Q2 revenue (+112% YoY) and ending the quarter with a $104B revenue backlog. The bullish case cites pricing gains (~25% across SKUs in July), $1.51B adjusted EBITDA (59% margin), and operating cash flow of +$679M, alongside raised 2026 capex guidance to $35–$39B. The primary risk is leverage and cash burn: Q2 free cash flow was -$5.74B, interest expense was $640M (guided $860–$940M in Q3), and liabilities of $72.05B versus $5.02B equity, with GAAP losses widening to -$626M and adjusted operating margin compressing to 5% from 16%. Overall, the article frames $85 as an attractive entry point contingent on backlog conversion, margin expansion into the low teens by Q4, and power scaling (1.5 GW now toward >1.85 GW by year-end).

Analysis

The equity story is no longer about demand; it is about funding the next leg of capacity. That shifts the risk from top-line growth to refinancing tolerance and execution on power, because a highly levered infrastructure platform can look cheap right up until the credit window narrows. If the market believes compute scarcity persists, the winners are the upstream picks-and-shovels: GPU vendors, power/cooling, and networking names, while hyperscalers like AMZN and GOOGL face some pricing pressure on specialized AI workloads but still retain the advantage of diversified cash flow.

Near term, the stock will trade like a credit-sensitive momentum name rather than a software multiple. The key catalyst over the next 1-3 months is whether margin expansion actually prints alongside capacity additions; if it does not, the market will haircut backlog more aggressively because contracted demand is only worth what the balance sheet can finance. A widening in high-yield spreads or any sign of customer concentration/renewal friction would likely matter more than another sell-side target change.

The contrarian read is that investors are probably underweighting pricing power but overweighting backlog quality. A 25% SKU repricing argues the market for AI compute is still tight, which supports the broader AI capex chain; however, backlog can re-rate downward fast if one or two large customers optimize away from external cloud capacity. Over 6-18 months, the thesis is either a rerating toward infrastructure utility-like economics or an equity dilution/refinancing overhang if capex keeps outrunning internally generated cash.

Key Decisions for Investors

• Long CRWV tactically on weakness, but size it as a financing-risk trade, not a secular compounder; best entry is after any post-event flush, with a stop if credit spreads or management guidance imply slower margin ramp.
• For a cleaner relative-value expression, pair long CRWV vs. short GOOGL over 3-6 months: thesis is specialized AI compute pricing and backlog scarcity versus hyperscaler bundling, with risk that GOOGL simply absorbs the demand through broader cloud relationships.
• If you want convexity instead of common equity, buy a 3-6 month CRWV bull call spread only if the next update confirms capacity additions and margin expansion; the reward is rerating, the loss is capped premium if execution slips.
• Do not short CRWV outright ahead of financing or quarterly updates; if bearish, use put spreads only after evidence of slower backlog conversion, weaker power buildout, or a widened credit market window.