Cathie Wood Is Doubling Down on This AI Stock During the Sell-Off

Cathie Wood’s ARK Innovation ETF (ARKK) added 100,000+ CoreWeave shares, including a $811.6k buy on July 8 and $2.0m on July 7, taking the stake to 1.6m shares (~$146m). Despite 1Q revenue growth of 114% to $2.1B and nearly $100B backlog, CoreWeave shares are down 23% since June 18 and fell another 14% after Meta flagged possible sale of excess computing power. The company remains loss-making (net loss $740m) while capex is extremely heavy ($6.8B in 1Q; guided $7B-$9B in 2Q; full-year capex raised to $31B-$35B) and debt has surged to ~$35B, making Wood’s “buy-the-dip” stance high-risk for average investors.

Analysis

The market is punishing the wrong variable if it thinks this is a pure demand story. For CRWV, the binding constraint is not backlog; it is the spread between contracted revenue and the cost of financing, powering, and refreshing an increasingly levered asset base. That makes equity value highly sensitive to utilization and refinancing terms, so even a modest slowdown in customer growth or pricing can translate into outsized multiple compression over the next 1-3 months.

The Meta angle matters more as a signal than as a single customer issue. If a hyperscaler can monetize surplus compute, the market will start questioning the scarcity premium embedded across the neocloud cohort, which is negative for CRWV and other GPU-rental models, but potentially constructive for META if it can self-fund more of its AI stack. Second-order beneficiaries remain the picks-and-shovels names tied to capex throughput, especially NVDA and networking/power infrastructure, though that support weakens if the market concludes the buildout is normalizing rather than accelerating.

Contrarian view: the selloff may be partially overdone because the backlog suggests near-term revenue visibility is still decent. The real debate is margin durability, not top-line growth. A clean falsifier would be either a confirmed customer expansion from Meta or a material step-down in capex guidance / debt funding pressure; absent that, the stock can stay weak simply because the market is discounting dilution risk and lower future contract quality.

Key Decisions for Investors

• Short CRWV on any bounce back toward the prior gap area; use a 1-3 month horizon and cover if management confirms no change in utilization or funding profile. Risk/reward is attractive because leverage and capex make the equity sensitive to small revisions in growth assumptions.
• If Meta confirms monetization of excess compute, initiate META/CRWV pair trade: long META, short CRWV. Thesis is that META can turn stranded infrastructure into cash flow while CRWV loses a perceived strategic anchor customer. Falsify if Meta explicitly recommits capacity internally or signs a long-term external contract.
• Long NVDA vs short CRWV as a cleaner way to own AI capex without the balance-sheet risk. NVDA captures spend regardless of which neocloud wins; CRWV is exposed to financing, utilization, and pricing compression. Use this as a 1-3 month relative-value trade.
• Avoid chasing CRWV until the next earnings update proves margin conversion. A revenue beat without capex moderation or leverage improvement is not enough; watch for any deterioration in guidance, bond spreads, or dilution language as the key downside catalyst.
• If you want optionality rather than outright short exposure, use CRWV put spreads for the next earnings cycle; the best-case outcome is limited if the stock stabilizes, while downside accelerates if financing or customer-concentration concerns intensify.