CoreWeave Is Down Around 30% in 6 Months. Has It Become a Steal of a Deal?

CoreWeave, which went public in 2025 and rents GPU capacity for AI workloads, reported revenue growth that more than doubled and a revenue backlog that nearly doubled to over $55 billion, anchored by a multi‑year Meta agreement worth more than $14 billion. Despite strong top‑line momentum, the company remains unprofitable—net losses of $715.3 million in the first nine months of 2025 (vs. $812.1 million prior year) and $841.4 million in net interest expense over the past three quarters versus only $43.6 million of operating profit—contributing to a drop in market cap from a peak near $88 billion to about $45 billion and a ~30% decline in the past six months, underscoring capital‑intensity and valuation risk for investors.

Analysis

Market structure: Hyperscalers (META, AMZN, GOOGL) and Nvidia (NVDA) are net beneficiaries as they capture pricing power and secure capacity; small/mid-cap GPU-rentals like CRWV face margin pressure because capital intensity and high interest expense (CRWV reported $841M net interest over three quarters) compress returns. Supply/demand remains tight for cutting-edge GPUs (backlog >$55B implies demand > supply today), supporting NVDA pricing but straining less-capitalized renters that must finance capacity at floating rates. Cross-asset: stressed credit for CRWV should widen its CDS/spreads and lift idiosyncratic equity volatility; a macro rate move up would further punish levered infra names while supporting USD and pressuring EM risk assets.

Risk assessment: Tail risks include a sharp rise in US rates (+100bp over 90 days) that blows out CRWV interest expense, an Nvidia supply ramp causing GPU price deflation >20% within 12 months, or a regulatory export shock constraining GPU shipments from NVDA. Immediate (days) risk is elevated IV and headline-driven swings; short-term (weeks/months) depends on quarterly backlog conversion and refinancing; long-term (quarters/years) hinges on CRWV achieving positive free cash flow or securing low-cost capital. Hidden dependency: customer concentration (multi-year Meta deal >$14B) creates counterparty and negotiation risk if Meta internalizes capacity.

Trade implications: Direct: initiate a modest short of CRWV (2% portfolio) via 3–6 month put spread to cap downside, funded by a 2–4% long in NVDA or META for lower leverage risk; pair trade: long NVDA vs short CRWV (1:1 notional) to express spread between equipment owner and renter economics. Options: buy CRWV 6-month puts or put spreads, sell 3–6 month covered calls on NVDA/META to finance; exit/roll if CRWV announces refinancing or interest expense cut >50% in next two quarters. Sector rotation: reduce small-cap AI-infra exposure by 5–10% over 30 days and reallocate to NVDA, large cloud (META, AMZN) and diversified semis.

Contrarian angles: Consensus underweights the strategic value of CRWV’s multi-year backlog and customer relationships — if GPU scarcity persists and CRWV converts backlog at high gross margins, upside is meaningful. The market may be overpricing financing risk: a successful refinancing or equity raise could halve net interest burden and trigger >50% rerating. Historical parallel: early cloud infrastructure providers traded poorly during capex cycles before consolidating into profitable scale players; unintended consequence of a large short position is a squeeze if META or a hyperscaler opts for M&A to secure capacity.

Key Decisions for Investors

• Establish a 2% portfolio short position in CRWV using a 3–6 month bear put spread (buy 1 ITM put / sell lower-strike put) sized to limit max loss to ~2–3% of portfolio; trim/close if CRWV announces refinancing that reduces quarterly net interest expense by >50% or if share price rallies >30% from entry.
• Fund the above by allocating 2–4% long to NVDA or 2–3% long to META (buy-and-hold 3–6 months) to capture durable demand exposure with stronger balance sheets; rebalance if NVDA/CRWV relative outperformance exceeds 20% in 3 months.
• Deploy an options yield strategy: buy 6-month CRWV puts or put spreads (protective downside) and sell 3–6 month covered calls on NVDA/META to offset cost; target net premium ~0–1% portfolio and maintain defined-risk profiles.
• Reduce aggregate exposure to small-cap AI-infrastructure names by 5–10% within 30 days and reallocate to large-cap semis/cloud (NVDA, META, AMZN) to lower funnel leverage and interest-rate sensitivity.
• Set monitoring triggers: add to long CRWV only if quarterly net interest expense falls below $200M or interest coverage (EBITDA/interest) >1 for two consecutive quarters, or if a hyperscaler announces an acquisition of CRWV within 6 months.