CoreWeave Owes $21 Billion. Its Customers Are Its Biggest Risk.

CoreWeave carries $21.0B of debt at an average ~11% interest rate, meaning interest expense consumes roughly 25% of last year’s $5.1B revenue. The company has a $66.8B backlog (including >$20B from OpenAI), is guiding to >$12B revenue for 2026, but lost nearly $1.2B last year and derives ~70% of revenue from Microsoft, creating concentration and competitive risk as hyperscalers may internalize AI compute. OpenAI’s projected $115B cash burn through 2029 and potential Fed rate hikes that would raise borrowing costs further amplify downside risk.

Analysis

CoreWeave’s capital structure is the effective lever on its operational risk: heavy fixed-financing obligations make small revenue shocks non-linear for equity and credit. That amplifies timing risk — a quarter or two of customer re-pricing or slower utilization can convert an earnings miss into covenant stress, and secondary effects (forced asset sales, distressed GPU liquidations) would further compress recoveries for creditors.

Concentration among a handful of hyperscalers creates asymmetric optionality for those customers: they can buy time by outsourcing now and internalize later with software+chip stack investments, converting a vendor relationship into forfeited backlog. The economics of internalization create a durable two-tier market — a small set of captive hyperscaler demand at strategic pricing and a larger elastic market where pricing power belongs to suppliers of integrated solutions (cloud providers and vertically integrated chip companies).

Catalyst map and reversal paths are clear. Near term (weeks–months) look for covenant language, quarterly utilization trends, and any non-dilutive funding announcements; medium term (6–24 months) watch large counterparty capex cycles and corporate funding rounds that could absorb excess demand or force reorgs. The asymmetric payoff: failure to refinance or a significant customer pullback likely produces >50% equity downside, while the upside is limited absent a strategic sale or a transformational margin recovery tied to diversified end markets.

Key Decisions for Investors

• Short CRWV equity sized 1–2% of portfolio notional, horizon 3–12 months; stop-loss at 20% adverse move; target asymmetric payoff of ~3:1 — primary risk is a strategic capital raise or acquisition that rerates the name.
• Buy 3–5 year protection on CRWV credit (CDS or senior bond puts) sized to cover 2–3% portfolio exposure, horizon 6–24 months; expect premium outlay ~5–10% of notional for meaningful tail protection — reward is concentrated in a distressed recovery scenario where bond prices fall >30–50%.
• Pair trade: long NVDA (or long-dated NVDA LEAPs 12–24 months) / short CRWV equity, 1:1 delta-adjusted, horizon 6–18 months. Rationale: NVDA benefits if CoreWeave must liquidate GPU inventory or if demand consolidates around chip incumbents; risk is broad market derisking that hits both.
• Hedge operational tail: buy protection on MSFT exposure tactically (e.g., buy puts or underweight) only if MSFT signals accelerated internalization timelines in earnings — timeframe 3–12 months; this limits residual counterparty contagion risk to the portfolio.