CoreWeave's Interest Expense Hit $640 Million Last Quarter, 2.4 Times What It Was a Year Ago

CoreWeave’s interest expense surged to $640M in Q2 (2.4x $267M a year earlier) even as management cut its weighted-average cost of debt by ~300 bps, highlighting balance-sheet growth as the driver. Q2 revenue grew 112% to about $2.6B with a ~$104B backlog, but guidance implies Q3 interest expense of $860M–$940M (~41% higher at the midpoint) versus $200M–$260M adjusted operating income. With capex guided to $35B–$39B full-year (about $19B–$23B more in 2H) against only ~$5.5B cash, the interest/operating-profit gap appears set to widen amid a still-high bond-rate backdrop (30Y Treasury yield ~5.27%).

Analysis

CoreWeave is moving from a growth story to a financing spread trade: equity upside now depends less on backlog quality and more on whether incremental dollars of debt can still earn an IRR above an ~8% blended funding cost. That is a materially tougher hurdle when the company is still in heavy build mode, because every extra turn of leverage pushes the equity farther down the capital structure and makes the common stock behave more like a levered call on future refinancing conditions than on revenue growth alone.

The near-term beneficiary is not necessarily the stock; it is the hardware ecosystem. Continued capex supports demand for NVIDIA accelerators and adjacent networking/storage vendors in the next 1-2 quarters, but a financing squeeze would eventually slow order velocity or shift share toward hyperscalers that can self-fund capex from cash flow. The second-order loser is any smaller AI cloud player without CoreWeave’s scale of contracted revenue, because a rising rate regime makes balance-sheet size itself a moat and raises the bar for independent GPU clouds.

The key risk is timing: the business can look fine operationally for months while the credit markets do the damage. What would reverse the bearish case is a clear decline in debt issuance yields, faster-than-expected conversion of backlog into cash EBITDA, or evidence that capex falls well below guidance without harming growth. Absent that, the setup remains vulnerable to a rerating from a revenue multiple story to a leverage-adjusted cash flow story.

Contrarianly, the market may be over-penalizing the stock if it assumes the capex is pure burn rather than paid-in-advance infrastructure for contracted revenue. If management can keep funding costs near current levels and demonstrate sub-12 month payback on incremental capacity, the equity could stabilize despite ugly headline interest expense. Until then, the burden of proof is on management to show that growth is outrunning financing drag, not merely matching it.

Key Decisions for Investors

• Short CRWV into strength or buy 1-3 month put spreads; thesis is that equity rerates toward financing risk before operating leverage can show up. Falsify if 3Q interest expense lands below guidance midpoint and capex steps down without backlog damage.
• Pair trade: long NVDA / short CRWV for the next 1-2 quarters. This isolates hardware demand from levered balance-sheet risk; thesis breaks if CRWV proves it can fund growth at materially lower coupons or if hyperscaler capex weakens.
• Avoid chasing CRWV on revenue growth headlines alone for now; wait for evidence that incremental capex is falling and free cash flow inflection is real. Reassess after the next debt raise or refinancing window.
• Watch the 10Y/30Y Treasury complex and primary credit spreads as the main catalyst set. A sustained decline in long rates would be the cleanest reason to cover CRWV shorts.
• Consider a relative-value long basket of self-funded AI capex names vs. CRWV if AI infrastructure spend broadens. The trade works if capital markets stay tight and smaller independents lose share to hyperscalers.