Why I'm Still Not Buying CoreWeave Stock

Q4 revenue surged 110% YoY to roughly $1.6B and full-year revenue reached $5.1B (+168%), with management guiding 2026 revenue of $12–13B (≈144% growth at midpoint) and reporting a $66.8B backlog (+342% YoY). Management expects capex to double to $30–35B in 2026 while Q4 net loss was $452M and interest expense more than doubled to $388M, creating material balance-sheet and interest-service pressure. At a ~$45B market cap (~3.5x management's 2026 revenue target), the author argues the stock already prices in flawless execution and a major swing to profitability, and thus prefers to stay on the sidelines.

Analysis

CoreWeave’s path is less a pure technology story and more an execution-and-capital cycle play: whoever supplies and finances the racks wins or loses as much as the company running them. That elevates GPU OEMs and financing intermediaries into primary beneficiaries while turning any single mis-timed build milestone into a liquidity event that can rapidly reset equity valuations. Expect second-order pressure on power, interconnect, and real-estate vendors as accelerated demand compresses lead times and forces premium pricing for turnkey capacity.

The most immediate tail risks are funding and supply friction rather than end-customer demand: rising borrowing costs, covenant squeezes, or GPU delivery delays can turn a backlog into deferred revenue quickly. Over a 3–12 month horizon watch three triggers — new debt raises at unfavorable yields, material slippage in build timelines, and any signs of hyperscaler insourcing ramping — any of which could disproportionately hurt an asset-light competitor. Conversely, a strategic JV or non-dilutive financing facility would be a positive catalyst that materially derisks the balance sheet before margins normalize.

A contrarian read is that the market may be underestimating pricing optionality for on-demand AI compute: persistent GPU tightness would allow providers to pass through price increases and compress payback periods on recent builds, accelerating free cash flow conversion. That optionality is binary and concentrated — it favors counterparties with secured supply relationships and flexible pricing, not necessarily the largest headline market caps. Positioning should therefore be asymmetric: protect against the downside of capex/execution failure while retaining selective upside exposure to GPU scarcity-driven pricing power.

Key Decisions for Investors

• Initiate a tactical short CRWV (size 1–2% NAV) vs long NVDA (size 1–2% NAV) pair trade, 3–9 month horizon. Rationale: hedge idiosyncratic execution risk in the infrastructure operator while maintaining exposure to GPU demand; target a 2:1 payoff if CRWV equity falls 30–50% while NVDA holds or outperforms. Use a stop-loss on the short at +15% and hedge with calls on NVDA if volatility spikes.
• Buy a CRWV 6–9 month put spread (sell higher strike) to cap premium cost — allocate no more than 0.5% NAV. Risk/reward: limited downside cost with asymmetric payoff if funding or build delays surface; roll or close if debt issuance is completed at market-friendly rates.
• Long NVDA 6–12 month call spread (debit), sized to offset put spread cost, as a directional play on sustained GPU tightness. Target 2–3x gross return if GPU ASPs remain elevated; reduce position on any signs of meaningful hyperscaler inventory destocking or NVDA margin compression.
• Monitor capital markets windows: buy protection (short-dated puts) on CRWV ahead of any announced debt raises or equity offers, and be ready to add concentrated short exposure if issuance occurs at punitive yields. This event-driven approach limits carry while capturing convex downside.