Bernstein reiterates CoreWeave stock Underperform rating at $56 By Investing.com

Bernstein reiterated an Underperform on CoreWeave with a $56 price target and values the firm at 28.4x FY2027 EBIT; the stock trades at $81.96, ~46% above Bernstein's target. CoreWeave carries $29.8B of total debt (D/E 8.94) and levered FCF of -$7.25B, with Bernstein forecasting >$80B in capex over the next five years. Positive operational developments include adding Nvidia HGX B300 chips, a 300 MW AI data center project (CoreWeave securing 140 MW), and partnerships with Cerebras, BCE, Zonos and Cline, while BofA resumed coverage with a buy stance, highlighting exposure to the $79B AI infrastructure market.

Analysis

CoreWeave’s asset-heavy expansion amplifies two structural frictions: liquidity runway sensitivity to rate moves, and capex timing mismatches with modelled demand. Depreciation and long-lived asset schedules will mechanically mute EBITDA conversion even as top-line throughput grows, creating recurring covenant and refinancing pinch points on any 6–24 month funding horizon.

GPU supply and vendor pricing are the choke points that determine whether scale converts into sustainable gross margins or a capital treadmill. If suppliers preserve aftermarket scarcity or increase OEM pricing, smaller builders face margin pressure and extended payback periods; conversely, any sudden capacity loosening would compress ASPs and hurt returns for late-cycle builders disproportionately.

Second-order beneficiaries include local utilities, construction contractors, and power-equipment OEMs in regions hosting large campuses — these players see multi-year, lumpy revenue tails tied to grid upgrades and long-term service contracts. Financially, the clearest inflection is balance-sheet optionality: an equity or asset sale could reset leverage quickly, while covenant breaches or rising rates would force dilutive financing or asset firesales within 9–18 months.

A contrarian upside is plausible but binary: if CoreWeave secures multi-year take-or-pay contracts with strong counterparties or executes a structured JV with a hyperscaler, valuation multiples could re-rate to industry growth comps; absent that, market will likely price a sustained discount for financing risk and capex intensity.

Key Decisions for Investors

• Short CRWV equity or buy 6–9 month put spreads sized for 2–4% portfolio exposure — target a 25–40% downside scenario over 3–12 months. Rationale: capex cadence + refinancing risk; hedge with a small-long NVDA position to offset systemic AI demand moves. Max loss = premium paid; target return 2–4x if downside realized.
• Pair trade: long NVDA (or long-dated NVDA calls, 9–18 months) vs short CRWV (equity or calls sold) — horizon 6–12 months. Expect NVDA to capture pricing power and supply control while smaller cloud builders face margin squeeze; aim for asymmetric 3:1 upside potential vs limited carry cost on the short leg.
• Buy BCE (or regional utility/infrastructure exposure) for 12–24 months as a defensive play on recurring colocation and power upgrade revenues in data-center hubs. Position size 3–5% of risk budget; reward is stable dividend plus 10–20% upside if multi-megawatt builds proceed, with low downside relative to pure growth names.
• If concerned about immediate volatility around financing/earnings windows, establish a hedged options collar on CRWV (buy puts and finance with OTM calls) for 3–6 months to limit downside to pre-defined levels while retaining some upside. This reduces funding shock risk at the cost of capped upside — effective for balance-sheet-sensitive exposures.