Bernstein: CoreWeave — Hyperscalers Are Situationships, Not Spouses
Summaries are public. PDF access requires an active membership; all members have the same access. Sign in
Bernstein initiates coverage of CoreWeave (CRWV) with an Underperform rating and a $56 price target, implying roughly 31% downside from the March 13, 2026 closing price of $81.11. While CoreWeave has capitalized on surging demand for dedicated GPU cloud capacity — growing revenue from $229 million in 2023 to $5.1 billion in 2025 — Bernstein argues this growth trajectory is unsustainable.
The core thesis centers on hyperscalers building approximately 23 GW of proprietary data center capacity expected online by end-2028. Once this capacity relieves market tightness, hyperscaler contracting with new cloud providers like CoreWeave should decline sharply. Bernstein also disputes CoreWeave's claimed software moat, contending that hyperscaler scale and spending power will prove decisive if they compete directly in GPU cloud services.
The financial profile adds concern. CoreWeave carries high debt levels, and Bernstein's model projects negative cumulative free cash flow to equity over the forecast horizon. On a non-cumulative basis, Bernstein expects FCFE to turn positive only in 2030, a year later than consensus.
Valuation appears stretched. The market implies a $162 billion backlog, but Bernstein estimates only $117 billion through 2030, against a current reported backlog of $66.8 billion. Bernstein projects roughly $20 billion in new contract signings for 2026 and $15 billion for 2027, below market expectations.
Major existing contracts include OpenAI ($22 billion), Microsoft ($10 billion), Meta ($14 billion), IBM ($3 billion), and NVIDIA ($6.3 billion standby). Key catalysts include hyperscaler capacity buildout progress, GPU-as-a-Service traction, and CoreWeave's capital expenditure and debt trajectory.