How can CoreWeave's (CRWV.US) financing dilemma be resolved? Bernstein lowers its funding gap estimate, but the hundred-billion-dollar "cash crunch" still hangs overhead like a sword.

Title context: How can CoreWeave's (CRWV.US) financing dilemma be resolved? Bernstein lowers its funding gap estimate, but the hundred-billion-dollar "cash crunch" still hangs overhead like a sword. Text: Bernstein's latest research report recalculates the capital expenditure required for CoreWeave to deliver its approximately US$129 billion backlog of contracted orders, revising its previous estimate of about US$75 billion in incremental capital to US$44-68 billion, with the external financing gap narrowing to US$18-42 billion. But analysts believe the company's real risk lies in the dilemma of rising financing costs coupled with shortening customer contract terms: expansion depends on continuously "borrowing new money," while the contracts underpinning that financing are getting shorter. Even if AI compute remains in short supply in the near term, once industry capacity is released in a concentrated wave, CoreWeave could be the first and hardest hit. Maintains "Underperform" rating, target price slightly raised to US$54. Model revision: funding gap "slims down," but the hundred-billion-dollar financing pressure remains Since October 14, CoreWeave's share price has been on a sustained decline, with the market's core doubt about this AI compute upstart centered on the source of funds for its capacity expansion. Bernstein previously estimated that the company would need about US$75 billion in incremental capital to deliver its approximately US$129 billion in signed backlog orders; this report concedes that the old model overestimated funding needs. The old model modeled ramp-up based on a single customer contract, which is suitable for demand-side analysisit can closely tie capacity scale to major customers such as Microsoft and Metabut applying it directly to funding calculations depresses the implied ratio of revenue to capital expenditure, pushing more spending into later construction cycles and thereby exaggerating the financing gap. This time, a power installed-capacity model is used for recalculation: based on historical unit installed capital expenditure of about US$42 million/MW and a baseline assumption of US$40 million/MW, delivering the existing signed backlog orders requires US$44-68 billion in additional capital expenditure; after deducting construction in progress, existing cash, and existing debt facilities, external financing of US$18-42 billion is still needed. If further including signed capacity of nearly 4.2GW in total to be brought online by the end of 2028, total additional capital expenditure would rise to US$95-122 billion, with US$70-96 billion in external financing still pending. The analysis notes that this portion of new capacity would in theory expand in tandem with new contracts. It is worth noting that the estimates remain constrained by severely insufficient company disclosure: the megawatt scale corresponding to signed orders, the annual revenue of the order backlog (RPO is disclosed only on a combined 24-month basis), and GPU purchase commitments are all undisclosed, bringing significant uncertainty to the estimates. The real concern: the dilemma of rising financing costs and shortening contract terms Bernstein's core argument is not whether existing orders can be financed, but that both sides of the financing equation are under pressure at the same time. Business expansion requires sustained large-scale external financing, but the terms of enterprise customer contracts are continuously shortening, and the contract base supporting financing is weakening, causing the financing cost per additional dollar of capital to keep rising, creating a strategic dilemma. In the near term, the current tight supply of AI data centers does indeed support CoreWeave. The company will remain loss-making over the next two yearsadjusted EPS is forecast at -US$5.14 in 2026 and -US$6.60 in 2027and operating profit will not improve significantly until 2027. Although EBITDA is growing rapidly (US$68.26 million in 2026 and US$147.35 million in 2027), interest and other expenses remain high, eroding profit. Analysts expect that as industry-wide capacity is released and supply tightness eases, CoreWeave will be among the first and hardest-hit companies. On valuation, Bernstein maintains a 23.5x forward EV/EBIT multiple, corresponding to a target price of US$54 (slightly raised from the previous US$51), but still below the current share price, implying about 2% downside. Rating and risks: bearish stance unchanged, upside factors warrant vigilance Although the estimated funding gap has narrowed, Bernstein reiterates its "Underperform" rating. Its rating standard is underperformance of more than 15 percentage points relative to the S&P 500 over the next 12 months. The report also flags three upside risks: delays in cloud giants' self-built data centers, which would benefit third-party compute services; CoreWeave's proprietary software platform sustaining customer demand stickiness; and continued power shortages at data centers supporting its pricing power. As of October 26, 2026, CRWV closed at US$55.19, with a market capitalization of US$30.439 billion, down 22.9% year-to-date and down 58.4% over the past 12 months, significantly underperforming the S&P 500. Amid the intertwined frenzy of AI compute capital expenditure and supply shortages, whether this "AI shovel seller's" financing narrative can be delivered will remain the focus of ongoing market debate.