CoreWeave’s financing gap revised lower, but expansion still faces a test from funding costs
After recalculating, Bernstein lowered its estimate of the additional financing needed to fulfill existing orders from approximately 75 billion dollars to 18 billion–42 billion dollars; the funding requirement would rise substantially if it covered all contracted power capacity scheduled to come online by the end of 2028.
Original source：BernsteinFirst published on Unhidea：Original source date：
Key takeaway
The financing gap for existing orders is smaller than Bernstein previously estimated, but new capacity still requires substantial capital, and shorter contract terms could raise financing costs.
Why was the financing gap revised down sharply?
Bernstein previously estimated that CoreWeave needed to raise approximately 75 billion dollars to fulfill around 129 billion dollars of contracted orders. After switching to a power-capacity-based method, the firm lowered the gap to approximately 18 billion–42 billion dollars.
The previous model estimated requirements based on the ramp-up schedule of an individual customer contract. Bernstein believes that applying the same framework to financing analysis understated the ratio of revenue to capital expenditure and allocated too much spending to the remaining construction phase, thereby overstating financing needs.
How is the funding requirement for existing orders calculated?
Bernstein estimates that fulfilling existing orders requires approximately 1.25 billion–1.5 billion watts of additional power capacity. At 35 million–45 million dollars per megawatt, incremental capital expenditure would be approximately 44 billion–68 billion dollars. The base case is 1.3 billion watts at 40 million dollars per megawatt, totaling 52 billion dollars.
After deducting approximately 11.9 billion dollars in construction in progress, 5 billion dollars in cash available for the calculation, and 9.1 billion dollars in existing delayed-draw term loan capacity from base-case capital expenditure, the remaining financing requirement is approximately 26 billion dollars. The cash assumption is 90% of cash on the balance sheet, with the remainder reserved for operations.
The estimate is sensitive to capacity and unit costs: the company has not disclosed the power capacity associated with signed orders, annual order revenue, or GPU purchase commitments. Bernstein uses the bill of materials for a Vera Rubin rack to support its base-case cost assumption; as of the second quarter of 2026, it calculated historical cumulative capital expenditure at approximately 42 million dollars per megawatt.
How much wider would the gap be if expansion plans were included?
If the calculation covers all contracted power capacity CoreWeave plans to bring online by the end of 2028, totaling nearly 4.2 billion watts, Bernstein estimates that incremental capital expenditure would rise to 95 billion–122 billion dollars, of which approximately 70 billion–96 billion dollars would still need to be raised.
This range covers more capacity than is required for existing orders. Bernstein expects the additional capacity to be supported by new contracts and substantial order growth; the two sets of figures therefore correspond to different construction scopes.
Why does Bernstein remain cautious after the gap narrowed?
The firm is more concerned that additional capital will become increasingly expensive: business growth requires continued financing, while the company is shifting toward shorter-term corporate contracts, shortening the contract terms supporting that financing.
Tight data center supply currently benefits CoreWeave; Bernstein expects the company could be affected earlier and to a greater extent as capacity expands and supply constraints ease. Conversely, delays in bringing major cloud providers’ self-built capacity online, continued power shortages, or the development of dedicated software platforms could support the company’s demand and pricing power.
Based on lower expected capital expenditure and corresponding debt financing for fiscal 2027, Bernstein raised its price target from 51 dollars to 54 dollars, while maintaining its underperform rating and target enterprise-value-to-operating-profit multiple of 23.5 times.
