For investors in compute infrastructure giant CoreWeave, Inc. (CRWV), the central debate has shifted from "how big is demand" to "where will the money come from."
According to market news, Goldman Sachs tackled this head-on in an October 21 research note, delivering an exceptionally rigorous mathematical estimate of CoreWeave's funding gap. The core conclusion: to digest a massive backlog of $129 billion through August 2026, CoreWeave faces a roughly $60 billion capital expenditure (capex) wall.
However, Goldman Sachs believes investors need not panic excessively. By cleverly using customer prepayments (roughly 20% of total contract value), CoreWeave can potentially compress the actual amount of new debt it needs to raise to $20 billion to $23 billion. If it further converts its entire locked-in 4.2GW power reserve into production capacity to support $260 billion in potential future orders, the company would need about $38 billion in new debt.
Goldman Sachs maintains a "Buy" rating on the stock with a $106 price target, implying 71.5% upside from the current share price of $61.82. It also notes that CoreWeave's valuation realization will depend entirely on its ability to balance leverage between debt financing and customer prepayments.
Digesting the existing $129 billion order book: up to $60 billion in capex needed
To understand CoreWeave's funding pressure, one must first look at its existing order book. Goldman Sachs estimates that under an extreme baseline scenario with no customer prepayments (assuming a 4.5-year GPU useful life), CoreWeave needs about $60 billion in capital expenditure to digest its current $129 billion backlog.
How would this huge sum be raised? Goldman Sachs provides a detailed breakdown:
Debt financing ceiling: Based on CoreWeave's existing delayed draw term loan (DDTL) structure, 80% of capex can be debt-financed, or up to $48 billion. After deducting approximately $11 billion in committed but undrawn DDTL, the company would still need to raise $37 billion in new debt.
Cash gap: The remaining 20% (about $12 billion) must be funded by cash (SPV contributions). Goldman Sachs calculates that CoreWeave's total available cash is approximately $10.4 billion (including $5.5 billion in cash from Q2 2026, minus a $3 billion operating expense buffer, plus September convertible bond net proceeds, $2 billion in revolver capacity, and 35 million shares issued through ATM at $64).
Final gap: This means that under the no-prepayment scenario, the company still has a direct cash shortfall of $1.6 billion, which would need to be filled through customer prepayments, vendor financing, or operating cash flow after debt service.
The key to unlocking the problem: customer prepayments slash new debt to $23 billion
Goldman Sachs points out that CoreWeave's business model contains a massive pressure-relief valve: customer prepayments. Currently, 70% of the company's contracts include customer prepayment terms, averaging 15%-25% of total contract value (TCV).
When Goldman Sachs incorporates 20% customer prepayments into the model, the funding pressure drops sharply:
Under matched contract duration (4.5-year GPU useful life), the required new debt drops significantly to about $23 billion.
Under this scenario, the debt-to-loan-value ratio falls to the 50%-75% range, meaning CoreWeave could comfortably choose a debt leverage ratio below 80%.
A more optimistic assumption: If the effective GPU useful life is assumed to be 6 years (implying positive residual value), with prepayments included, the new debt requirement could drop even further to $19.5 billion.
It is worth noting that over the past six months, as the company scaled up and capital costs declined, the prepayment ratio has decreased somewhat; but Goldman Sachs believes that in an environment of low debt financing visibility, CoreWeave is fully capable of raising this ratio again.
Ultimate expansion: activating 4.2GW power reserve requires $38 billion in new debt
Beyond the existing $129 billion order book, the bigger market imagination lies in CoreWeave's vast power reserve. As of August 2026, the company has locked in 4.2GW of power capacity (compared to only 1.5GW of active capacity in Q2 2026).
Goldman Sachs estimates that if all 4.2GW of power were converted into live capacity:
It would support approximately $263 billion in potential backlog (nearly double the $129 billion order book as of August).
Under the assumption of 20% customer prepayments, CoreWeave would need to raise approximately $38 billion in new debt to complete this ambitious expansion plan.
Valuation and price target: where does the 71.5% upside come from?
Based on the above estimates, Goldman Sachs provides a 12-month price target of $106, based on an 18x multiple of expected EBIT for the fifth through eighth quarters (Q5-Q8) ahead. The current share price of $61.82 implies 71.5% potential upside.
It is worth noting that CoreWeave shares have corrected significantly since the beginning of the year:
Over the past 3 months: down 22.3%, underperforming the S&P 500 by 24.4 percentage points; over the past 6 months: down 46.3%, underperforming the S&P 500 by 50.8 percentage points; over the past 12 months: down 50.6%, underperforming the S&P 500 by 56.8 percentage points.
Against this backdrop, Goldman Sachs' Buy rating and $106 price target essentially bet that the market has priced in financing risk too pessimistically, and that CoreWeave's order backlog and power assets provide sufficiently strong fundamental support.
However, Goldman Sachs also clearly highlights four core risks investors must watch:
Financing obstacles: the process of raising tens of billions of dollars in debt financing may encounter hurdles.
Capacity ramp-up delays: execution risk in bringing data center capacity online.
Time mismatch and inflation: potential timing mismatch between contract pricing and cost inflation when actual capacity comes online.
Deteriorating supply-demand dynamics: a reversal in the compute market's supply-demand relationship, leading to reduced pricing power for new or renewed contracts.
CoreWeave's investment thesis is highly dependent on the continued openness of debt capital markets and sustained strong AI compute demand. If the financing environment tightens or AI demand expectations reverse, the company's highly leveraged structure will face a severe test. Goldman Sachs' estimates show that under the most optimistic scenario (customer prepayments + 6-year GPU life), new debt demand is about $19.5 billion—still a massive financing task that must be completed in the capital markets.
Comments