The biggest question hanging over the AI cloud trade isn’t demand. It’s whether the companies building the infrastructure can keep paying for it.
CoreWeave Inc. (NASDAQ:CRWV) gave Bank of America an answer it likes. In a note on Wednesday, analyst Tal Liani kept his Buy rating and $140 price objective after the third-quarter results.
The stock rose 20.9% to $58.92 in early trading Wednesday. Even after that move, Bank of America’s target implies about 138% upside.
So why would a bank double down on a stock that has lost more than a quarter of its value in a month?
A Quarter That Beat on Every Line That Matters
Revenue grew 168% year over year to $3.66 billion. That was 700 basis points above Street expectations.
The bigger surprise was profitability. Operating margin reached 10.4%, up 540 basis points from the prior quarter and 350 basis points ahead of consensus. Operating income of $379 million was well above Bank of America’s $243 million estimate.
Liani said the outperformance came from “continued active power conversion and strategic capacity allocation to shorter-term contracts.”
Active power rose by 200 megawatts in the quarter to 1.7 gigawatts. Remaining performance obligations climbed 138% year over year to about $133 billion. More than half of the contracts behind that figure are expected to start generating revenue by year-end.
The Price of a Megawatt Keeps Climbing
The figure Wall Street should watch is the price per megawatt.
Long-dated contracts of four to five years are now being signed at $20 million to $25 million per megawatt. Earlier this year they went for about $14 million. Shorter-term contracts are going for around $40 million per megawatt.
That pricing points to a scarce resource, not a commodity. “We expect CoreWeave’s pricing power to persist as customers’ improving ROI on inference deployments support demand,” Liani wrote.
The supply imbalance is stark. Bank of America estimates that 2027 demand exceeds expected supply by 10 times, and that 2028 demand is already several times larger than planned capacity.
The bank now models operating margin rising from 10.1% in fiscal 2026 to 18.1% in 2027 and 20.8% in 2028.
The Funding Question, Answered — for Now
The bear case on CoreWeave has always been about the balance sheet, not the business.
Third-quarter capital expenditures were $14.5 billion, about $2 billion above Bank of America’s estimate. The bank projects negative free cash flow of $23 billion this year and $27.3 billion in 2027. Quarterly interest expense reached $846 million, compared with $298.5 million a year earlier.
Liani isn’t dismissing the scale of the problem. He argues the toolkit is wider than the market assumes.
“CoreWeave’s funding needs are manageable, in our view, supported by a diversified set of financing options,” he wrote.
CoreWeave doesn’t expect to tap the unsecured debt market through 2027. Instead, it can fund extra capex with customer prepayments, vendor financing and contract-backed delayed-draw term loans (DDTLs).
Customer prepayments were part of 70% of the deals signed this year, and the bank expects them to become more common. CoreWeave also plans another DDTL soon, targeting a 7% to 7.5% yield. Liani called this “currently the most attractive” option.
His conclusion is direct: “We do not forecast financing to be a constraint on CoreWeave’s capacity ramp over the next few years.”
Bank of America expects active power to reach 3.3 gigawatts by the end of 2027 and 4.6 gigawatts by the end of 2028.
The Tension Beneath the Upgrade
Not every number moved up.
Bank of America raised its revenue estimates for 2026, 2027 and 2028. At the same time, it cut its 2027 EPS estimate from a loss of 90 cents to a loss of $1.60, and lowered its 2028 EPS forecast from $3.63 to $1.82. Heavier interest costs absorb much of the operating improvement.
The $140 target is based on 26 times the bank’s calendar 2027 EV/EBIT estimate. That is well above the 14x to 19x range for the cloud data center group, a premium Bank of America attributes to CoreWeave’s growth.
The bank lists customer concentration, supplier concentration and competition from Microsoft Azure, Google Cloud and Amazon Web Services as the main downside risks.
The question is no longer whether demand exists. It’s whether the cost of capital will stay friendly long enough for CoreWeave to grow into its own backlog.
Image: Shutterstock