CoreWeave stock moved higher in premarket trading after Nvidia delivered stronger-than-expected results and signaled that AI infrastructure demand remains supply-constrained. For CoreWeave, the most important figure remains its roughly $129 billion backlog, a contracted revenue base that exceeds the company’s equity market value by a wide margin.
Yet the rally also revives the central risk in the story. CoreWeave is scaling rapidly, but it is doing so with about $35 billion of debt and quarterly interest expense of $640 million, a burden that continues to overshadow improving operating performance.
The market reaction reflects both sides of the investment case: accelerating AI demand and supplier support on one hand, and refinancing risk plus cash burn on the other.
Key Facts
- CoreWeave held a market capitalization of about $49.09 billion after closing near $89.03, with shares rising toward $94 in premarket trading.
- The company reported approximately $129 billion of backlog as of August 11, 2026, including more than $25 billion of new commitments signed early in the third quarter.
- Second-quarter 2026 revenue reached $2.58 billion, up 112% year over year, while net loss widened to $626 million.
- Quarterly interest expense climbed to $640 million, versus adjusted operating income of $128 million in the same period.
- Full-year 2026 capital expenditure guidance was raised to $35 billion to $39 billion, above projected revenue of $12.4 billion to $13.2 billion.
CoreWeave backlog and debt outlook
The immediate catalyst for CoreWeave was Nvidia’s latest quarter. Nvidia posted revenue of $96.2 billion, above the $92.07 billion consensus, while Data Center revenue reached $89.0 billion, up 117% from a year earlier. It also guided the October quarter to $108.0 billion, ahead of the $103.9 billion expected, and indicated supply remains the main constraint rather than customer demand.
That matters directly to CoreWeave because its business model depends on securing GPU supply, building data center capacity, and converting long-term contracts into active revenue. Two Nvidia disclosures stood out. First, the Vera Rubin platform entered full production, important because CoreWeave completed the first bring-up and validation of that architecture in the second quarter. Second, Amazon Web Services committed to buying 2 million Nvidia GPUs, undercutting the idea that hyperscalers are abandoning third-party accelerators in favor of only custom silicon.
For CoreWeave, the bull case rests on contract visibility. Backlog climbed from $66.8 billion at the end of 2025 to about $104 billion by June 30, 2026, then to roughly $129 billion by August 11 after fresh commitments. That is around 10 times the midpoint of 2026 revenue guidance. The company’s customer roster includes Meta, Anthropic, OpenAI, Databricks, Jane Street, Cohere, Mistral and Hudson River Trading, showing demand across AI labs, hyperscalers, enterprises and quantitative trading firms.
CoreWeave’s valuation hinges on a simple question: can a $129 billion backlog outrun a capital structure carrying $640 million of quarterly interest expense?
Why Nvidia’s update matters so much
CoreWeave’s business is unusually sensitive to upstream supply conditions because it rents high-end AI infrastructure rather than designing its own chips. Nvidia’s supply commitments more than doubled to $279 billion from $119 billion in one quarter, driven largely by memory procurement. For investors, that is a signal that the hardware pipeline supporting future CoreWeave deployments is still expanding.
Management has also argued that pricing remains firm. New contracts reportedly carry contribution margins 5 to 10 percentage points higher than recent quarters, while legacy GPU systems continue to hold value. A notable example is a contract tied to Nvidia A100 chips, first introduced in 2020, that extends through 2029. If older hardware can remain commercially useful for close to a decade, the economics of depreciation and collateral value look more durable than many skeptics assumed.
Implications for Investors
For equity investors, CoreWeave remains a high-beta AI infrastructure trade with unusually large upside and downside paths. On valuation, the stock appears inexpensive against backlog, trading at roughly 0.38 times its $129 billion contracted revenue base and around 3.7 to 4.0 times projected 2026 sales based on market capitalization. That discount partly reflects concerns that backlog is not the same as near-term cash flow.
The bigger issue is financing. CoreWeave has about $35 billion of debt, with roughly $9.7 billion due within one year based on the latest debt breakdown. Its quarterly interest expense of $640 million annualizes to about $2.56 billion, far above full-year adjusted operating income guidance of $960 million to $1.15 billion. In practical terms, the underlying business is growing quickly, but lenders still absorb more than the current operating engine generates.
Bond yields therefore matter almost as much as customer wins. When long-term Treasury yields rose above 5.3% in late August, CoreWeave shares came under pressure despite a healthy commercial pipeline. Investors should watch three variables closely over the next several quarters: the pace of backlog conversion into recognized revenue, the company’s ability to refinance upcoming maturities at manageable rates, and execution on power and capacity buildout.
There is also a strategic split in how investors may view the stock. Bulls see a leveraged call option on AI compute demand through 2028, supported by named counterparties and a market that still appears supply-constrained. Bears see a capital-intensive company spending $35 billion to $39 billion this year on capex against roughly $12.8 billion of expected revenue, with limited room for operational missteps or tighter credit conditions.
Operationally, CoreWeave is still proving that scale can overcome financial drag. Active power capacity reached 1.5 gigawatts by June 30 across 51 active data centers, with contracted capacity of about 3.7 gigawatts and a target of more than 8 gigawatts by 2030. If that infrastructure comes online on schedule, backlog conversion could accelerate. If projects slip, interest expense continues to accrue even before revenue is recognized.
CoreWeave now sits at the center of the AI infrastructure buildout, with Nvidia’s production ramp offering fresh support for the demand story. The next phase for investors will depend less on whether demand exists and more on whether CoreWeave can finance, deploy and monetize capacity quickly enough to justify the leverage.