CoreWeave shares jumped 14% in extended trading on Tuesday after the AI infrastructure provider reported results than topped Wall Street expectations.
Here's how the company did relative to LSEG consensus:
- Earnings per share: Loss of $1.03 adjusted vs. loss of $1.20 expected
- Revenue: $2.58 billion vs. $2.56 billion expected
Revenue climbed 112% during the quarter from a year earlier, CoreWeave said in a statement. Net loss of $626 million increased from $290 million, or 60 cents per share, a year ago.
The company's revenue backlog now stands at $104 billion, a figure that excludes over $25 billion in new commitments from the third quarter, and it boasted 1.5 gigawatts of active power.
With respect to guidance, management called for $3.4 billion to $3.6 billion in third-quarter revenue, which would imply 158% growth at the middle of the range. Analysts polled by LSEG were looking for $3.43 billion.
For 2026, CoreWeave now sees $960 million to $1.15 billion in adjusted operating income on $12.4 billion to $13.2 billion in revenue. Analysts surveyed by LSEG had expected $12.63 billion in full-year revenue. In May it was projecting $900 million to $1.1 billion in adjusted operating income on $12 billion to $13 billion in revenue.
The company called for over 1.85 gigawatts in active power by year end, with $35 billion to $39 billion in annual capital expenditures. The forecast as of May included capital spending between $31 billion and $35 billion.
The 8-year-old company has been racing cloud market leaders Amazon, Google and Microsoft to open data centers filled with chips that can run generative artificial intelligence models. Unlike them, CoreWeave isn't profitable. As of quarter end, CoreWeave had $35 billion in debt on its balance sheet to cover the cost of Nvidia graphics processing units and other equipment.
Across the U.S., people have become more critical of data center construction. In July New York's governor, Kathy Hochul, signed an executive order establishing a moratorium on new large-scale data centers.
"When we talk through the numbers with you guys, we're basing our progress on where we are today and what we have guided here," CEO Mike Intrator said on a conference call with analysts. "None of those numbers will be impacted by the regulatory pushback as of today."
Still, the opposition isn't helping.
"There is no question that when when parts of the U.S. become unwilling to even engage in those conversations, that it becomes more challenging," Intrator said in an interview with CNBC.
CoreWeave it seeing favorable trends on renting Nvidia chips, Intrator said. "Pricing and margins for our Blackwell and Vera Rubin SKUs are setting new highs, while pricing for prior generation SKUs is at or above where it was years ago," he said.
The company is passing component price hikes down to customers, said Nitin Agrawal, its finance chief.
During the quarter, Meta said it would spend an additional $21 billion with CoreWeave, which also announced a multi-year agreement with Anthropic and a $6 billion commitment from quantitative trading firm Jane Street.
Meanwhile, competition is ramping. SpaceX has begun selling excess computing capacity, and Meta has considered launching a cloud business. Rival Nebius gained 5% in extended trading.
"Even with this increased competition, we're seeing demand, pricing and margin all expanding, which is a signal for the growth in the CoreWeave ... product, as well as our growth overall in an already massive TAM that exists," Agrawal said.
As of Tuesday's close, CoreWeave shares had gained 26% year to date, while the S&P 500 was up almost 13%. The stock debuted on Nasdaq in March 2025.
Facts Only
* CoreWeave shares rose 14% in extended trading on Tuesday.
* Quarterly revenue was $2.58 billion, a 112% increase from the previous year.
* Adjusted earnings per share was a loss of $1.03.
* Net loss for the quarter was $626 million.
* Revenue backlog is $104 billion, excluding $25 billion in new third-quarter commitments.
* Active power stands at 1.5 gigawatts.
* Third-quarter revenue guidance is $3.4 billion to $3.6 billion.
* 2026 revenue projection is $12.4 billion to $13.2 billion with adjusted operating income of $960 million to $1.15 billion.
* Projected year-end active power is over 1.85 gigawatts with annual capital expenditures of $35 billion to $39 billion.
* Total debt on the balance sheet at quarter end was $35 billion.
* Meta committed an additional $21 billion to CoreWeave.
* Jane Street provided a $6 billion commitment.
* New York Governor Kathy Hochul signed an executive order in July establishing a moratorium on new large-scale data centers.
* CoreWeave debuted on the Nasdaq in March 2025.
Executive Summary
CoreWeave is experiencing rapid hyper-growth driven by the demand for generative AI infrastructure, evidenced by a 112% year-over-year revenue increase and a massive $104 billion backlog. Despite beating Wall Street expectations on revenue and adjusted earnings per share, the company remains unprofitable, reporting a net loss of $626 million and carrying $35 billion in debt to finance high-cost Nvidia hardware. Management is aggressively scaling, increasing capital expenditure forecasts and projecting a shift toward profitability by 2026.
The company faces a complex competitive and regulatory landscape. While securing major commitments from Meta, Anthropic, and Jane Street, CoreWeave is competing against hyperscalers like Amazon, Google, and Microsoft, as well as new entrants like SpaceX. Additionally, regulatory headwinds are emerging, specifically in New York, where a moratorium on large-scale data centers has been implemented. While leadership maintains that current guidance remains unaffected by these restrictions, they acknowledge that increasing regional opposition to data center construction creates operational challenges.
Full Take
The strongest version of this narrative is one of a specialized "pure-play" AI infrastructure provider successfully carving out a niche against tech giants by moving faster and specializing deeper in Nvidia hardware deployment. The growth trajectory is numerically staggering, suggesting a market where demand for compute currently dwarfs available supply.
However, this narrative rests on a precarious financial architecture. The company is effectively operating as a leveraged bet on AI longevity; it is borrowing tens of billions to build capacity in hopes that the "Blackwell" and "Rubin" cycles maintain high margins. The reliance on debt to fund hardware—which historically depreciates rapidly—creates a high-risk profile. There is a tension between the exuberant stock market reaction and the reality of a widening net loss.
The root cause is the "Compute Gold Rush" paradigm, where the assumption is that AI demand is an infinite upward curve. This echoes the build-out phases of previous infrastructure booms (such as fiber optics in the late 90s), where capacity was built far ahead of actual utilization.
The implications for human agency are significant: as compute power concentrates in the hands of a few providers funded by massive debt and quantitative firms, the "entry fee" for AI innovation rises, potentially gatekeeping the technology.
Bridge Questions:
1. If AI model efficiency improves rapidly, reducing the need for massive GPU clusters, what happens to a balance sheet built on $35 billion of specialized hardware debt?
2. How does the transition from "growth at all costs" to the 2026 profitability target affect service pricing for smaller AI startups?
Counterstrike Scan: A coordinated campaign to inflate a stock would emphasize "backlog" and "gigawatts" (large, impressive numbers) while burying the net loss and debt in the middle of the text. While the data here is transparent, the framing focuses heavily on the "beat" and the "growth," which mirrors a bullish promotional playbook. However, the inclusion of regulatory hurdles and the lack of current profitability prevents this from being a pure promotional piece.
Patterns detected: none
Sentinel — Human
The text functions effectively as standard financial reporting, blending quantitative data with contextual commentary, which strongly indicates human journalistic construction rather than pure synthetic generation.