CoreWeave, Inc. (CRWV) Stock Analysis and Metrics

CoreWeave

NASDAQ: CRWV

$71.85 ▼ -9.25 (-11.40%)

At close: Jul 24, 2026 · 3:59 PM UTC

Financial Ratios

Market Cap37.91 Bn

P/E-23.80

P/S6.09

Div. Yield0.00

ROIC (Qtr)0.00

Total Debt (Qtr)24.86 Bn

Revenue Growth (1y) (Qtr)111.61

Add ratio to table…

About

CoreWeave is a purpose built AI cloud provider that delivers infrastructure and software optimized for artificial intelligence workloads. The company designs, builds, and operates a network of purpose built data centers equipped with high density GPU clusters, advanced networking, and AI optimized storage. Its platform enables the full lifecycle of AI including model training, inference, data movement, and agentic workflows. CoreWeave serves customers ranging from leading research labs to large enterprises seeking to accelerate AI innovation and deployment.

The company generates revenue by offering Infrastructure Services, Managed Software Services, and Application Software Services through multi year committed contracts on a take or pay basis and through on demand pay as you go access. As of December 31 2025, remaining performance obligations totaled $60.7 billion, reflecting strong demand for multi year capacity. Revenue reached $5.1 billion in 2025, $1.9 billion in 2024, and $229 million in 2023. CoreWeave finances infrastructure growth through asset level debt backed by customer contracts, supplemented by corporate equity and debt financing.

The company operates through the following segments: Infrastructure Services, Managed Software Services, and Application Software Services, each delivering distinct layers of the AI cloud stack.

• Infrastructure Services: provides purpose built foundational infrastructure including GPU clusters, high performance networking, and AI optimized storage that ingests, moves, and serves large volumes of data for training and inference workloads.

• Managed Software Services: delivers orchestration, automation, and observability tools such as CoreWeave Mission Control and Slurm on Kubernetes to provision infrastructure, schedule workloads, monitor performance, and enable intelligent unified control across training and inference environments.

• Application Software Services: offers developer tools including Weights & Biases, Marimo, OpenPipe, and Monolith to support model training, agent development, model serving, and the creation of reliable AI agents at scale.

CoreWeave holds a strong position in the AI cloud market by offering a purpose built platform that delivers high performance computing for training and inference workloads. Its competitive advantages include speed to market with the latest GPU generations, proprietary software such as CoreWeave Mission Control, robust security standards, and an attractive total cost of ownership. Primary competitors are hyperscale cloud providers like Amazon Web Services, Microsoft Azure, and Google Cloud Platform, as well as smaller specialized cloud vendors. The company’s focus on AI native infrastructure allows it to achieve superior utilization and efficiency compared with general purpose cloud offerings.

The company serves leading enterprises, AI research labs, and technology companies that deploy artificial intelligence workloads at scale across industries such as healthcare, finance, manufacturing, and entertainment. CoreWeave’s customers rely on its multi year contracts and on demand access to scale AI initiatives from pilot projects to production deployments.

Investment Thesis

▲ Bull case

CoreWeave ended the quarter with contracted revenue backlog approaching $100 billion, a figure that reflects strong demand across both established hyperscalers and new enterprise verticals. Of this backlog, 36% is expected to be recognized in the next 24 months and 75% within the next four years, providing a clear runway for revenue recognition. This visibility allows the company to plan capacity additions with confidence and reduces the risk of shortfall in near term results. As the infrastructure that supports this backlog matures, the initial negative contribution margin phase will pass and contribution margins are expected to stabilize in the mid 20s range. The combination of large backlog and near term recognition creates a foundation for sequential margin expansion in the coming quarters.
Customer diversification has moved beyond the original AI native foundation labs, with financial services backlog now approaching $10 billion driven by commitments from firms such as Jane Street and Hudson River Trading. This shift means that over 70% of the backlog is tied to investment grade counterparties, which improves the credit quality of the revenue stream and reduces reliance on speculative early stage companies. The presence of ten customers each committed to spend at least $1 billion further concentrates revenue in high quality, long term contracts. These enterprise relationships often involve multi year agreements that include not just GPU compute but also storage networking and software layers. As a result, the revenue base is becoming more resilient to fluctuations in the AI research cycle and more aligned with predictable enterprise budgeting cycles.
The company closed an $8.5 billion Delayed Draw Term Loan 4.0 Facility that received an A equivalent rating and was priced at an implied cost below 6%. This transaction represents the first investment grade HPC infrastructure backed loan in the market, establishing a new asset class for financing AI cloud build outs. By securing debt at such a low rate, CoreWeave lowers its weighted average cost of debt and reduces the interest burden associated with its rapid capital expenditure program. The facility’s nonrecourse structure also protects the parent company while allowing the borrower to match drawdowns with actual infrastructure deployment milestones. Over time, access to cheaper financing should improve operating leverage and support margin expansion as revenue grows faster than interest expense.
Management stated that the overwhelming majority of components and power needed to deliver 2026 revenue are already secured, which limits execution risk for the current guidance period. This includes GPUs, memory, storage and networking gear as well as long term power contracts with data center partners. By locking in these inputs ahead of time, the company can avoid surprise cost increases that could arise from supply chain bottlenecks or commodity price spikes. The secured pipeline also supports the goal of reaching more than 1.7 gigawatts of active power by the end of 2026 and more than 3.5 gigawatts of contracted power by the end of 2027. With the bulk of required inputs in hand, the focus shifts to execution and installation rather than sourcing, which should improve timing and reduce the likelihood of delays.
The mix of workloads is shifting toward inference, with management noting that materially more than 50% of consumed power is now used for inference rather than training. Inference workloads tend to be less capital intensive and more directly tied to revenue generation, which can improve utilization and pricing power. As inference demand grows, the company can expect higher average selling prices for its GPU fleets and better contribution margins per unit of power deployed. This trend also encourages customers to adopt additional layers of the stack such as storage and networking, which further increases wallet share. Over time, a higher inference share should contribute to more stable and predictable revenue streams compared with the more volatile training cycle.

CoreWeave ended the quarter with contracted revenue backlog approaching $100 billion, a figure that reflects strong demand across both established hyperscalers and new enterprise verticals. Of this backlog, 36% is expected to be recognized in the next 24 months and 75% within the next four years, providing a clear runway for revenue recognition. This visibility allows the company to plan capacity additions with confidence and reduces the risk of shortfall in near term results. As the infrastructure that supports this backlog matures, the initial negative contribution margin phase will pass and contribution margins are expected to stabilize in the mid 20s range. The combination of large backlog and near term recognition creates a foundation for sequential margin expansion in the coming quarters.
Customer diversification has moved beyond the original AI native foundation labs, with financial services backlog now approaching $10 billion driven by commitments from firms such as Jane Street and Hudson River Trading. This shift means that over 70% of the backlog is tied to investment grade counterparties, which improves the credit quality of the revenue stream and reduces reliance on speculative early stage companies. The presence of ten customers each committed to spend at least $1 billion further concentrates revenue in high quality, long term contracts. These enterprise relationships often involve multi year agreements that include not just GPU compute but also storage networking and software layers. As a result, the revenue base is becoming more resilient to fluctuations in the AI research cycle and more aligned with predictable enterprise budgeting cycles.
The company closed an $8.5 billion Delayed Draw Term Loan 4.0 Facility that received an A equivalent rating and was priced at an implied cost below 6%. This transaction represents the first investment grade HPC infrastructure backed loan in the market, establishing a new asset class for financing AI cloud build outs. By securing debt at such a low rate, CoreWeave lowers its weighted average cost of debt and reduces the interest burden associated with its rapid capital expenditure program. The facility’s nonrecourse structure also protects the parent company while allowing the borrower to match drawdowns with actual infrastructure deployment milestones. Over time, access to cheaper financing should improve operating leverage and support margin expansion as revenue grows faster than interest expense.
Management stated that the overwhelming majority of components and power needed to deliver 2026 revenue are already secured, which limits execution risk for the current guidance period. This includes GPUs, memory, storage and networking gear as well as long term power contracts with data center partners. By locking in these inputs ahead of time, the company can avoid surprise cost increases that could arise from supply chain bottlenecks or commodity price spikes. The secured pipeline also supports the goal of reaching more than 1.7 gigawatts of active power by the end of 2026 and more than 3.5 gigawatts of contracted power by the end of 2027. With the bulk of required inputs in hand, the focus shifts to execution and installation rather than sourcing, which should improve timing and reduce the likelihood of delays.
The mix of workloads is shifting toward inference, with management noting that materially more than 50% of consumed power is now used for inference rather than training. Inference workloads tend to be less capital intensive and more directly tied to revenue generation, which can improve utilization and pricing power. As inference demand grows, the company can expect higher average selling prices for its GPU fleets and better contribution margins per unit of power deployed. This trend also encourages customers to adopt additional layers of the stack such as storage and networking, which further increases wallet share. Over time, a higher inference share should contribute to more stable and predictable revenue streams compared with the more volatile training cycle.

▼ Bear case

Capital expenditure guidance was increased to a range of $31 billion to $35 billion for the full year, with the low end raised specifically to reflect higher component pricing. This suggests that the cost of acquiring essential hardware such as GPUs, memory and networking equipment is trending upward and may continue to do so as demand for AI infrastructure remains intense. If component prices keep rising faster than the company’s ability to pass through costs to customers, gross margins could remain under pressure despite revenue growth. The elevated CapEx also means that cash outflow will remain high, potentially limiting flexibility for other strategic uses such as acquisitions or shareholder returns. Investors will need to monitor whether pricing power can keep pace with input cost inflation to avoid margin compression.
Operating expenses rose to $2.2 billion in the quarter, driven by the ongoing scale up of active power capacity and increased investment in sales, marketing and general and administrative functions. This increase in overhead contributed to a net loss of $740 million, which is substantially higher than the $315 million loss reported a year earlier. While revenue is growing rapidly, the business is still in a phase where fixed costs associated with building out a global cloud platform outweigh the incremental profit from additional revenue. Unless operating expenses begin to grow at a slower rate than revenue, the path to sustained profitability may be delayed. The high expense base also raises the break even point for adjusted operating income, requiring a larger scale of revenue to achieve meaningful margin expansion.
Interest expense for the quarter reached $536 million, up from $264 million a year earlier, reflecting the growing debt balance used to finance the rapid build out of data center capacity. Although the newest debt facilities are investment grade and carry relatively low interest rates, the absolute level of interest expense remains significant and will continue to weigh on the income statement until operating income expands sufficiently. The company’s strategy of using delayed draw term loans aligns funding with deployment milestones, but any delay in bringing capacity online could cause interest to accrue without corresponding revenue. Investors should watch the trend in interest coverage ratio to ensure that earnings before interest and taxes can comfortably service the debt load.
Gross and operating margins remain depressed by the timing lag between infrastructure deployment and revenue recognition, with management noting that new capacity typically runs at negative contribution margins for the first one to two months before stabilizing in the mid 20s range after about three months. If the rate of new capacity additions outpaces the ability to fill that capacity with paying workloads, the period of negative contribution could extend beyond the expected window. Any slowdown in customer uptake or deployment delays would keep a larger portion of the fleet in the low margin phase, dragging down overall profitability. Furthermore, as the base of active power grows, the incremental impact of each new deployment on total margins diminishes, but only if the ramp up continues smoothly.
Although non investment grade AI native companies now account for less than 30% of the backlog, a meaningful portion of revenue still depends on early stage customers whose spending can be volatile and subject to changes in venture capital funding or shifts in AI research priorities. A downturn in startup financing or a shift away from large model training could reduce demand from this segment and affect utilization rates, especially for the latest generation GPUs that are heavily sold out to these customers. The company’s reliance on a small number of large enterprise contracts also creates concentration risk; if one of those partners were to renegotiate or reduce its commitment, the impact on backlog would be material. Diversification helps but does not eliminate the exposure to cyclical trends in the broader AI market.

Capital expenditure guidance was increased to a range of $31 billion to $35 billion for the full year, with the low end raised specifically to reflect higher component pricing. This suggests that the cost of acquiring essential hardware such as GPUs, memory and networking equipment is trending upward and may continue to do so as demand for AI infrastructure remains intense. If component prices keep rising faster than the company’s ability to pass through costs to customers, gross margins could remain under pressure despite revenue growth. The elevated CapEx also means that cash outflow will remain high, potentially limiting flexibility for other strategic uses such as acquisitions or shareholder returns. Investors will need to monitor whether pricing power can keep pace with input cost inflation to avoid margin compression.
Operating expenses rose to $2.2 billion in the quarter, driven by the ongoing scale up of active power capacity and increased investment in sales, marketing and general and administrative functions. This increase in overhead contributed to a net loss of $740 million, which is substantially higher than the $315 million loss reported a year earlier. While revenue is growing rapidly, the business is still in a phase where fixed costs associated with building out a global cloud platform outweigh the incremental profit from additional revenue. Unless operating expenses begin to grow at a slower rate than revenue, the path to sustained profitability may be delayed. The high expense base also raises the break even point for adjusted operating income, requiring a larger scale of revenue to achieve meaningful margin expansion.
Interest expense for the quarter reached $536 million, up from $264 million a year earlier, reflecting the growing debt balance used to finance the rapid build out of data center capacity. Although the newest debt facilities are investment grade and carry relatively low interest rates, the absolute level of interest expense remains significant and will continue to weigh on the income statement until operating income expands sufficiently. The company’s strategy of using delayed draw term loans aligns funding with deployment milestones, but any delay in bringing capacity online could cause interest to accrue without corresponding revenue. Investors should watch the trend in interest coverage ratio to ensure that earnings before interest and taxes can comfortably service the debt load.
Gross and operating margins remain depressed by the timing lag between infrastructure deployment and revenue recognition, with management noting that new capacity typically runs at negative contribution margins for the first one to two months before stabilizing in the mid 20s range after about three months. If the rate of new capacity additions outpaces the ability to fill that capacity with paying workloads, the period of negative contribution could extend beyond the expected window. Any slowdown in customer uptake or deployment delays would keep a larger portion of the fleet in the low margin phase, dragging down overall profitability. Furthermore, as the base of active power grows, the incremental impact of each new deployment on total margins diminishes, but only if the ramp up continues smoothly.
Although non investment grade AI native companies now account for less than 30% of the backlog, a meaningful portion of revenue still depends on early stage customers whose spending can be volatile and subject to changes in venture capital funding or shifts in AI research priorities. A downturn in startup financing or a shift away from large model training could reduce demand from this segment and affect utilization rates, especially for the latest generation GPUs that are heavily sold out to these customers. The company’s reliance on a small number of large enterprise contracts also creates concentration risk; if one of those partners were to renegotiate or reduce its commitment, the impact on backlog would be material. Diversification helps but does not eliminate the exposure to cyclical trends in the broader AI market.

Peer Comparison

Companies in the Software - Infrastructure

S.No.	Ticker	Company	Market Cap	P/E	P/S	Total Debt (Qtr)
1	MSFT	Microsoft Corp	2,842.90 Bn	58.08	8.93	40.26 Bn
2	PAGS	PagSeguro Digital Ltd.	2,572.26 Bn	4,596.13	680.02	0.44 Bn
3	ORCL	Oracle Corp	329.59 Bn	17.64	4.89	122.34 Bn
4	RPAY	Repay Holdings Corp	314.63 Bn	-2,562.84	-	0.43 Bn
5	PLTR	Palantir Technologies Inc.	294.47 Bn	128.41	56.37	-
6	PANW	Palo Alto Networks Inc	227.51 Bn	177.48	23.00	-
7	CRWD	CrowdStrike Holdings, Inc.	183.28 Bn	-1,136.88	38.09	0.75 Bn
8	FTNT	Fortinet, Inc.	112.44 Bn	57.52	15.82	0.50 Bn

Research Tools

Developers

About

Investment Thesis

Geographical Breakdown of Revenue (2025)

Peer Comparison

Research Tools

Developers

About

Investment Thesis

Geographical Breakdown of Revenue (2025)

Peer Comparison

Create an account

Sign In

Reset Password