CoreWeave, Inc.

What they do

CoreWeave operates a vertically integrated AI cloud platform, delivering high-density GPU compute, networking, storage, and orchestration software (CoreWeave Mission Control, SUNK) to AI labs and enterprises. The company builds and manages purpose-built data centers with liquid cooling and high-speed interconnects, primarily serving customers under committed take-or-pay contracts. It also provides managed software services, including Weights & Biases, and developer tools for AI model development and deployment.

Revenue drivers

• Cloud compute services — Primary revenue source from committed and on-demand access to AI-optimized GPU infrastructure; revenue grew from $1.92B in 2024 to $5.13B in 2025.
• Managed software services — Includes services like Weights & Biases and developer tools, providing additional revenue and platform stickiness, though relative size not disclosed.
• Data center capacity expansion — Revenue scales with active power capacity: 360 MW at end-2024, 850 MW at end-2025, and 1.5 GW by mid-2026, supporting more customer workloads.

Recent performance

For Q2 2026, CoreWeave reported record revenue of $2.58B, up from $1.21B in Q2 2025, but recorded a net loss of $626M, compared to a $290M loss a year earlier. GAAP operating loss was $49M versus a $19M operating income in the prior-year quarter; adjusted EBITDA was $1.51B, a 59% margin. Revenue backlog reached approximately $104B as of June 30, 2026, with an additional $25B in commitments in early Q3. The company ended the quarter with $5.52B cash, $72.05B liabilities, and $5.02B equity.

Strategy

CoreWeave plans to continue expanding its AI cloud platform and data center footprint, with contracted power capacity of ~3.7 GW as of mid-2026. It is investing in software capabilities, cross-cloud interconnect with Google Cloud, and agentic AI tools like ARIA and Sandboxes. The company focuses on being first-to-market with new GPU generations, evidenced by the bring-up of NVIDIA Vera Rubin NVL72. It also pursues acquisitions, including Weights & Biases, OpenPipe, Marimo, and Monolith AI, to deepen its software stack and enterprise reach.

Risks

• Customer concentration — A substantial portion of revenue comes from a limited number of customers; losing one or more top customers would materially hurt results.
• Power and component supply constraints — Inability to access sufficient power or increased power costs, plus reliance on limited suppliers for critical hardware, could delay expansion or raise costs.
• High leverage and capital needs — The business requires substantial capital expenditures and has long-term debt of $17.31B; additional capital may be needed to fund growth, and asset-level debt is tied to customer contracts.
• Rapid growth and operational scaling — Recent growth may not be indicative of future performance, and any delays in data center availability or operational missteps could harm profitability.

Outlook

Management expects customer demand to accelerate, driven by enterprise adoption and platform deepening. The company plans to continue scaling power capacity and contracted backlog, with record revenue backlog supporting future revenue recognition. Management indicated an inflection point in operating leverage as the business scales, though GAAP profitability remains negative. No specific forward guidance was provided in the excerpts.