Active Capacity: The company expanded its active power by nearly 500 megawatts (MW) during the quarter, reaching a total of 1.5 gigawatts (GW).
Contracted Pipeline: Total contracted power grew to approximately 3.7 GW.
Portfolio Diversification: This power expansion included further diversification of its portfolio of providers and an increase in its powered land footprint.
Historic Revenue Backlog
The company's top-line growth is underpinned by a massive forward-looking pipeline. CoreWeave reported a revenue backlog of approximately $104 billion as of June 30, 2026. Furthermore, this backlog figure does not include more than $25 billion in net new customer commitments that were secured in early Q3.
Capital Structure and Financial Margins
Funding a multi-gigawatt infrastructure pipeline requires historic levels of capital, and CoreWeave executed several major financial maneuvers to support its growth:
Unsecured Debt & Eurobonds: The company raised more than $10 billion through unsecured debt and convertible bonds, which included CoreWeave's inaugural Eurobond issuance.
HPC-Backed Facility: CoreWeave closed a milestone $3.1 billion term loan, representing the first-ever publicly syndicated delayed draw facility backed specifically by HPC infrastructure.
Strategic Investment: Jane Street provided a $1 billion strategic investment following the expansion of their commercial relationship in Q1 2026.
On a profitability basis, CoreWeave generated $1.51 billion in Adjusted EBITDA for the quarter, yielding a 59% Adjusted EBITDA margin, compared to $753 million in Q2 2025. The company reported a GAAP net loss of $626 million, widening from a $290 million net loss in the prior-year period.
Hardware Validation and Agentic AI
CoreWeave also achieved several critical hardware and software milestones during the quarter. The company successfully completed the industry's first bring-up and validation of the NVIDIA Vera Rubin NVL72. Additionally, CoreWeave set new MLPerf records for training and inference using open-source models on the NVIDIA Grace Blackwell platform, achieving the lowest cost per token for inference in its test runs.
To support cross-cloud functionality for enterprise clients, the company launched CoreWeave Interconnect, offering private, high-performance fiber linking directly to other hyperscale platforms, starting with Google Cloud. CoreWeave also rolled out unified agentic AI capabilities, including CoreWeave ARIA (AI Research and Iteration Agent), designed to uncover hidden insights and drive continuous model improvement in production.
