Detailed narrative
Record bookings and backlog composition
CoreWeave signed more than $40 billion of new commitments in Q1 — described as its strongest booking quarter — lifting contracted revenue backlog to $99.4 billion, up nearly 50% sequentially and close to 4x year-over-year. The backlog is near-term weighted: 36% is expected to be recognized in the next 24 months and 75% within four years.
Most new commitments contribute toward 2027 targets and are expected to be highly contribution-margin positive, consistent with historically underwritten return profiles. Commitments from non-investment-grade AI-native companies and foundation labs now represent less than 30% of overall backlog, reflecting diversification, while the weighted average contract length for new capacity remains approximately five years.
Customer diversification into new verticals
The company added Anthropic and signed multiple new Meta orders including a $21 billion agreement announced in early April, stating the world's four preeminent AI model developers and nine of the ten AI leaders outside China now run on CoreWeave.
Financial Services has emerged as a $10 billion-approaching backlog vertical, driven by Jane Street adding $6 billion of capacity in Q1 and new customer Hudson River Trading. Physical AI and spatial computing surpassed $1 billion in backlog contributions with new customers World Labs, Physics X and Sunday Robotics.
CoreWeave now has 10 customers each committed to spending at least $1 billion.
Platform expansion beyond GPUs
More than 90% of reserved-instance customers use at least two CoreWeave products and more than 75% use three or more. Storage is multiplying quickly, and software, CPU and networking businesses are each expected to exceed $100 million of ARR by year-end.
New capabilities launched in Q1 include CoreWeave Trust Center for enterprise security/compliance, Flex reservation and spot pricing (both immediately oversubscribed), CoreWeave Interconnect in collaboration with Google Cloud, and CoreWeave Omni to deploy the full stack in customers' own data centers.
Perplexity will run next-generation inference on CoreWeave while using Weights & Biases for model training and management.
Capacity build-out and execution
CoreWeave surpassed 1 gigawatt of active power and grew total contracted power to more than 3.5 GW after adding over 400 MW in Q1 (versus 200 MW in Q4), all via long-term leases, with the substantial majority expected online by end of 2027.
The company operates across close to 50 data centers with no single data center provider delivering more than 17% of active infrastructure. It is on track for more than 1.7 GW of active power by end 2026 and targets more than 8 GW by 2030.
Alongside leases, CoreWeave is accelerating self-build sites (first online later this year) and evaluating its NVIDIA 5 GW relationship to opportunistically accelerate footprint.
Financing engine and cost of capital
CoreWeave closed the $8.5 billion DDTL 4.0 facility — the first-ever investment-grade (A- equivalent from Moody's, Fitch and DBRS) HPC-infrastructure-backed debt facility — priced at an implied cost below 6%, non-recourse to the parent, with an ABS-style draw feature unlocking an additional $1 billion upon contract stabilization.
It raised $2 billion of equity tied to the expanded NVIDIA relationship, and in Q2 secured more than $10 billion of additional debt and equity (convertible and high-yield offerings upsized), a $1 billion strategic investment from Jane Street, and priced a fifth DDTL facility (first syndicated in public loan markets) to finance OpenAI and Cohere contracts at 50 bps inside the initial marketing range — fully financing existing OpenAI commitments.
Total capital secured year-to-date exceeds $20 billion, and S&P moved the corporate outlook from stable to positive. Weighted average cost of debt has fallen ~600 bps from 2023 to 2025 and a further ~80 bps year-to-date, with no debt maturities until 2029 outside self-amortizing and vendor financing.
Margin trajectory and the timing dynamic
Adjusted operating income was $21 million (1% margin), described as the trough, versus $163 million a year ago, while adjusted EBITDA reached $1.2 billion (56% margin), up 91% year-over-year. Management framed the margin compression as timing-based, not economic: upon receiving a powered shell, CoreWeave incurs lease, power and depreciation costs during a 1-2 month fit-out with no revenue, running new deployments at negative contribution margin; by month three, deployments typically generate revenue with contribution margins normalizing to the mid-20s.
With active power nearly tripled since the start of 2025, management expects sequential margin expansion for the rest of the year, an inflection crossing Q2 to Q3, and a return to low-double-digit adjusted operating margin by Q4.