CoreWeave Earnings - May 2026
CoreWeave, Inc.
Executive Summary
Bottom Line: $99 billion backlog (up 50% in one quarter), revenue more than doubles, but the adjusted loss missed expectations by approximately 24% and 2026 capex moves higher. Hyperscale arrived; profitability didn't.
CoreWeave, Inc. (NASDAQ: CRWV) reported Q1 2026 results on May 7, 2026. Revenue of $2.078 billion grew 112% year-over-year and beat the $1.97 billion consensus by approximately $110 million. However, the GAAP net loss widened to -$740 million (vs. -$315 million prior year). GAAP EPS came in at -$1.40 (vs. -$1.49 prior year), and adjusted EPS of -$1.12 missed the -$0.90 LSEG consensus by approximately $0.22 (~24%). Adjusted EBITDA of $1.157 billion (56% margin) and operating cash flow of $2.984 billion remained strong, but interest expense at $536 million continued to expand alongside the debt-funded buildout.
The defining metric was the revenue backlog of $99.4 billion, up nearly 50% sequentially from $66.8 billion at year-end 2025 and close to 4x the prior-year level. CEO Mike Intrator: This was the strongest bookings quarter in CoreWeave's history. Active power surpassed 1 GW for the first time, contracted power expanded to 3.5 GW, and 36% of backlog is expected to convert to revenue within the next 24 months. Q2 2026 revenue guidance of $2.45-$2.6 billion (midpoint $2.53B) trailed the $2.69 billion LSEG consensus, and 2026 capex was raised to $31-$35 billion.
Our take: A bifurcated print that perfectly captures the CoreWeave investment debate. The revenue beat, the $99.4 billion backlog, the 1 GW active-power milestone, and the 10 customers committed to spending $1B+ each are all extraordinary data points, validating the AI-infrastructure-demand thesis. But the wider-than-expected adjusted loss (-$1.12 vs. -$0.90 consensus, a ~24% miss), the higher capex range, the soft Q2 revenue guide versus consensus, and continued elevated interest expense ($536M in Q1) keep the path-to-profitability question open. CRWV closed regular session May 7 at $128.84 (-6.62% on the day) on insider-selling concerns, then fell as much as -10% in extended trading after the print, settling near $116.90 by 8 PM. Mike Intrator's we have reached hyperscale framing matters: the question is no longer if CoreWeave can scale, but whether margins can ramp from low-single-digits in Q1 to low-double-digits by Q4 2026 as guided.
Headline KPIs at a Glance
Revenue and Profitability
Q1 2026 revenue of $2.078 billion grew 112% year-over-year and 32% sequentially from $1.568 billion in Q4 2025, beating the $1.97 billion LSEG consensus by approximately $110 million. The beat reflected continued strong execution in deploying capacity, with 1 GW of active power achieved during the quarter (vs. 850 MW at year-end 2025) and 10 customers now committed to spending at least $1 billion each on CoreWeave Cloud.
How the Print Looked Against Consensus
The bottom line was meaningfully weaker than expected. Adjusted EPS of -$1.12 missed the -$0.90 LSEG consensus by $0.22 (approximately 24% surprise miss). GAAP EPS was -$1.40 (vs. -$1.49 prior year). Net loss widened to -$740 million from -$315 million in Q1 2025. Adjusted net loss of -$589 million reflects ongoing margin compression as new capacity ramps before generating full-quarter revenue. Adjusted EBITDA of $1.157 billion (56% margin) was solid, and operating cash flow of $2.984 billion underscored the underlying cash-generation capability of the contracted-revenue model. Interest expense, net, of $536 million was in line with the $510-$590 million guide and represents the principal drag versus expectations.
The $99 Billion Backlog
Revenue backlog (commercial remaining performance obligations) ended Q1 2026 at $99.4 billion, up nearly 50% sequentially from $66.8 billion at year-end 2025 and close to 4x year-over-year. Per CFO Nitin Agrawal, the backlog is near-term weighted, with 36% expected to be recognized as revenue in the next 24 months and 75% in the next 4 years. The weighted-average contract length for new capacity remains approximately 5 years.
Customer Diversification
Customer diversification is now a board-level KPI, not just a narrative. CoreWeave disclosed:
- 10 customers committed to spending $1 billion+ each on CoreWeave Cloud (vs. essentially Microsoft-concentrated revenue in 2024, when 62% of revenue came from one customer).
- Less than 30% of total backlog now comes from non-investment-grade AI-native companies and foundation labs: a meaningful improvement in credit quality.
- Major April commitments confirmed and incorporated: Meta's $21 billion expansion through December 2032, Anthropic's multi-year multi-billion deal, and Jane Street's $6 billion contract with a $1 billion equity investment.
- New customer wins disclosed this quarter: Cohere and Mistral, alongside expanded relationships across the existing AI-native and enterprise base.
CoreWeave's contract structure ties forward revenue to customer take-or-pay commitments. The $99.4B backlog represents approximately 8x trailing twelve-month revenue and provides multi-year revenue visibility that few cloud businesses have ever had. Whether that backlog converts to revenue on the schedule the model assumes, however, is now the single-most-important execution question.
Capacity, Capex, and the Financing Story
CoreWeave surpassed 1 GW of active power during Q1 2026 (1.01 GW exactly), up from 850 MW at year-end 2025. Total contracted power expanded to 3.5 GW (from 3.1 GW at year-end). The long-term target was raised to more than 8 GW by 2030 (up from 5+ GW previously contracted). CEO Mike Intrator: We have reached hyperscale.
Q1 2026 capex was $7.695 billion, on track with the prior $6-$7B Q1 guide. However, full-year 2026 capex was raised to $31-$35 billion, with the low end moving up from $30 billion (cited by CFO Agrawal as a result of higher component prices). Intrator framed it as a deliberate decision: It's an issue, it's a problem, but we have an incredible capacity to navigate the supply chain. Property and equipment on the balance sheet stood at $36.4 billion as of March 31, 2026.
Capital Structure
- $8.5 billion DDTL 4.0 facility: New non-recourse delayed-draw term loan, investment-grade rated, closed in Q1 to fund continued buildout.
- $2 billion NVIDIA equity investment: Class A common stock investment closed during Q1, doubling NVIDIA's prior stake in CoreWeave.
- Meta-backed financing: A unique debt instrument backed by GPUs and Meta's contracted AI-compute payments was closed at quarter-end.
- S&P credit outlook upgraded to positive from stable per CFO Agrawal.
Forward Guidance and Surprises
Forward Guidance
| Metric | FY 2026 / Q2 Guidance | Change vs. Prior |
|---|---|---|
| Q2 2026 Revenue | $2.45B-$2.60B (mid $2.53B) | Below $2.69B consensus |
| FY 2026 Revenue | $12B-$13B (~140% YoY at midpoint) | Maintained |
| FY 2026 Capex | $31B-$35B | Low end raised from $30B |
| FY 2026 Adj. Op. Income | $900M-$1.1B | Q1 trough at low single digit margin |
| FY 2026 Margin Trajectory | Low-single-digits Q1 to low-double-digits Q4 | Reaffirmed |
| Long-Term Margin Target | 25-30% mature contribution | Reaffirmed |
| FY 2027 Run Rate | >$30B annualized exiting 2027 | Reaffirmed |
| Year-End 2026 Active Power | >1.7 GW | Reaffirmed |
Key Surprises and Stock-Price Implications
- Backlog at $99.4B. Up 50% sequentially. Up from $66.8B at year-end and approximately $50B increase YoY. The most positive single data point in the print.
- Adjusted EPS missed by ~24%. -$1.12 vs. -$0.90 LSEG consensus, a $0.22 miss. The single most negative data point. Drove the after-hours sell-off.
- Q2 revenue guide soft. Midpoint of $2.53B below the $2.69B consensus. Reinforced the gap between bookings and revenue recognition timing.
- FY26 capex floor raised to $31B. $1B incremental at the low end. Cited as component-price driven, not strategy-shift driven.
- 1 GW active power crossed. Operationally significant milestone. Validates the build-and-deliver execution narrative.
- 10 $1B+ customers. Customer-concentration risk meaningfully de-risked vs. the 62%-Microsoft structure of 2024.
- Long-term capacity target raised to 8+ GW. Previously 5+ GW. Reflects pipeline building beyond current contracted base.
- S&P outlook upgrade. Positive from stable. Modest but credit-positive signal during a heavy borrowing cycle.
Stock Reaction and Wall Street Sentiment
CRWV had a volatile May 7 even before earnings: the stock fell -6.62% during the regular session to close at $128.84, reportedly on insider-selling concerns and broader AI-infrastructure spending worries. After the earnings release at 4:05 PM ET, the stock continued to decline in extended trading, falling as much as -10% to approximately $116.90 by 8:00 PM ET. Year-to-date through May 7's regular close, CRWV was up approximately 80% while the S&P 500 had gained roughly 7%. The 52-week range of $49.06 to $187.00 illustrates the extraordinary volatility this name carries.
Analyst Sentiment
- Aggregate consensus: Moderate Buy / Buy. Of 35 covering analysts (Yahoo Finance pre-print), 11 Buy and 4 Hold, with no Sell ratings. Average target $131.06 (Yahoo); $133.21 (MarketBeat).
- Jefferies $160 target maintained pre-print (Buy), citing the post-Q4 deal flow. Likely to be tested in post-print revision cycle.
- Vanguard: Largest institutional holder at approximately 5.84%. Vanguard Total Stock Market ETF holds ~2.08%; Vanguard Mid-Cap ETF holds ~1.41%.
- NVIDIA: Closed $2 billion equity investment during Q1, signaling continued strategic alignment.
- Q1's strongest-bookings-quarter framing may motivate selective target raises in coming days, but the soft Q2 guide and the EPS miss are likely to keep aggregate revisions mixed.
Pricing Environment
Pricing power has held: average H100 GPU pricing in Q1 was within 10% of where it started 2025, and average A100 pricing actually increased in 2025. Demand for prior-generation GPU architectures remains robust, primarily for inference use cases. NVIDIA chips overall remain in short supply per Intrator. CoreWeave is targeting H2 FY26 deployment of NVIDIA Rubin platforms; first-mover advantage on Rubin is a meaningful pricing-power lever for 2027 revenue.
Key Risks and What to Watch
Primary Risks
- Margin trajectory execution. Q1 was the guided trough at low-single-digit operating margin; the path to low-double-digits by Q4 2026 is now the cleanest quarterly proof-point. Any meaningful deceleration in this ramp directly impairs the long-term 25-30% margin thesis.
- Capex inflation. FY26 capex floor moved from $30B to $31B citing component prices. Continued inflation in GPU, power, or data-center construction inputs could pressure returns on the contracted backlog.
- Customer concentration legacy. While diversification has improved meaningfully (10 $1B+ customers, less than 30% of backlog from non-IG AI natives), single-customer concentration remains a structural risk if any major customer consolidates spending or pivots to in-house capacity.
- Capital structure leverage. Interest expense at $536M in Q1 alone, plus continued aggressive borrowing through DDTL 4.0 ($8.5B) and GPU-backed instruments. Sensitive to rate environment and to any covenants triggered by margin softness.
- AI capex deceleration. CRWV is the highest-beta name in the AI infrastructure complex. Any meaningful sign that hyperscaler or foundation-model capex is decelerating would compress CRWV multiples disproportionately.
Catalysts to Watch
- Q2 2026 print (early August). First test of revenue-recognition velocity against the $2.45B-$2.60B guidance and the margin-ramp trajectory toward Q4.
- NVIDIA Rubin deployment (H2 2026). CoreWeave is targeting first-mover status. New SKUs typically ship at premium pricing. Rubin-attached contracts could meaningfully expand backlog.
- Cloud-stack licensing. Management indicated proprietary cloud stack may be licensed beyond CoreWeave-operated data centers starting in 2026; explicitly excluded from FY26 guidance, so any contribution would be incremental.
- New $1B+ customer adds. Adding to the 10 $1B+ customers is the cleanest concentration-risk-reduction signal.
- Annualized run-rate milestones. Exit 2026 at $17B-$19B annualized; exit 2027 at $30B+ annualized. Each quarter's pace check is a live indicator.
Bottom Line for the Holder
A messy print with two diametrically opposed signals. The $99.4B backlog (up 50% in one quarter), 1 GW active-power milestone, 10 $1B+ customers, and $2 billion NVIDIA equity investment are extraordinary validation of the AI-infrastructure-demand thesis. The $0.22 adjusted-EPS miss (~24%), $1B incremental capex floor, and soft Q2 revenue guide are equally clear near-term execution warnings. The stock's two-stage decline (regular session -6.62% on insider selling, then extended trading down to ~-10% post-earnings) is appropriate for the EPS shortfall but understates the strategic significance of the backlog. For long-term holders with high risk tolerance and conviction in the multi-year AI buildout, the print is consistent with thesis. For near-term traders, expect continued volatility through the margin-ramp quarters of FY26.
Ready to Get Started? Create Your Customized Financial Game Plan.
Before we can build a plan to help you meet your financial goals, we’ll take the time to get to know you and your financial vision. In this short exercise, answer questions about yourself and your future objectives. Then, request a consultation so that together, we can build a plan to help you get there.
