CoreWeave (CRWV) Report Interpretation

Overview

This earnings update argues that CoreWeave delivered a broad 2Q26 beat through faster capacity activation, strong demand, improved pricing and an outlook increase. Morgan Stanley retains its Equal-weight rating, judging that the stronger near-term execution is balanced by long-term delivery, financing and concentration risks.

Core views

Morgan Stanley characterizes the quarter as a beat on the metrics it considers most important: active-power delivery, profitability, contracted demand and guidance. Revenue was $2.575 billion, up 112% year on year, 0.5% above consensus and 2% above the midpoint of management’s guide. Adjusted operating-income margin was 5%, above Morgan Stanley and Street expectations of roughly 2-3%, while capex of $9.35 billion exceeded the $7-$9 billion guide and the firm’s and Street’s $8 billion expectations. The institution views the results as evidence that CoreWeave is executing in a supply-constrained environment rather than merely benefiting from demand. Capacity expansion was a central positive. CoreWeave added nearly 500 MW of net new active power during the quarter, taking active power above 1.5 GW versus consensus around 1.25 GW; this was the company’s largest quarterly addition and more than triple the year-earlier pace. Contracted power rose to 3.7 GW at quarter-end from 3.5 GW in Q1 and was about 4.2 GW quarter-to-date. Management reiterated a target of at least 8 GW by 2030. Morgan Stanley treats the power ramp as important because capacity availability is the operational constraint on converting AI-compute demand into revenue. Demand remained ahead of supply across the business, according to management. Backlog reached $104.2 billion, implying $7.4 billion of Q2 bookings, and excludes more than $25 billion of net new customer commitments signed early in Q3 across customers and contract sizes and durations. The report also notes sustained demand for older chip generations and strong renewal economics, even as contracts shift modestly toward shorter-duration and enterprise arrangements. It highlights an A100 contract extending into 2029 at an attractive price as support for continuing monetization of prior-generation GPUs. The report sees improving unit economics alongside scale. Contribution margins on Q2 contracts were said to be 5-10 percentage points better, and CoreWeave implemented a 25% across-the-board price increase in July. Morgan Stanley says these pricing actions, stronger contract margins and Vera Rubin economics support the longer-term margin case. Managed-inference platform booked ARR grew from $1 million to more than $100 million and is expected to reach at least $250 million exiting 2026, adding evidence of broader platform monetization. Management raised FY26 guidance: revenue by 2% to $12.4-$13.2 billion, ARR by 3% to $18.5-$19.5 billion, active power to more than 1.85 GW, and adjusted operating income by 6% to $960 million-$1.15 billion. FY26 capex increased by 12% at the midpoint to $35-$39 billion. Q3 guidance calls for $3.45-$3.60 billion of revenue, $200-$260 million of adjusted operating income and $11.5-$13.5 billion of capex. Revenue guidance is slightly above Street expectations, while adjusted operating income is slightly below the $262 million consensus and capex exceeds the Street’s $10 billion expectation. The lower implied Q3 margin, at 5.8%-7.2% versus 7.6% consensus, means CoreWeave needs a more meaningful Q4 margin ramp; management expects sequential expansion to the low-teens range in Q4. Morgan Stanley’s risk-reward framing balances a large GenAI data-center opportunity, estimated at $361 billion, and CoreWeave’s rapid scaling with a limited operating and financial history, expected debt of about $38 billion by end-CY26, and significant customer concentration. Its base-case range is $80-$131, based on 27-33x CY35 EV/FCF of $8.4-$8.5 billion, with CY35 revenue of $47-$78 billion, a 24%-31% CAGR, and operating margin of about 23%. The report states that the shares reasonably price a wide range of outcomes, which underpins the retained Equal-weight rating despite the strong operational update.

Analysis framework

Morgan Stanley evaluates the quarter by comparing revenue, margin, capex and active-power delivery with guidance and consensus, then links capacity, bookings, pricing and contract margins to the outlook for growth and profitability. It frames valuation through long-term EV/FCF scenarios and weighs the growth opportunity against execution, debt and customer-concentration risks.

Methodology notes

• Industry AnalysisSupply-demand framework

  AI-compute demand relative to available power and GPU data-center capacity

  The report assesses whether CoreWeave can convert strong demand into growth by tracking active and contracted power, backlog and management’s statement that demand exceeds supply.

• Valuation methodsEV/EBITDA valuation

  Long-term EV/FCF scenario valuation

  Morgan Stanley values bull, base and bear outcomes using enterprise value relative to projected CY35 free cash flow, rather than a near-term earnings multiple.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• CoreWeave (CRWV.O)

  Primary covered company; increased AI-compute capacity and demand conversion support growth, while long-term execution and financing needs constrain the rating.

  Strengths

  Active power above 1.5 GW, expanding contracted power, $104.2 billion backlog, stronger pricing and contribution margins, and growing managed-inference ARR.

  Weaknesses

  Limited operating and financial history, high capex requirements and a Q3 margin outlook below consensus.

  Comparison

  2Q26 active power exceeded Morgan Stanley/Street expectations of roughly 1.2 GW; the report says about 6x EV/CY27 revenue using CY27 net debt is broadly in line with peers.

  Risks

  Customer concentration, debt expected to expand to about $38 billion by end-CY26, AI-demand normalization, GPU useful-life pressure and difficulty raising capital at attractive debt costs.

Key data

• 2Q26 revenue$2.575 billionUp 112% year on year; 0.5% above consensus and 2% above the guide midpoint.
• Active power>1.5 GWVersus 1.25 GW consensus after nearly 500 MW of net new capacity in the quarter.
• Contracted power3.7 GW at quarter-end; ~4.2 GW quarter-to-dateUp from 3.5 GW in Q1.
• Backlog$104.2 billionImplies $7.4 billion of Q2 bookings and excludes more than $25 billion of early-Q3 commitments.
• FY26 revenue guidance$12.4-$13.2 billionRaised 2%.
• FY26 adjusted operating-income guidance$960 million-$1.15 billionRaised 6%.
• FY26 capex guidance$35-$39 billionMidpoint raised 12%.

Impact & implications

The report says the quarter strengthens the evidence for CoreWeave’s demand and execution momentum, particularly through power delivery and pricing. However, the required margin ramp, large capital needs, expected debt growth and customer concentration keep Morgan Stanley from taking a more positive relative-rating stance.

Risks

• AI demand could normalize.
• GPU instance pricing may not remain stable.
• Industry trends could shorten the useful life of AI servers.
• Large customers may choose to build AI-compute capacity in-house rather than lease it.
• Capital may be difficult to raise at an attractive or improving cost of debt.
• Customer concentration and limited operating and financial history remain material concerns.

What to watch

• Delivery against the more than 1.85 GW FY26 active-power target and the at-least-8-GW 2030 target.
• Conversion of the $104.2 billion backlog and more than $25 billion of early-Q3 commitments into revenue.
• The path from Q3’s guided 5.8%-7.2% adjusted operating margin to management’s expected low-teens Q4 margin.
• Pricing stability, contract contribution margins and monetization of older GPU generations.
• Managed-inference ARR progress toward at least $250 million exiting 2026.
• Customer decisions on leasing versus owning data centers and progress in reducing customer concentration.