CoreWeave demand and GPU pricing remain strong, while software monetization opens up margin potential, but Goldman Sachs maintains its Neutral rating

Overview

The report reviews CoreWeave's 2Q26 results and assesses its growth and profitability trajectory through the lenses of compute supply and demand, GPU pricing, data center expansion, and monetization of high-margin software. Goldman Sachs believes the company has a strong competitive position in AI infrastructure, but software and platform services remain at an early stage; it therefore maintains its Neutral rating while raising its price target.

Core views

CoreWeave's shares rose 19% following its August 12 earnings release. Revenue was in line with consensus, while the EBIT margin was approximately 200 basis points above Visible Alpha consensus; the company's 2026 revenue guidance was 1% above consensus, 4Q ARR guidance was 3% above consensus, and margin guidance was 40 basis points above consensus, although capital expenditure guidance was also 11% higher. Goldman Sachs attributes the market reaction to three key themes: demand continues to exceed available supply, pricing remains firm across both new and older GPU generations, and there is a credible path to margin expansion as new capacity comes online and serves a larger installed base. Demand and pricing are the report's first core theme. Revenue backlog reached $104.0 billion, up 5% quarter over quarter; although quarterly additions were more moderate than in previous quarters, the company signed more than $25.0 billion of additional commitments in early 3Q, indicating that momentum remained strong at quarter-end. Pricing for next-generation chips such as Blackwell and Vera Rubin has reached new highs, while pricing for older-generation products remains at or above levels from approximately one year ago. The company also recently signed an A100 contract extending through 2029. Near-term capacity is effectively sold out, and contracts signed in 2Q carry contribution margins 5% to 10% higher than recent contracts, while CoreWeave's target contribution margin is in the mid-20% range. The customer mix is also broadening, with a greater share from enterprise customers such as Caterpillar, IBM, Nissan, and ZF, which Goldman Sachs views as a positive signal. The second theme is that capacity development is progressing according to plan. Active power increased from 1GW in 1Q to more than 1.5GW; contracted power rose from 3.5GW to 3.7GW at quarter-end and reached 4.2GW on August 11, 2026. Management raised its FY26 active power target from more than 1.7GW to more than 1.85GW while reiterating its target of more than 8GW by 2030. This long-term target is supported by 4.2GW of contracted power and options on more than 1.5GW of powered land, with the first self-built site expected to come online later in 2026. Goldman Sachs believes new capacity deployments will not only unlock currently constrained demand but also improve operating leverage through a larger installed base. The third theme is monetization of high-margin software and platform services. CoreWeave's Managed Inference business expanded booked ARR from $1 million to more than $100 million within several months, and management expects it to reach $250 million by the end of 2026, although current growth remains capacity constrained. ARR from higher-margin platform services, including CPU, storage, networking, and software, has exceeded $400 million and is expected to continue growing as customers consolidate more spending on the CoreWeave platform. CoreWeave Omni, which allows customers to use their own GPUs and deploy CoreWeave's cloud stack in their own data centers, has signed its first contract, with revenue expected to begin scaling in 2027. Goldman Sachs believes these businesses can increase ARR per MW and provide additional high-margin monetization channels beyond core infrastructure, but more evidence of scale is still needed. Goldman Sachs revised its forecasts based on the 10-Q and updated management guidance. FY26/27/28 revenue forecasts were raised from $12.9 billion, $26.5 billion, and $38.0 billion to $13.0 billion, $27.2 billion, and $44.2 billion, respectively; capital expenditure forecasts for the same periods were increased substantially from $35.0 billion, $41.7 billion, and $43.0 billion to $39.1 billion, $57.0 billion, and $58.5 billion. The new FY26/27/28 EPS forecasts are -$3.64, -$2.90, and -$1.15, respectively, compared with -$3.80, -$0.07, and $2.24 previously. The forecast table shows revenue growth slowing from 152.5% in FY26 to 110.3% in FY27 and 62.2% in FY28, while the EBIT margin rises from 8.2% to 15.4% and 18.2%. Net debt/EBITDA is expected to decline from 5.7x in FY26 to 2.7x in FY28, but free cash flow yields over the same period remain -48.1%, -45.3%, and -28.1%, respectively, reflecting the continued substantial capital investment required for expansion. On valuation, Goldman Sachs maintains its Neutral rating and raises its 12-month price target from $121 to $139. The new price target applies a 22x enterprise value/next-12-month non-GAAP EBIT multiple, compared with 24x previously; the lower multiple reflects declining peer valuations. The price target corresponds to 17x 2028 EBIT, consistent with the previous methodology. Overall, Goldman Sachs continues to view CoreWeave as one of the best-positioned AI infrastructure providers, but it remains Neutral until software and platform services become more meaningful sources of growth and profitability.

Analysis framework

Goldman Sachs first compares actual 2Q results and management guidance with consensus expectations, then assesses demand strength through revenue backlog, new commitments, pricing across GPU generations, and customer mix. It subsequently evaluates capacity execution using active power, contracted power, and long-term land resources, while assessing how Managed Inference, platform services, and Omni could increase revenue per MW and margins. Finally, the report updates revenue, capital expenditure, and earnings forecasts based on the 10-Q and new guidance and determines the price target using a next-12-month non-GAAP EBIT multiple.

Methodology notes

• Industry/Sector Analysis FrameworkSupply-demand framework

  AI Compute Supply-Demand and Pricing Analysis

  The report uses revenue backlog, quarterly new commitments, the sellout status of near-term capacity, and pricing for new and older GPUs to assess whether demand continues to exceed supply and, based on this, explains pricing resilience and improving margins on new contracts.

• Valuation Method

  EV/EBIT Valuation

  Goldman Sachs applies a 22x enterprise value/next-12-month non-GAAP EBIT multiple to CoreWeave's earnings forecast to derive a 12-month price target of $139; the multiple was reduced from 24x previously to reflect declining peer valuations.

Asset mapping & comparison

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

• CoreWeave Inc. (CRWV.US)

  AI compute demand exceeding supply, capacity expansion, and software monetization collectively support revenue growth and potential margin improvement.

  Strengths

  Revenue backlog has reached $104.0 billion, near-term capacity is essentially sold out, and pricing for both new and older GPUs remains firm; active and contracted power continue to increase, the enterprise customer mix is broadening, and ARR from Managed Inference and platform services is growing rapidly.

  Weaknesses

  Expansion requires substantial capital expenditure, and free cash flow yield is expected to remain negative throughout the forecast period; software and platform services are still at an early stage and have not yet demonstrated that they can become sufficiently meaningful sources of growth and profitability.

  Comparison

  Due to declining peer valuations, the price target valuation multiple was reduced from 24x to 22x next-12-month non-GAAP EBIT; the price target corresponds to 17x 2028 EBIT.

  Risks

  A decline in demand for generative AI compute, intensified competition from hyperscalers and neocloud providers, disruptions to data center equipment procurement, and an inability to secure additional financing at competitive interest rates.

Key data

• Share Price Performance After Earnings ReleaseUp 19%Following the release of 2Q26 results on August 12, 2026
• Revenue Backlog$104.0 billionUp 5% quarter over quarter
• New Commitments in Early 3QMore than $25.0 billionSigned after the end of 2Q
• Contribution Margin on New 2Q Contracts5%-10% higher than recent contractsThe company's target contribution margin is in the mid-20% range
• Active PowerMore than 1.5GW1GW in 1Q
• Contracted Power4.2GWAs of August 11, 2026; 3.7GW at quarter-end, compared with 3.5GW previously
• FY26 Active Power TargetMore than 1.85GWPreviously more than 1.7GW
• 2030 Active Power TargetMore than 8GWManagement reiterated the target
• Booked Managed Inference ARRMore than $100 millionExpanded from $1 million within several months
• Managed Inference ARR Target by Year-End 2026$250 millionGrowth is currently capacity constrained
• High-Margin Platform Services ARRMore than $400 millionIncludes CPU, storage, networking, and software services
• FY26/27/28 Revenue Forecasts$13.0 billion/$27.2 billion/$44.2 billionPreviously $12.9 billion/$26.5 billion/$38.0 billion
• FY26/27/28 Capital Expenditure Forecasts$39.1 billion/$57.0 billion/$58.5 billionPreviously $35.0 billion/$41.7 billion/$43.0 billion
• FY26/27/28 EPS Forecasts-$3.64/-$2.90/-$1.15Previously -$3.80/-$0.07/$2.24
• 12-Month Price Target$139Previously $121; rating maintained at Neutral
• Price Target Valuation Basis22x next-12-month non-GAAP EBITPreviously 24x; the price target still corresponds to 17x 2028 EBIT

Impact & implications

The report believes that a strong backlog, sold-out near-term capacity, and pricing resilience across GPU generations improve visibility into revenue and margin realization, while expanding power resources support medium- to long-term growth. Managed Inference, platform services, and Omni could increase revenue per MW and improve the business mix, but the significantly higher capital expenditure forecasts, persistently negative free cash flow, and funding requirements mean that software monetization and operating leverage still require further validation.

Risks

• Upside scenarios include faster AI adoption, causing compute demand to continue exceeding supply.
• Sustained technological differentiation in future GPU generations could generate greater-than-expected upside.
• Achieving positive free cash flow sooner could lead to better-than-expected performance.
• Stronger-than-expected demand for CoreWeave's broader software solutions could improve growth and margins.
• A decline in overall demand for generative AI compute is the primary downside risk.
• Intensifying competition from hyperscalers and neocloud providers could constrain growth or pricing.
• Disruptions to data center equipment procurement could delay capacity deployment.
• The company may have difficulty raising additional capital at competitive interest rates.

What to watch

• Monitor whether software and platform services can become more meaningful contributors to growth and profitability and increase ARR per MW.
• Monitor whether Managed Inference ARR can reach $250 million by the end of 2026 despite capacity constraints.
• Monitor whether CoreWeave Omni's first contract can begin generating revenue at scale in 2027 as expected.