CoreWeave, Inc. (CRWV) stock analysis

Business Model

CoreWeave rents specialized AI cloud capacity for training, inference, data movement, and agentic workflows. Its platform combines GPU compute, high-speed networking, storage, orchestration and observability software, managed services, and developer tools.

Customers include AI labs, hyperscalers, startups, enterprises, and public-sector or defense buyers that need large, rapidly deployable accelerator clusters. Revenue depends on deployed power, accelerator availability, utilization, contract duration, workload pricing, and mix. It is primarily supported by committed, often take-or-pay contracts; on-demand usage is secondary. In its Q2 2026 earnings release, CoreWeave reported $104.2 billion of revenue backlog.

Market Opportunity

CoreWeave monetizes rented AI cloud capacity and related software, not the full AI-infrastructure stack of chips, servers, power equipment, real estate, or enterprise AI spending.

Gartner estimates worldwide end-user spending on AI-optimized IaaS at $37.5 billion in 2026, with inference representing 55% of that total. Inference is more continuous than episodic training, but it also raises reliability, latency, and cost-performance requirements.

IDC forecasts broader AI infrastructure spending of $487 billion in 2026 and more than $1 trillion by 2029. That supports the demand backdrop but overstates CoreWeave’s directly serviceable market because it includes categories customers may buy or build outside its cloud.

GPU-as-a-Service forecasts show definitional uncertainty. Market research firms Grand View Research, Mordor Intelligence, and Fortune Business Insights estimate 2026 market size from $5.1 billion to $8.66 billion, with CAGRs from 16.0% to 44.3%.

Power and data-center scarcity affect pricing, margins, and execution. CBRE reported Q1 2026 vacancy of 0.3% in Northern Virginia and 1.0% in Atlanta. JLL estimates global data-center capacity could add 100 GW by 2030 and require roughly $3 trillion of total expenditures. The IEA projects data-center electricity consumption rising from 485 TWh in 2025 to 950 TWh in 2030.

Regionally, the U.S. remains CoreWeave’s main scale market and the largest cloud market. Europe matters for sovereignty, latency, and renewable-power positioning, while Asia-Pacific is a growth vector for AI capacity. CoreWeave’s filings do not disclose regional revenue, limiting assessment of contribution or margin differences by geography.

Position, Advantages & Constraints

CoreWeave is advantaged when customers value speed, dense GPU clusters, and dedicated capacity more than hyperscaler bundling. Its disclosed wins with Meta and Anthropic support credibility with large AI buyers.

Competition remains structurally difficult. Synergy Research Group estimated Q1 2026 cloud infrastructure services spending at $128.6 billion, with AWS, Microsoft, and Google holding 28%, 21%, and 14% global share, respectively, and 67% combined public-cloud share. Those companies can self-build, bundle services, absorb pricing pressure, and act as both partners and competitors.

Execution depends on GPUs, networking gear, liquid cooling, financing, and rapid accelerator cycles. Customer concentration compounds that exposure: one customer represented 67% of 2025 revenue, making reported growth sensitive to a small number of counterparties.

Leadership is relevant because execution is infrastructure- and finance-intensive. Co-founder Michael Intrator has led the company since 2017, and CFO Nitin Agrawal joined in 2024 after finance roles at Google Cloud, Mapbox, and AWS.

Financial Snapshot

Performance & Profitability

CoreWeave continues to scale rapidly, but Q2 2026 still reflected buildout economics rather than mature GAAP profitability. Revenue more than doubled year over year, and each quarter from Q3 2024 through Q2 2026 reached a new high, pointing to capacity- and demand-driven expansion rather than seasonality.

Gross margin remains the strongest part of the model, though below prior peaks. That suggests attractive unit economics from scarce compute capacity, with incremental pressure from infrastructure costs, depreciation, customer mix, or less favorable scale economics.

Operating leverage improved sequentially from Q1 2026 but remained unproven: Q2 operating margin was still negative despite the revenue step-up. Gross profit continues to be absorbed by capacity expansion, platform operations, headcount, public-company costs, depreciation, and other scale-related spending.

Diluted EPS has been negative in every reported quarter. The latest sequential improvement in loss per share is constructive, but depreciation, lease costs, and financing expense are core costs of delivering AI infrastructure at scale, not incidental items.

Negative ROE and ROA reinforce that the capital base is not yet producing positive accounting earnings. ROA is especially important because value creation depends on long-lived compute and data-center assets earning returns above financing and replacement costs.

Margin readings can differ depending on whether they use consolidated net income or income attributable to common stockholders. That affects comparability, not the conclusion.

Balance Sheet & Resilience

The balance sheet is the main financial constraint. Cash is meaningful in absolute terms, but current liabilities substantially exceed current assets, working capital is deeply negative, and liquidity ratios indicate limited short-term flexibility.

Leverage is aggressive: liabilities absorb most of the asset base, leaving a thin equity cushion while losses persist. Debt and capital lease obligations support rapid capacity buildout when utilization remains strong, but increase downside if growth slows or refinancing conditions tighten.

Solvency depends on earnings improvement and continued access to capital. Interest coverage is negative, so current operating earnings do not support interest expense on a conventional coverage basis. Until operating income turns sustainably positive, CoreWeave’s ability to carry the capital structure relies on cash balances, customer receipts, capital-market access, and lender or lessor confidence.

The asset base provides some resilience, but it is not a complete safety net. AI infrastructure assets can generate substantial revenue if utilization, pricing, and contract duration hold up. They can also depreciate quickly, become less competitive as accelerator generations change, or lose value if power, data-center capacity, or customer demand shifts.

Revolver availability, covenant headroom, and a detailed debt maturity schedule are missing from the filing narrative, limiting the ability to date the refinancing wall or quantify covenant cushion.

Financial Health Scores & Insider Activity

Financial Health Scores

The underlying nine Piotroski criteria are not disclosed, so the composite score cannot be decomposed.

Recent Insider Activity

Valuation Signals

Quote-Linked Snapshot

Implied Expectations & Margin of Safety

CoreWeave’s valuation is priced more on AI-compute scarcity and future cash-flow conversion than on current GAAP earnings support. As noted in Financials, the key valuation question is whether adjusted profitability can bridge to GAAP earnings and free cash flow quickly enough to justify revenue and enterprise-value multiples.

The margin of safety is limited. A premium to book value and no dividend leave few conventional anchors, so execution must carry the valuation case.

What Would Change the View

Rerating catalysts

• Faster conversion of deployed capacity into high-utilization revenue, with evidence that EBITDA gains are becoming GAAP profit and free cash flow.
• Customer diversification, longer commitments, or refinancing progress that lowers the discount rate investors apply to the model.

Compression risks

• Slower demand absorption, weaker pricing, or renegotiated customer economics that lower expected revenue per unit of capacity.
• Higher capital needs from deployment delays, power constraints, supplier bottlenecks, accelerator refresh cycles, or hardware obsolescence.

Management Guidance & Strategic Priorities

CoreWeave’s strategy remains demand-led but capital-intensive. After Q2 2026, management raised revenue expectations and the capex needed to deliver capacity.

Management has not provided adjusted EBITDA, EPS, or margin guidance. Future filings therefore need to show whether utilization and gross profit can absorb depreciation and interest expense as new capacity comes online.

Competitive Position

As noted in Company Intro, CoreWeave’s differentiation rests on specialization: GPU-dense clusters, AI-workload optimization, fast deployment cycles, and close alignment with the NVIDIA ecosystem.

The constraint is that scarcity is contestable. Hyperscalers can fund larger builds, bundle services, and steer customers toward proprietary chips; specialist GPU-cloud rivals are also competing for power and accelerators. Power-site competition, including Core Scientific’s AMD-related activity, also shows that alternative accelerator ecosystems are developing.

Customer power remains the key competitive weakness. The Q2 2026 results showed the top three customers accounted for about 72% of revenue. Expansion with Meta, OpenAI, Anthropic, Jane Street, enterprise inference workloads, and government opportunities such as the Leidos partnership would improve visibility only if concentration declines and contribution margins hold.

Supplier exposure is the operational counterpart: NVIDIA dependence supports current positioning but creates risk from supply disruption, pricing pressure, export controls, or shifts toward AMD and custom silicon alternatives.

Catalysts & Risks

The main strategic checkpoint is backlog conversion. The Q2 2026 results showed backlog of roughly $104B, versus FY 2026 revenue guidance in the table. Downside signals would include backlog revisions, customer deferrals, or capex rising faster than revenue.

Capacity delivery is the next operating test. Active power and deployed GPU capacity need to rise on schedule, including next-generation NVIDIA platforms such as Rubin/Vera Rubin. Grid access, permitting, construction, networking, or energization delays would defer revenue while capital costs continue.

As noted in Financials, refinancing and operating cash-flow improvement remain central to whether expansion becomes equity-accretive.

Technical Snapshot

As of: 2026-08-14

Trend & Momentum

CRWV’s near-term trend is constructive after a volatile rebound, while the limited trading history leaves the longer-term trend less defined.

Momentum remains mixed: RSI is neutral and weekly MACD is still bearish. Given high volatility, the support/resistance range is best treated as a risk-control frame rather than evidence of a durable trend.

Investment Thesis

CoreWeave is a high-upside, high-risk way to own AI compute scarcity. The equity case depends on proving that rapid capacity deployment can translate into durable GAAP profitability, free cash flow, and acceptable returns on long-lived infrastructure assets.

Bull Case

• Contracted demand supports rapid scaling if deployments arrive on time and clusters remain highly utilized.
• Specialized AI infrastructure may allow CoreWeave to outperform broader cloud platforms in performance-sensitive workloads.
• Operating leverage improves if revenue growth outpaces depreciation, financing costs, and recurring capital needs.

Bear Case

• Capital intensity prevents strong adjusted profit from becoming strong equity returns.
• Expanding AI compute supply compresses pricing.
• Customer concentration creates exposure to renegotiation, slower workload ramps, or migration away from the platform.

Thesis Scorecard

Thesis Breakers

• A major customer reduces commitments, delays workload ramps, or renegotiates pricing materially lower.
• Refinancing conditions tighten while near-term obligations remain high.
• GPU supply loosens or alternative silicon reduces CoreWeave’s scarcity premium.

Catalysts

• Backlog converts into revenue faster than expected with high utilization.
• GAAP margins and free cash flow improve as deployed capacity matures.
• Liquidity improves through refinancing, better working capital, or strategic capital at favorable terms.

CoreWeave offers real exposure to AI infrastructure demand, but the risk/reward is stretched because the market is already capitalizing a large share of future success. The stock looks rich, and the thesis needs visible profit conversion to become more balanced.

Filings & Company Disclosures

• SEC — CoreWeave 2025 Form 10-K — Mar. 2, 2026
• CoreWeave Investor Relations — CoreWeave Reports Strong First Quarter 2026 Results — May 7, 2026
• CoreWeave SEC filing — Q1 2026 Form 10-Q business and revenue model. (sec.gov)
• CoreWeave Q2 2026 earnings release and 8-K exhibit, supporting Q2 revenue, backlog, power expansion, financing actions, and current reported period. (sec.gov)
• CoreWeave Q2 2026 Form 10-Q, supporting debt, leases, customer concentration, power risk, supplier/NVIDIA dependence, competition, and capital investment discussion. (sec.gov)
• CoreWeave press release index for 2026 product, MLPerf, Gartner, Sweden, and platform milestones. (coreweave.com)

Market Data & Reproducible Calculations

• Quarterly income statements: Financial Modeling Prep /income-statement?symbol=CRWV&period=quarter; GAAP revenue, net income, EPS, EBITDA, and margin numerators in the Financial Snapshot.
• Quarterly balance sheets: Financial Modeling Prep /balance-sheet-statement?symbol=CRWV&period=quarter; liquidity, leverage, cash, debt, and book-equity inputs documented in the Sources and Methodology section.
• Quarterly cash-flow statements: Financial Modeling Prep /cash-flow-statement?symbol=CRWV&period=quarter; operating cash flow and capex where cited.
• Financial health scores: Financial Modeling Prep /financial-scores?symbol=CRWV (Altman Z-Score and Piotroski score).
• Insider transactions: Financial Modeling Prep /insider-trading/search?symbol=CRWV.
• Computed financial ratios: Liquidity, leverage, return, and margin figures in the Financial Snapshot are sourced from FMP /ratios-ttm and /ratios?period=quarter when available; otherwise they are calculated locally from the filing inputs above. Formulas appear in the Sources and Methodology section at the end of the report.
• Share price (August 14, 2026) and quote timestamp (August 14, 2026, 08:00 PM UTC): Financial Modeling Prep /quote endpoint, symbol=CRWV.
• Trailing revenue, GAAP net income, and EBITDA: sum of the latest four quarterly filings via FMP /income-statement (symbol=CRWV, period=quarter), ending June 30, 2026.
• Cash, debt, book equity, and period-end share count: latest quarterly balance sheet via FMP /balance-sheet-statement (symbol=CRWV, period=quarter), as of June 30, 2026.
• Company filings: Financial Modeling Prep /sec-filings-search/symbol?symbol=CRWV.
• Earnings releases: Financial Modeling Prep /news/press-releases?symbols=CRWV.
• Contemporaneous reporting of company outlook: Financial Modeling Prep /news/stock?symbols=CRWV.
• Peer set: Financial Modeling Prep /stock-peers?symbol=CRWV.
• Weekly adjusted prices: Financial Modeling Prep /historical-price-eod/full?symbol=CRWV resampled to weekly adjusted closes.
• 13-week support and resistance: Computed locally from weekly adjusted closes (13-week min/max of adjusted close).
• RSI / MACD: Computed locally from the same weekly adjusted-close series (14-week RSI; 12/26/9 weekly MACD).

Third-Party Research & News

• CoreWeave — About Us — accessed Aug. 11, 2026
• CoreWeave — Mike Intrator Executive Leadership — accessed Aug. 11, 2026
• CoreWeave — Nitin Agrawal Executive Leadership — accessed Aug. 11, 2026
• CoreWeave — Why CoreWeave Cloud — accessed Aug. 11, 2026
• Nasdaq — CoreWeave to Join Nasdaq-100 Index — Jun. 12, 2026
• CoreWeave — CoreWeave Named a Visionary in the 2026 Gartner Magic Quadrant for Cloud AI Infrastructure — Jul. 9, 2026
• CoreWeave Investor Presentation — March 2026 financing, backlog, data-center footprint, and platform positioning. (s205.q4cdn.com)
• Gartner — AI-optimized IaaS spending forecast. (gartner.com)
• IDC — AI infrastructure spending forecast. (idc.com)
• Grand View Research — GPU-as-a-Service market forecast. (grandviewresearch.com)
• Mordor Intelligence — GPU-as-a-Service market forecast. (mordorintelligence.com)
• Fortune Business Insights — GPU-as-a-Service market forecast. (fortunebusinessinsights.com)
• Synergy Research Group — Q1 2026 cloud infrastructure market share and spending. (srgresearch.com)
• CBRE — Global Data Center Trends 2026. (cbre.com)
• JLL — 2026 Global Data Center Outlook. (jll.com)
• IEA — Key Questions on Energy and AI, 2026 update. (iea.org)
• CoreWeave customer and strategic deal releases, including Meta, Anthropic, Jane Street, and OpenAI. (coreweave.com)
• Leidos and CoreWeave collaboration announcement for defense, national security, and intelligence AI workloads. (leidos.com)
• Core Scientific merger termination and AMD/Core Scientific infrastructure partnership, supporting the competitive power-capacity risk. (investors.corescientific.com)

Methodology Notes

Financial Snapshot Methodology:

• Balance-sheet date: June 30, 2026 (Q2 2026).
• Snapshot liquidity ratios (as of June 30, 2026): Sourced from FMP /ratios-ttm.
• Liquidity trend charts: Sourced from FMP /ratios?period=quarter (period ratios by quarter, not TTM).

Working capital remains computed from GAAP balance-sheet line items on that date (total current assets − total current liabilities).

• Debt: Total debt = short-term debt + long-term debt as reported on the balance sheet. Finance leases, project financing, and other obligations are included only if they appear in the short-term and long-term debt line items on the balance sheet.
• Snapshot debt-to-equity and interest coverage (as of June 30, 2026): Sourced from FMP /ratios-ttm.
• Leverage trend charts: Sourced from FMP /ratios?period=quarter (period ratios by quarter, not TTM).
• Snapshot margins (as of June 30, 2026): Sourced from FMP /ratios-ttm.
• Margin trend charts: Sourced from FMP /ratios?period=quarter (period margins by quarter, not TTM).
• ROE / ROA trend charts: Sourced from FMP /ratios?period=quarter (period returns by quarter, not TTM).
• Precision: Ratios use FMP provider figures when available; otherwise GAAP line items and dates shown above. Values are rounded for display; underlying filings may differ in classification or decimal precision.

Valuation Snapshot Methodology:

• Share price: $105 from FMP live quote (/quote).
• Market close date: August 14, 2026. Quote data timestamp (UTC): August 14, 2026, 08:00 PM UTC.
• Shares outstanding (market cap): 545.57 million (actual shares outstanding (2026-08-15 03:13:55; FMP /shares-float)).
• Market cap: Computed as share price × actual shares outstanding = $57.43 billion.
• Balance-sheet inputs (as of June 30, 2026): cash $6.40 billion; total debt $35.66 billion (short-term debt $8.10 billion + long-term debt $27.56 billion); book equity $5.02 billion (total shareholder equity on the balance sheet (June 30, 2026)).
• TTM diluted EPS (P/E denominator): $-3.65 = sum of four quarterly diluted EPS figures (September 30, 2025 $-0.22 + December 31, 2025 $-0.89 + March 31, 2026 $-1.40 + June 30, 2026 $-1.14). Each quarter uses company-reported diluted EPS or that quarter's net income ÷ its own diluted weighted-average shares—not the latest quarter's share count applied retrospectively to all four quarters.
• TTM revenue / net income: Sum of the four quarters ending June 30, 2026 ($7.59 billion revenue; $-1.93 billion GAAP net income).
• EBITDA denominator (TTM GAAP): $2.55 billion summed from four quarterly GAAP EBITDA values as reported on the income statement.
• Enterprise value: Market cap + total debt − cash and cash equivalents = $86.69 billion.
• P/E: FMP price-to-earnings ratio TTM (FMP /ratios-ttm).
• P/S: FMP price-to-sales ratio TTM (FMP /ratios-ttm).
• P/B: FMP price-to-book ratio TTM (FMP /ratios-ttm).
• EV/Revenue: FMP EV/revenue TTM (FMP /key-metrics-ttm).
• EV/EBITDA: FMP EV/EBITDA TTM (FMP /key-metrics-ttm) = 40.27.
• Technical price note: The Price & Technicals section uses the latest weekly adjusted close from historical end-of-day prices; the valuation snapshot uses the live FMP quote above.
• Precision: Multiples use the FMP inputs and quote dates shown above and are rounded for display; filings may differ in classification, restatements, or decimal precision.