Anysphere / Cursor Startup Diligence Report
The central diligence question is whether Cursor can convert rapid AI-coding adoption into durable, secure, high-gross-margin enterprise software revenue while managing model-provider dependency, code-data privacy/IP risk, and intense competition from copilots, IDE vendors, and frontier model providers.
Anysphere / Cursor Startup Diligence Report
Anysphere/Cursor has unusually strong public growth and financing signals for a private AI developer-tools company, including a public $2.3B Series D at a $29.3B post-money valuation and company-reported $1B+ annualized revenue. The diligence posture should be positive on public traction but conservative on valuation quality until audited revenue, gross margin, retention, model-cost exposure, code/privacy controls, and customer concentration are validated.
Cursor depends on frontier model access, inference economics, model-provider policies, and internal model quality while serving sensitive code workflows.
Diligence request: Review model/provider contracts, cost curves, evals, fallback plans, security architecture, and data-flow maps.
Developer AI tooling is crowded, with IDE vendors, copilots, model providers, open-source tools, and enterprise platforms competing aggressively.
Diligence request: Request win/loss, churn reasons, pricing benchmarks, model quality benchmarks, and customer references.
The $29.3B valuation and $2.3B Series D are public, but audited revenue, gross margin, cash, debt, burn, runway, and financing terms are not publicly verifiable.
Diligence request: Request audited financials, ARR bridge, cash/debt, board plan, and Series D documents.
ARR and adoption claims are company-reported; customer concentration, NRR, churn, contract terms, discounts, and support burden require data-room validation.
Diligence request: Request customer cohorts, ARR waterfall, top-account contracts, retention, churn, usage, and discounting.
Cursor processes developer inputs and suggestions, making privacy, IP leakage, training-use commitments, subprocessors, audit logs, and security incidents central legal diligence topics.
Diligence request: Review DPA, SOC report, subprocessors, model-provider terms, incident history, IP assignments, and customer security reviews.
Public evidence identifies rapid expansion, a CEO/founder quote, and over 250 team members, but HRIS, attrition, succession, compensation, and management depth are not public.
Diligence request: Request org chart, HRIS export, attrition, hiring plan, comp bands, executive agreements, board composition, and succession plan.
Anysphere / Cursor has public financing, valuation, and selected operating signals, but full financial quality remains data-room dependent.
not publicly verifiable confidence: high
Public financial evidence is limited to selected company or secondary signals; audited revenue, margin, cash, burn, debt, and runway are not public.
| metric | public signal | verification status | diligence request |
|---|---|---|---|
| Annualized revenue | Company press release says Cursor exceeded $1B in annualized revenue. | partially_verified | ARR definition, revenue-recognition policy, monthly ARR bridge, retention, collections, and audit support. |
| Enterprise revenue growth | Release summary says enterprise revenue grew 100x in 2025 YTD. | partially_verified | Enterprise ARR starting base, bookings bridge, expansion, churn, and contract quality. |
| Gross margin/model costs | Not publicly disclosed. | not_publicly_verifiable | Model-provider costs, inference margin, credits, committed spend, and gross-margin bridge. |
| Cash, burn, runway, debt | Not publicly disclosed. | not_publicly_verifiable | Cash balance, debt, burn, runway, board plan, and financing uses. |
partially verified confidence: medium
Public valuation and financing anchors support unicorn status, but capitalization terms, preferences, ownership, and investor rights are private.
| date | round or event | amount | valuation | investors or sources | verification status |
|---|---|---|---|---|---|
| 2024-12-20 | CB Insights unicorn tracker | not disclosed in tracker row | $29.3B | Andreessen Horowitz, Thrive Capital, OpenAI Startup Fund listed by tracker | verified |
| 2025-11-13 | Series D | $2.3B | $29.3B post-money | Thrive, a16z, Accel, DST, Coatue, NVIDIA, Google | verified |
| stakeholder | public position | diligence caveat |
|---|---|---|
| Series D investors | Existing investors Thrive, a16z, Accel, DST; new investors Coatue, NVIDIA, Google. | Ownership percentages, preferences, side letters, and governance rights not public. |
| OpenAI Startup Fund / earlier investors | CB Insights tracker lists OpenAI Startup Fund among investors. | Investment timing, security class, and strategic rights require documents. |
| Employees/founders/options | No stock ledger or option plan public. | Request fully diluted cap table, option pool, vesting, and tender history. |
Public sources verify Anysphere / Cursor's product surface and monetization model at a high level; detailed usage, reliability, margin, and dependency economics are private.
partially verified confidence: medium
Public product pages and secondary sources verify the product surface, but usage depth, roadmap delivery, and service-level economics need data-room support.
| product or surface | audience | public features | verification status |
|---|---|---|---|
| Cursor IDE / code editor | Professional developers and engineering teams | AI-assisted code writing, reading, review, and codebase workflows. | verified |
| Agents / cloud agents | Individual, team, and enterprise developers | Agent requests, cloud agents with shared team context, automations, and security review agent on Teams. | verified |
| Enterprise controls | Large organizations | SAML/OIDC, SCIM, audit logs, model controls, privacy mode, account support. | verified |
| Frontier model access | Developers choosing models | OpenAI, Anthropic, Gemini, xAI, and Cursor model options. | verified |
partially verified confidence: medium
Pricing and delivery model are partly public; real monetization, cost-to-serve, infrastructure/vendor costs, and exceptions are not public.
| tier | public price | included or feature signal | monetization question |
|---|---|---|---|
| Hobby | Free | Limited agent requests and tab completions. | Free-to-paid conversion and model-cost exposure. |
| Individual | $20/month | Access to frontier models and individual usage. | Usage allowances, overage economics, and gross margin. |
| Teams | $40/user/month | Cloud agents, shared team context, team rules, SSO, privacy mode, analytics, centralized billing. | Seat expansion, churn, support load, and enterprise upsell. |
| Enterprise | Custom | Pooled usage, invoice/PO billing, SCIM, AI code tracking API, audit logs, granular model controls, priority support. | Discounting, committed spend, model-cost pass-through, and renewal terms. |
| dependency | role | public evidence | diligence gap |
|---|---|---|---|
| Frontier model providers | Model quality, feature availability, cost, latency, and compliance. | Cursor supports OpenAI, Anthropic, Gemini, xAI, and Cursor models. | Provider contracts, committed spend, fallback strategy, evals. |
| Code-data privacy controls | Enterprise trust and IP protection. | Privacy Mode/ZDR, no training use of inputs/suggestions except defined exceptions. | DPA, logs, subprocessor flow-downs, SOC controls, incident history. |
| Cloud agents and shared context | Team productivity and security surface. | Teams plan includes cloud agents and shared team context. | Architecture, isolation, access controls, retention, and audit logs. |
Customer/adoption evidence is meaningful but incomplete: public scale or logo claims do not disclose retention, concentration, cohort quality, or reference outcomes.
partially verified confidence: medium
Public adoption/customer signals are useful but do not disclose concentration, retention, churn, expansion, or customer health.
| segment | public signal | verification status | concentration question |
|---|---|---|---|
| Developers | Press release says Cursor serves millions of developers. | partially_verified | Active paid developers, activity cohorts, and conversion from free usage. |
| Teams | Press release says over 50,000 teams globally. | partially_verified | Paid team count, seat count, NRR, churn, and team-size distribution. |
| Large enterprises | Press release says majority of the Fortune 500. | partially_verified | Named logos, contracts, security exceptions, and renewal dates. |
partially verified confidence: medium
Public partner and supplier relationships identify dependencies but not contract terms, exclusivity, SLAs, or contingency plans.
| partner or supplier | nature | public evidence | diligence gap |
|---|---|---|---|
| Model providers | AI model ecosystem and inference dependency. | OpenAI, Anthropic, Gemini, xAI, Cursor model choices. | Contracts and cost exposure. |
| Strategic/investor ecosystem | Investors include NVIDIA and Google in Series D plus existing major venture investors. | Series D release. | Strategic rights, exclusivity, commercial commitments. |
| Enterprise security subprocessors | Data processing and compliance dependency. | Security page says subprocessors are published on trust portal and re-reviewed annually. | Subprocessor list, DPA, incident and audit evidence. |
The competitive landscape is active and cross-category; market-share and win/loss evidence are not public.
partially verified confidence: medium
Competitive alternatives are identifiable from the public market context, while share, win/loss, and displacement rates are not public.
| competitor | segment | funding or scale | product overlap | differentiator or risk |
|---|---|---|---|---|
| GitHub Copilot / Microsoft | AI coding assistant integrated into developer workflow | Large incumbent | Code suggestions, chat, agentic developer workflow | Distribution through GitHub, Visual Studio Code, and enterprise Microsoft contracts. |
| Codeium/Windsurf and other AI IDEs | AI-native coding tools | Private/startup and acquired/strategic alternatives | AI code editing, agents, enterprise controls | Fast feature parity and price pressure. |
| JetBrains, Replit, Sourcegraph, model-native tools | IDE, cloud dev, code search, and AI assistants | Mixed incumbents/startups | Developer productivity and code intelligence | Existing developer loyalty and enterprise procurement access. |
partially verified confidence: medium
Defensibility claims are plausible but require customer evidence, switching-cost data, product usage, and economic benchmarks.
| axis | target position | competitor position | evidence |
|---|---|---|---|
| Model quality and latency | Strong public positioning via frontier model access and Cursor models. | Model providers and copilots can integrate comparable models. | Company pages cite multiple frontier models. |
| Enterprise trust | SOC 2 availability, SSO/SCIM, audit logs, privacy mode. | Large incumbents often have broad compliance procurement paths. | Pricing/security pages. |
| Workflow lock-in | Codebase context, shared rules, automations, cloud agents. | Lock-in strength requires usage and retention evidence. | Public product claims, but no retention data. |
Public GTM channels are visible, but CAC, payback, pipeline conversion, discounting, and channel attribution require internal data.
partially verified confidence: medium
Observable channels support a preliminary GTM map, but CAC, payback, and sales productivity are not public.
| channel | region or segment | public evidence | diligence gap |
|---|---|---|---|
| Self-serve individual developer | Global developers | Free Hobby and $20/month Individual plan. | Free-to-paid conversion, CAC, support cost. |
| Teams self-serve/prosumer | Professional teams | $40/user/month Teams plan with shared context and centralized billing. | Seat expansion, team churn, margin by usage. |
| Enterprise sales | Large enterprises and Fortune 500 | Custom Enterprise plan, invoice/PO billing, account support, Fortune 500 demand claim. | Pipeline, win/loss, discounting, sales productivity. |
partially verified confidence: medium
Public marketing claims need reconciliation to booked revenue, paid usage, and retained cohorts.
| signal | public claim | diligence use |
|---|---|---|
| Series D and investor syndicate | $2.3B Series D with major strategic/financial investors. | Brand and credibility signal; validate terms and governance. |
| Developer/customer scale | Millions of developers, 50,000+ teams, majority Fortune 500. | Adoption signal; validate active, paid, retained, and concentrated users. |
| Enterprise revenue growth | Enterprise revenue grew 100x in 2025 YTD. | Momentum signal; validate starting base and booked/recognized revenue. |
R&D and technology signals show product ambition, but roadmap, architecture, incident history, and IP/control evidence require diligence.
partially verified confidence: medium
Roadmap and R&D direction are visible only at a high level; actual staffing, budget, milestones, and reliability history are private.
| initiative or personnel | status | expected or public timing | verification status |
|---|---|---|---|
| Technical research and frontier model training | Funded by Series D proceeds per release. | Post-Series D | verified |
| Composer agentic coding model | Referenced as recently launched in release. | Publicly referenced in 2025 release. | partially_verified |
| Security/privacy engineering | SOC 2, Privacy Mode, model blocklists, enterprise controls public. | Current controls; roadmap not public. | partially_verified |
partially verified confidence: medium
Technology and IP assets require deeper review of architecture, vendors, data rights, security, IP assignments, and open-source obligations.
| asset or process | public evidence | ip or control question |
|---|---|---|
| Model-routing and provider ecosystem | Model choices across OpenAI, Anthropic, Gemini, xAI, Cursor. | Internal model IP, provider contracts, fallback plans, eval data. |
| Privacy Mode/ZDR | Controls and contractual requirements to prevent model-provider storage/training of code data when enabled. | Enforcement, logs, exceptions, incident history. |
| AI code tracking API/audit logs | Enterprise plan includes AI code tracking API and audit logs. | Accuracy, retention, customer audit needs, legal discovery. |
Public leadership and people signals are enough for screening, not for management-depth, attrition, compensation, or governance reliance.
partially verified confidence: medium
Leadership evidence is public for selected executives/founders, but board composition, governance rights, succession, and compensation are not complete.
| name | role | tenure or background | source |
|---|---|---|---|
| Michael Truell | Co-founder and CEO | Quoted in Series D release. | Business Wire / Cursor release |
| Brian McCarthy | President of Global Revenue and Field Operations | Related Business Wire newsroom item visible in release page; not deeply diligenced. | Business Wire release page pointer |
| Board and broader executive team | Not fully public in accessed sources | Requires company confirmation. | Diligence request |
partially verified confidence: medium
People-scale signals are public in part; current HRIS, attrition, hiring plan, and employee-relations matters are private.
| function or region | public signal | source | diligence question |
|---|---|---|---|
| Company team | Over 250 engineers, researchers, designers, and operators. | Series D release | Current HRIS, function split, attrition, contractor mix. |
| Offices | San Francisco and new New York office mentioned. | Series D release | Employment compliance, hiring plan, lease obligations. |
| Executive succession/retention | Only limited public executive roster found in accessed sources. | Diligence observation | Management agreements, compensation, board oversight, key-person risk. |
Legal and compliance conclusions are limited screening observations; counsel schedules and specialist searches are required before reliance.
partially verified confidence: medium
Known legal/regulatory evidence is limited to public pages and secondary summaries; full liability assessment requires counsel records.
| matter | venue or agency | status | diligence request |
|---|---|---|---|
| Public litigation docket review | Not completed from specialist databases | not_publicly_verifiable | Counsel schedule and docket pulls. |
| Privacy/data-protection obligations | Global privacy laws via company privacy policy | partially_verified | DPA, DPIAs, regulator correspondence, incident log. |
| Security certifications | SOC 2 Type II report available on request | partially_verified | Obtain SOC report and bridge letters. |
partially verified confidence: medium
Public IP/privacy/security evidence identifies diligence areas but not complete ownership, licenses, incidents, insurance, or regulator correspondence.
| asset or area | jurisdiction or scope | public status | diligence request |
|---|---|---|---|
| Code and user input privacy | Service-wide | Inputs/Suggestions not used for training except defined exceptions. | Validate contracts, logs, product settings, and enterprise commitments. |
| Model/IP ownership | Cursor models and third-party models | Cursor model and third-party model options public; ownership/licensing not public. | Model training rights, IP assignments, open-source compliance, provider licenses. |
| Enterprise audit and AI code tracking | Enterprise customers | Enterprise plan includes API and audit logs. | Audit-log retention, legal discovery processes, customer security obligations. |
| ID | Claim | Status | Sources |
|---|---|---|---|
| EC-001 | CB Insights lists Anysphere as a San Francisco enterprise-tech unicorn valued at $29.3 billion. | verified medium | SRC-001 |
| EC-002 | Cursor announced a $2.3 billion Series D at a $29.3 billion post-money valuation in November 2025. | verified high | SRC-002SRC-007 |
| EC-003 | Cursor publicly positions itself as an AI development platform for writing, reviewing, and maintaining code over a codebase. | verified high | SRC-003 |
| EC-004 | Cursor publishes self-serve and enterprise pricing tiers with Individual at $20/month, Teams at $40/user/month, and Enterprise custom. | verified high | SRC-004 |
| EC-005 | Cursor claims more than $1 billion of annualized revenue, millions of developers, over 50,000 teams globally, and a majority of the Fortune 500. | partially verified medium | SRC-002 |
| EC-006 | Cursor has a public SOC 2 Type II attestation availability statement, subprocessor program, SSO/SCIM, audit logs, and Privacy Mode/ZDR controls. | verified high | SRC-004SRC-005 |
| EC-007 | Audited financial statements, cash, debt, burn, gross margin, retention, and cap-table terms were not public. | not publicly verifiable high | SRC-001SRC-002SRC-003SRC-004SRC-005SRC-006 |
| EC-008 | Cursor depends on third-party and first-party AI model ecosystems including OpenAI, Anthropic, Gemini, xAI, and Cursor models. | verified high | SRC-003SRC-005 |
| EC-009 | Customer concentration, retention, expansion, developer activity, and enterprise contract quality are not public. | not publicly verifiable high | SRC-002SRC-003 |
| EC-010 | The AI developer-tools market includes strong incumbents and fast-moving substitutes. | partially verified medium | SRC-003SRC-004 |
| EC-011 | Cursor has published enterprise security and administration features, but independent product defensibility and switching-cost data are not public. | partially verified medium | SRC-004SRC-005 |
| EC-012 | Cursor uses self-serve, team, and enterprise GTM motions. | verified high | SRC-002SRC-004 |
| EC-013 | Public marketing signals emphasize AI coding productivity, frontier models, enterprise demand, and rapid revenue growth. | partially verified medium | SRC-002SRC-003 |
| EC-014 | Cursor says Series D proceeds will fund technical research, product development, frontier model training, and team expansion. | verified high | SRC-002 |
| EC-015 | Cursor privacy and security claims reduce but do not eliminate code-data, IP, and model-provider risks. | partially verified medium | SRC-005SRC-006 |
| EC-016 | Cursor publicly reported a team of over 250 engineers, researchers, designers, and operators across San Francisco and New York. | partially verified medium | SRC-002 |
| EC-017 | Michael Truell is publicly quoted as Cursor co-founder and CEO. | verified high | SRC-002 |
| EC-018 | No comprehensive public litigation or regulatory-docket review was completed for Anysphere; public privacy/security terms are the main legal evidence reviewed. | not publicly verifiable high | SRC-005SRC-006 |
| ID | Publisher | Title | Accessed |
|---|---|---|---|
| SRC-001 | CB Insights | The Complete List of Unicorn Companies | 2026-05-17 |
| SRC-002 | Business Wire / Cursor | Cursor Secures $2.3 Billion Series D Financing at $29.3 Billion Valuation | 2026-05-17 |
| SRC-003 | Cursor | Cursor home page | 2026-05-17 |
| SRC-004 | Cursor | Cursor pricing | 2026-05-17 |
| SRC-005 | Cursor | Cursor security | 2026-05-17 |
| SRC-006 | Cursor | Cursor privacy policy | 2026-05-17 |
| SRC-007 | DuckDuckGo | DuckDuckGo results for Anysphere/Cursor funding | 2026-05-17 |
This report is a public-evidence diligence snapshot, not investment advice. Important financial, legal, technical, and contractual facts remain non-public and should be verified directly with management and primary documents before any investment decision.