Skip to content
Deepak Gupta markGTM Map

AI-native · founded 2022 · acquired by SpaceX (2026)

How Cursor went to market

Cursor shipped an AI code editor to the public within months, then grew for two years almost entirely by word of mouth among developers while the founders left sales and marketing alone. In 2025 it moved to consumption pricing and built an enterprise sales team on top of that base, and in 2026 SpaceX acquired it.

Buyer:
Developers, Enterprise
Contract size:
Mixed
Checked:
October 17, 2026

Draft: this record is sourced but has not been through final editorial review. It is not indexed yet.

01The phases

PrimarySupportingTried, then dropped

  1. 2022 to 2023

    Build for ourselves, launch in public

    The founders built the editor for their own use, launched it to the public about three months after the first line of code, and measured paid power users rather than signups.

    • Product-led

      A self-serve editor launched within months and improved in public. The team tracked paid power users, people using the AI four or five days a week, as the number to grow.[15],[16]

    • Developer-led

      The founders were the first users, living in the editor full time after about five weeks. Developers are both the entry user and the person who decides to adopt it.[14],[15]

    • Inbound content

      A co-founder built a following by posting about AI and used it, with a demo and a waitlist, to bring in the first users. The team then stepped away from it and, in the CEO's words, lived like monks through 2023.[17]

  2. 2024 to 2025

    Word of mouth compounding

    Product improvements showed up directly in usage and users told other developers. Sales and marketing were deliberately left alone, and growth engineering sprints did not move the numbers the way product work did.

    • Product-led

      The CEO says the product spread from word of mouth: in a market where end users pick their tools, making the best product meant people heard about it. Two-month growth engineering sprints washed away next to product work.[17],[15]

    • Developer-led

      By January 2025 Cursor described itself as used by millions of programmers as their editor of choice.[2]

    • Education

      From May 2025 verified students in eligible countries and schools could claim a year of Pro free. The current student page no longer states that offer.[13]

  3. 2025 to now

    Enterprise push and consumption pricing

    With large companies already using it, Cursor added a sales path, enterprise security credentials, and a field sales team, and moved its plans from request limits to compute-based usage.

    • Usage pricing

      In June 2025 Pro moved from request limits to compute limits with $20 of included usage, and a $200 Ultra tier was added. Cursor apologised that the change came as a surprise and offered refunds. The CEO later described the direction as a consumption model.[5],[6],[18]

    • Sales-assist

      Sales was layered onto a product-led base. The CEO describes starting incredibly lean and product-led, then building out for larger customers. The enterprise page routes to Contact sales and lists SOC 2 Type II, SSO, and SCIM.[15],[8],[9]

    • Land and expand

      The CEO calls teams the atomic unit Cursor serves. A customer quote on the enterprise page describes adoption growing from 150 to over 500 engineers.[18],[8]

    • Community

      A funded ambassador program runs meetups, workshops, and hackathons. Cursor reports more than 800 events and 300 ambassadors in over 80 countries.[11],[10]

    • Sales-led

      By October 2026 the careers page lists a full field organisation: regional sales VPs, strategic and federal account executives, and a deal desk. Job listings show structure, not revenue mix.[12]

    • Outbound

      Sales development roles are open in London, New York, San Francisco, and Australia and New Zealand. No source describes their results yet.[12]

02The numbers

MetricValueAs ofBasis
CustomersOver 40,000[1]August 22, 2024Company-stated
Recurring revenueOver $100M[2]January 16, 2025Company-stated
Annual recurring revenueOver $500M[3]June 6, 2025Company-stated
HeadcountGoing on 60 people[15]May 1, 2025Company-stated
Annualized revenueOver $1B[4]November 13, 2025Company-stated
Annualized revenueOver $2B[19]March 2, 2026Reported
Share of revenue from large corporate buyersAbout 60%[19]March 2, 2026Reported
Acquisition price (SpaceX, all stock)$60B[21]June 16, 2026Reported
Fortune 500 companies using Cursor64%[8]October 17, 2026Company-stated

“Reported” figures come from trade press, not the company, and are shown with their date.

03Myth checks

  • “Cursor did zero marketing.”

    Partly true

    The founders did leave sales and marketing alone through the word of mouth years. But the launch used a co-founder's social following, a demo, and a waitlist, and by 2026 Cursor funds an ambassador and events program and is hiring paid media and field marketing roles.[17],[15],[10],[12]

  • “Cursor grew with no sales team.”

    Partly true

    True for the first stretch: the CEO says it was the co-founders and a few people until growth was dialed in. Not true from 2025, when Cursor built enterprise engineering, a Contact sales path, and a field sales organisation.[16],[15],[18],[12]

04What other companies can take from it

  • When the user is also the buyer, let product quality carry growth before you build a growth team.

    Works when: The end user picks the tool personally, can install it without approval, feels quality differences every day, and the price is low enough to pay or expense alone.

    Backfires when: The buyer is not the user (security, compliance, HR systems), procurement gates installation, or quality differences are invisible in a short trial.

    Based on: Word of mouth compounding, Product-led; Build for ourselves, launch in public, Developer-led

  • Track one metric tied to a paid habit, not signups.

    Works when: The product is used daily and every use has a real cost, such as inference or compute.

    Backfires when: Use is naturally occasional, like quarterly reporting or incident response, so a daily-use bar would flag healthy customers as churn risks.

    Based on: Build for ourselves, launch in public, Product-led

  • Add sales after enterprises arrive through the product, and sell to teams.

    Works when: Many signups already come from company email domains and large accounts have dozens of individual users before any contract.

    Backfires when: There is no bottom-up base to expand from, so field sales becomes cold outbound with none of the product-led advantage.

    Based on: Enterprise push and consumption pricing, Sales-assist; Enterprise push and consumption pricing, Land and expand

  • A move to consumption pricing needs over-communication and a refund path.

    Works when: Cost per user varies widely, so flat pricing loses money on heavy users.

    Backfires when: Users cannot see or predict their spend, and the switch reads as a hidden price rise to the community that drove word of mouth.

    Based on: Enterprise push and consumption pricing, Usage pricing

05 · GTM Now

What changes in the AI era, and what to do now

Analysis by Deepak Gupta. This is opinion built on the sourced record above, not part of it.[16],[14],[20],[6],[10]

Still works

  • Product quality as the growth engine in a market where switching is cheap. Developers still hear about tools from peers, which AI answers weaken far less than they weaken search-driven content.
  • A paid-habit metric. When every use costs compute, a free user who never pays is a cost, not a lead.

Breaks

  • Cheap switching cuts both ways. The same low switching cost that let Cursor take share from GitHub Copilot exposes it to Claude Code and Codex.
  • Flat seat pricing, when compute per user varies by orders of magnitude.

What replaces it

  • Included-usage plans with overage, plus team and enterprise contracts, where reported margins are positive.
  • A funded ambassador and meetup network in place of the founder posting that started the launch.

Apply it now

If you are

A seed or Series A developer tool, used daily by individual developers, priced under $1K a seat

  1. Ship to the public within weeks and use the product yourselves every day.
  2. Define one paid-habit metric, such as paid users active four or more days a week, and review it weekly before any growth spend.
  3. Run at most one time-boxed growth experiment a quarter, and drop it if it does not beat the lift from your latest product release.
  4. Use founder posting and a waitlist for the launch only, then put that time back into the product.

You will know it works when: Week-over-week growth in paid power users, and the share of new signups who say a colleague or friend sent them.

Based on Cursor's Build for ourselves, launch in public, Word of mouth compounding

If you are

A product-led AI tool where heavy users cost far more than light ones, with company domains showing up in signups

  1. Move heavy usage to included allowance plus overage, and publish worked examples of monthly cost before you switch.
  2. Announce with a dated changelog, keep the old plan available for a fixed period, and offer a refund window.
  3. Have team billing, SSO, SCIM, and a SOC 2 report in place before you hire account executives.
  4. Hire expansion account executives to work accounts that already have many individual users.

You will know it works when: Gross margin by segment (individual, team, enterprise) and seat growth inside accounts that started as individual use.

Based on Cursor's Enterprise push and consumption pricing, Enterprise push and consumption pricing, Enterprise push and consumption pricing

Motions in this record

Product-ledDeveloper-ledInbound contentEducationUsage pricingSales-assistLand and expandCommunitySales-ledOutbound

Compare with every other company on the GTM matrix.

Sources

  1. [1]CompanySeries A (August 22, 2024)
  2. [2]CompanySeries B (January 16, 2025)
  3. [3]CompanySeries C (June 6, 2025)
  4. [4]CompanySeries D (November 13, 2025)
  5. [5]CompanyNew tier: Ultra, and Pro moves to compute limits (June 16, 2025)
  6. [6]CompanyClarifying our pricing (July 4, 2025)
  7. [7]CompanyCursor is now a part of SpaceX (August 14, 2026)
  8. [8]CompanyCursor for Enterprise (October 17, 2026)
  9. [9]CompanyCursor Security (August 25, 2026)
  10. [10]CompanyCursor Community (October 17, 2026)
  11. [11]CompanyCursor Ambassadors (October 17, 2026)
  12. [12]CompanyCursor Careers (snapshot) (October 17, 2026)
  13. [13]CompanyStudent Discount: Details, Updates, Q&As (staff post) (May 8, 2025)
  14. [14]Founder interviewLex Fridman Podcast #447: Cursor Team (transcript) (Michael Truell, Sualeh Asif, Aman Sanger, October 6, 2024, at 00:11:33, 00:13:24, 00:13:59)
  15. [15]Founder interviewThe rise of Cursor (Lenny's Podcast) (Michael Truell, May 1, 2025, at 24:40 to 29:15, 58:06 to 59:07)
  16. [16]Founder interviewCursor CEO: Going Beyond Code (Y Combinator, How to Build the Future) (Michael Truell, June 11, 2025, at 26:28 to 31:02, 35:05)
  17. [17]Founder interviewMichael Truell: Building Cursor At 23 (Y Combinator) (Michael Truell, September 3, 2025, at 20:14 to 24:19)
  18. [18]Founder interviewWhy Cursor's CEO believes OpenAI, Anthropic competition won't crush his startup (TechCrunch, Fortune Brainstorm AI) (Michael Truell, December 9, 2025, at direct quotes in the article body)
  19. [19]Trade pressCursor has reportedly surpassed $2B in annualized revenue (TechCrunch, March 2, 2026)
  20. [20]Trade pressCursor in talks to raise $2B+ at $50B valuation as enterprise growth surges (TechCrunch, April 17, 2026)
  21. [21]Trade pressSpaceX to acquire the AI coding startup Cursor for $60 billion (CNBC, June 16, 2026)

Every phase, motion, and myth verdict cites a filing, the company's own pages, or a founder speaking on the record. See the methodology.