Insights
Article 4 of 14 - State of the Market
The analysts are undercounting it. So what is the forecast measuring?
Read Time: 4 min
The AI coding vendors are now bigger than the forecasts for their own market. The fastest-growing of them, OpenAI's Codex, is not even in the count. The forecasts are not wrong. They are late. Treat every published market size as a floor, and buy as if the vendors' growth rate is the real one.
Add up the run-rates of eight AI coding vendors and you get about $9.5 billion. That covers Cursor, Claude Code, Cognition, Replit, Lovable, GitHub Copilot, Base44, and Emergent, on figures published between February and September 2026. Mordor Intelligence forecasts the whole AI code tools market at $9.4 billion for 2026. The vendors have already passed it, before the year is out. And that total leaves out OpenAI, which does not disclose what Codex earns.
What the vendors are reporting
The largest vendors are growing three to five times a year. No published forecast assumes anything close.
Cursor passed $2 billion in March and was estimated near $4 billion by June. SpaceX bought it for $60 billion in August, so it no longer reports on its own. Anthropic's run-rate passed $65 billion by the end of July. The last figure it gave for Claude Code alone was $2.5 billion. Lovable reached $600 million in September, up from $500 million in June. Replit reached $525 million in April. Cognition was at $492 million in May and on course for about $1 billion in September. Emergent and Base44 are at about $100 million each.
GitHub does not disclose Copilot revenue. At 4.7 million paid seats and $11–15 a seat, it is worth about $0.65 billion. These are run-rates, not audited revenue. Where a figure is an estimate, we say so.
OpenAI is the biggest gap in that sum. It does not publish Codex revenue. It does publish usage. Codex had under a million weekly users in February, 3 million in April, and more than 5 million in June. By October, Codex and ChatGPT Work together had 40 million active users, though OpenAI did not say over what period.
OpenAI's total run-rate passed $40 billion in August, with Codex named as a driver. It told investors September was about $50 billion. That came days after a report of nearly $70 billion, which shows how loosely the term gets used. Whatever Codex earns sits inside those numbers. No category forecast counts it. Neither does ours.
Buyers and investors see the same picture. Coding took $4 billion of enterprise generative AI spend in 2025, more than half of all departmental AI spend. SpaceX paid $60 billion for Cursor. Lovable was valued at $13.3 billion. Nobody pays that for a $10 billion market growing at 25% a year.
Figure 1: Leading AI coding vendors are reaching multi-billion-dollar revenue run-rates at unprecedented speed.
Why the forecasts trail the market
The forecasts are not wrong. They are late. Mordor Intelligence puts the 2026 market at $9.4 billion, Fortune Business Insights at $10.1 billion, Grand View Research at about $10.4 billion, and MarketsandMarkets at about $12.1 billion. Their growth rates run from 22.5% to 48% a year. Only MarketsandMarkets' 48% fits what the vendors are reporting.
Four things explain the gap.
Category lines. Claude Code sits inside a model lab's API revenue. Codex sits inside ChatGPT Business seats. Copilot sits inside Microsoft's cloud numbers, and Cursor now sits inside SpaceX. Forrester's definition leaves out app generators like Lovable altogether. A 'code tools' forecast misses several of the fastest sellers by design.
Base years. A forecast that compounds a 2024 base at 25% a year cannot describe vendors growing three to five times in twelve months.
Pricing. GitHub moved Copilot to usage-based billing in April. Codex is billed in credits per million tokens. A model built on seats cannot see revenue that tracks tokens.
Substitution. The money is coming from budgets the forecasts do not watch. Some comes from labour and services, where Bain puts today's gains from assistants at only 10–15%. Some comes from the software budget, where McKinsey finds 32% of companies have decided to build software they would once have bought. February's sell-off in software stocks priced in the same thing.
Figure 2: Observed vendor revenue is challenging analyst forecasts for the AI coding tools market.
What a buyer should plan against
Plan against the vendors' growth, not the analysts' forecast. Three things follow.
Cost. Usage-based pricing moves the model-cost risk to you. Last year's predictable seat licence is this year's metered bill, and usage is compounding. Budget for it. Ask for caps. Prefer suppliers who price the outcome, not the tokens.
Churn. At these valuations, consolidation is a schedule, not a risk. Cognition bought Windsurf in July 2025. AWS is retiring Amazon Q Developer in favour of Kiro. Ona joined OpenAI in June. SpaceX bought Cursor in August. GitHub now runs Claude and Codex agents side by side in the same repository. The tool is not the asset. Make sure you own the code, the tests, the infrastructure definitions, and the evidence, in formats that outlive the vendor.
Governance. Gartner's first quadrant for enterprise AI coding agents made verification and auditability mandatory. Gartner also expects more than 40% of agentic AI projects to be cancelled by the end of 2027. Buy the evidence layer, not the model of the month.
None of this is a reason to wait. Work is moving from people to agents faster than anyone has written down. Buy on three assumptions: usage will grow faster than planned, vendors will change hands, and what you keep is what you own.
What we hold ourselves to
We plan our own business on the vendors' numbers, not the analysts'. We assume the models we use today will be replaced within eighteen months, so a change of model changes nothing about the evidence. A second, independent model vendor reviews design and code. That guards against error, and against lock-in. Every client owns everything we produce for them: source, tests, infrastructure definitions, pipeline, and executed evidence, in standard formats and in their own accounts. That is why Forge is model-agnostic. The model is the part that changes. The evidence is the part the client keeps.
The forecasts will catch up. MarketsandMarkets has started. Until the rest do, assume the market is bigger and faster than anyone has published, and buy in a way that works whichever number turns out to be right.
So, a question for your own organisation. If usage of your coding tools triples next year, who pays? And if the vendor is bought the year after, what do you still own?
Sources:
1. Mordor Intelligence, AI code tools market size, share and 2031 growth trends, July 2026. https://www.mordorintelligence.com/industry-reports/artificial-intelligence-code-tools-market
2. TechCrunch, Cursor has reportedly surpassed $2B in annualised revenue, 2 March 2026. https://techcrunch.com/2026/03/02/cursor-has-reportedly-surpassed-2b-in-annualized-revenue/
3. Sacra, company revenue estimates – Cursor, Cognition, Replit, Base44, Anthropic, 2026 (analyst estimates, unaudited). https://sacra.com/c/cognition/
4. OpenAI, The next phase of enterprise AI (Codex at 3 million weekly active users; enterprise more than 40% of revenue), 8 April 2026. https://openai.com/index/next-phase-of-enterprise-ai/
5. OpenAI, Codex is becoming a productivity tool for everyone (more than 5 million weekly active users), 2 June 2026. https://openai.com/index/codex-for-knowledge-work/
6. Unite.AI (reporting Bloomberg), OpenAI says Codex and ChatGPT Work hit 10 million users, 21 July 2026. https://www.unite.ai/openai-says-codex-and-chatgpt-work-hit-10-million-users/
7. Yahoo Finance (reporting Bloomberg), OpenAI revenue run rate tops $40 billion, 13 August 2026. https://finance.yahoo.com/technology/ai/articles/openai-revenue-run-rate-tops-224009196.html
8. OpenAI Help Center, Codex rate card (credit-based pricing for Business, Enterprise, and Edu), September 2026. https://help.openai.com/en/articles/20001106-codex-rate-card
9. OpenAI, Codex named a Leader in the 2026 Gartner Magic Quadrant for Enterprise AI Coding Agents, May 2026. https://openai.com/index/gartner-2026-agentic-coding-leader/
10. VentureBeat, Anthropic says it hit a $30 billion revenue run-rate, May 2026. https://venturebeat.com/technology/anthropic-says-it-hit-a-30-billion-revenue-run-rate-after-crazy-80x-growth
11. TechCrunch, Lovable says it has hit $500M in annualised revenue, 9 June 2026. https://techcrunch.com/2026/06/09/lovable-says-it-has-hit-500m-in-annualized-revenue-with-1-million-new-projects-a-week/
12. TechCrunch, Emergent hits $100M ARR eight months after launch, 17 February 2026. https://techcrunch.com/2026/02/17/emergent-hits-100m-arr-eight-months-after-launch-rolls-out-mobile-app/
13. Wix.com, Q4 2025 earnings call (Base44 revenue), 4 March 2026. https://finance.yahoo.com/news/wix-com-q4-earnings-call-165003576.html
14. Microsoft, FY26 Q2 earnings release, January 2026. https://www.microsoft.com/en-us/investor/events/fy-2026/earnings-fy-2026-q2
15. Menlo Ventures, 2025: The state of generative AI in the enterprise, December 2025. https://menlovc.com/perspective/2025-the-state-of-generative-ai-in-the-enterprise/
16. TechCrunch, Lovable confirms new $13.3B valuation, raises another $400M, 12 August 2026. https://techcrunch.com/2026/08/12/lovable-confirms-new-13-3b-valuation-raises-another-400m/
17. Fortune Business Insights, AI code tools market size, share, trends, 2034, August 2026. https://www.fortunebusinessinsights.com/ai-code-tools-market-111725
18. Grand View Research, AI code assistants market size, share, industry report 2033, 2026. https://www.grandviewresearch.com/industry-analysis/ai-code-assistants-market-report
19. MarketsandMarkets, AI code assistants market report 2025–2032, September 2026. https://www.marketsandmarkets.com/Market-Reports/ai-code-assistants-market-53503659.html
20. Forrester, Agentic software development: defining the next phase of AI-driven engineering tools, March 2026. https://www.forrester.com/blogs/agentic-software-development-defining-the-next-phase-of-ai-driven-engineering-tools/
21. GitHub, GitHub Copilot is moving to usage-based billing, April 2026. https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/
22. Bain & Company, From pilots to payoff: generative AI in software development (Technology Report 2025), September 2025. https://www.bain.com/insights/from-pilots-to-payoff-generative-ai-in-software-development-technology-report-2025/
23. Yahoo Finance, reporting McKinsey: 32% of companies decided against buying software they could build, September 2026. https://finance.yahoo.com/technology/ai/articles/build-vs-buy-shift-32-113806700.html
24. Fortune, Anthropic's Claude Opus 4.6 upgrade triggers software stock sell-off, 6 February 2026. https://fortune.com/2026/02/06/anthropic-claude-opus-4-6-stock-selloff-new-upgrade/
25. AWS, Amazon Q Developer end-of-support announcement, April 2026. https://aws.amazon.com/blogs/devops/amazon-q-developer-end-of-support-announcement/
26. Ona (formerly Gitpod), Ona is joining OpenAI, June 2026. https://ona.com/blog
27. GitHub, Pick your agent: use Claude and Codex on Agent HQ, February 2026. https://github.blog/news-insights/company-news/pick-your-agent-use-claude-and-codex-on-agent-hq/
28. Virtualization Review, reporting Gartner's 2026 Magic Quadrant for Enterprise AI Coding Agents, 5 June 2026. https://virtualizationreview.com/articles/2026/06/05/ai-firms-push-cloud-giants-from-leaders-quadrant-in-gartner-ai-coding-report.aspx
29. Gartner, over 40% of agentic AI projects will be cancelled by end of 2027, 25 June 2025. https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027
30. Bloomberg News (via Advisor Perspectives), SpaceX completes its $60 billion Cursor acquisition, 14 August 2026. https://www.advisorperspectives.com/articles/2026/08/14/spacex-completes-60-billion-cursor-acquisition
31. CEO Insights (reporting Reuters), Anthropic revenue run rate crosses $65 billion in July 2026, 18 August 2026. https://www.ceoinsightsindia.com/news/anthropic-revenue-run-rate-crosses-65-billion-in-july-2026-nwid-24708.html
32. El Ecosistema Startup (reporting TechCrunch), Lovable crosses US$600M ARR, 24 September 2026. https://ecosistemastartup.com/lovable-cruza-us600m-arr-y-consolida-el-vibe-coding/
33. RuntimeWire (reporting Bloomberg), Cognition's revenue run rate reaches $1B, four months after $492M, 25 September 2026. https://runtimewire.com/article/cognition-billion-annualized-revenue-scott-wu
34. Bloomberg Government (reporting Axios), OpenAI's annualized revenue nears $70 billion, 29 September 2026. https://news.bgov.com/artificial-intelligence/openais-annualized-revenue-nears-70-billion-axios-says
35. Zawya (reporting Reuters and the Financial Times), OpenAI's September annualized revenue nears $50bln, less than previously indicated, 9 October 2026. https://www.zawya.com/en/special-coverage/openais-september-annualized-revenue-nears-50bln-less-than-previously-indicated-source-says-1586243
36. ToolNavs, Codex and ChatGPT Work hit 40 million active users, 8 October 2026. https://toolnavs.com/article/2404-codex-and-chatgpt-work-hit-40-million-active-users-a-big-number-with-a-catch
