# AI Coding and Developer Tools
> US-led sector of startups using large language models to write and review code, attracting record venture funding and the largest startup acquisition in history by mid-2026.

**Meta:** type: reference · date: 2026-07-03 · heads:  · 4 takes · 4 lenses · 2 regions

## What it is

AI coding and developer tools is the category of products and startups that apply large language models to writing, reviewing, and maintaining software. Three distinct form factors have emerged: IDE plug-ins that suggest and complete code as a developer types (GitHub Copilot, Tabnine, Codeium); AI-first integrated development environments built around a context-aware model rather than bolted on to an existing editor (Cursor, Windsurf); and command-line agents that take a task, open a branch, write commits, and file a pull request with minimal human steps (Anthropic's Claude Code, OpenAI's Codex). As of mid-2026, the sector has moved past autocomplete and into agentic work, where the model can own an issue from triage to merged pull request.

## History

GitHub launched Copilot in technical preview in June 2021, built on OpenAI's Codex model and trained on public code repositories. It became generally available in June 2022 and grew to 1 million paid subscribers within that year. The next structural shift came in 2023 when Anysphere released Cursor, an AI-first fork of VS Code built for whole-file and multi-file context rather than line-by-line suggestion. Codeium launched Windsurf in late 2024 with a similar IDE-centric design. Anthropic released Claude Code, a terminal-native agentic tool, in early 2025. By the time Stack Overflow published its 2025 Developer Survey, 84 percent of respondents reported using or planning to use AI tools in their workflows, up from 76 percent the year before.

## Current state

As of mid-2026, GitHub Copilot leads on raw scale: it crossed 20 million all-time users in July 2025 and reached 4.7 million paid subscribers and more than 77,000 adopting organizations by early 2026. Cursor's parent company Anysphere reached US$4bn in annualised revenue, with two-thirds of the US Fortune 500 as enterprise clients, before SpaceX agreed to acquire it for US$60bn in June 2026 (see [スペースXがCursor開発元Anysphereを600億ドルで買収、スタートアップ史上最大の買収案件](/ja/n/spacex-acquires-cursor-2026)), the largest acquisition of a venture-backed startup on record. GitHub introduced its own autonomous coding agent at Microsoft Build in May 2025, available to Copilot Enterprise and Pro+ users; the agent takes a GitHub issue, pushes commits to a draft pull request, and requires human approval before any CI/CD pipeline runs. Supabase, a PostgreSQL-as-a-service platform, raised US$500m at a US$10.5bn valuation in June 2026 after Claude Code and similar agents drove 600 percent year-on-year database growth (see [Supabase raises $500m at $10.5bn on 600% database growth driven by AI agents](/ja/n/supabase-series-f-2026)). Trust remains qualified: the 2025 Stack Overflow survey found 46 percent of developers actively distrust AI output accuracy versus 33 percent who trust it, and 66 percent cite near-miss suggestions as their biggest friction.

## Relationships

The sector is organized around a handful of platform providers: Microsoft, which controls GitHub (Copilot) and Visual Studio Code; Anthropic, which makes Claude Code; and SpaceX, which is acquiring Cursor. Amazon's CodeWhisperer, rebranded as Amazon Q Developer in 2024, competes for US enterprise accounts. On the venture side, Y Combinator, Andreessen Horowitz, and Sequoia have backed companies across the category; Thrive Capital, Accel, and GIC led the largest late-stage rounds. Growth in AI coding is tightly coupled to the database and compute infrastructure layer, which explains the trajectory of companies like Supabase and the rapid expansion of managed-cloud databases as agents became primary consumers of backend services.

## What to watch

Whether the agentic model, where AI autonomously handles a GitHub issue end-to-end, displaces the assistant model, where a human still drives the tool, will determine which companies command the largest share of enterprise budgets. The US$60bn SpaceX-Anysphere deal faces US antitrust review in Q3 2026. Legal questions around ownership and liability for AI-generated code remain unresolved across most jurisdictions. Developer sentiment, tracked annually by Stack Overflow and JetBrains, is a leading indicator: declining trust scores, if sustained, could compress premium-tier conversion rates even as overall adoption continues rising.

## Regional takes (batched by bias / lens)

### survey data
- **Stack Overflow 2025 Developer Survey** (global, en) — Annual survey of 65,000+ developers; 84 percent use or plan to use AI coding tools, up from 76 percent in 2024; only 33 percent trust AI output accuracy versus 46 percent who actively distrust it.
  Source: https://survey.stackoverflow.co/2025/ai

### official record
- **GitHub Newsroom** (United States, en) — GitHub's May 2025 announcement of an autonomous coding agent for Copilot Enterprise and Pro+ users; notes 150 million GitHub developers and 77,000 organizations on Copilot as of that date.
  Source: https://github.com/newsroom/press-releases/coding-agent-for-github-copilot

### markets / growth
- **TechCrunch** (United States, en) — Reports GitHub Copilot crossing 20 million all-time users as of July 2025, citing Microsoft CEO Satya Nadella on the company's earnings call; notes 5 million new users joined in the preceding three months.
  Source: https://techcrunch.com/2025/07/30/github-copilot-crosses-20-million-all-time-users/

### analysis
- **Stack Overflow Blog** (global, en) — Stack Overflow's editorial analysis of its 2025 survey, documenting the gap between rising adoption and falling trust; 66 percent of developers cite near-miss AI suggestions as their biggest friction point.
  Source: https://stackoverflow.blog/2025/12/29/developers-remain-willing-but-reluctant-to-use-ai-the-2025-developer-survey-results-are-here/

## Across the graph
- Related: [[spacex-acquires-cursor-2026]], [[supabase-series-f-2026]]
- Entities: AI Coding Startups, Startup Ma, AI Agent Startups, AI Infra Startups, Foundation Model Startups

---
Canonical: https://rbtfl.xyz/ja/n/ai-coding-startups-dossier