AI coding agents are the most-commercialised AI agent category in 2026. Cursor reached $2 billion ARR with approximately 7 million monthly active users. Cognition Labs is reportedly in talks at a $25 billion valuation. Claude Code is tied with Cursor at 18 percent workplace adoption per the January 2026 JetBrains survey. GitHub Copilot crossed 4.7 million paid users at 42 percent market share. This page consolidates the market structure, pricing, capability benchmarks, and the consolidation dynamics.
Key Findings
- Anysphere (the company behind Cursor) reached approximately $2 billion ARR with approximately 7 million monthly active users by Q1 2026, making it the fastest-scaling software product on record by revenue trajectory.
- Cognition Labs, the company behind Devin, was reportedly in talks at a $25 billion valuation in Q1 2026. Cognition acquired Windsurf in late 2025 for an undisclosed amount; Windsurf had approximately $82 million ARR at the time.
- The January 2026 JetBrains workplace AI survey showed Claude Code and Cursor each at 18 percent adoption among professional developers, with GitHub Copilot at 42 percent overall and 24 percent among workplace users specifically.
- GitHub Copilot reached approximately 4.7 million paid users with approximately 90 percent of Fortune 100 deployments by May 2026.
- SWE-Bench Verified scores: Claude 4.7 Opus with Claude Code at approximately 82 percent, OpenAI o5-codex with agentic tooling at approximately 78 percent, Gemini 3.1 Pro with Code Mode at approximately 73 percent.
AI Coding Agent Market (May 2026)
| Product | Company | Paid Users / ARR | 2026 Valuation or Status |
|---|---|---|---|
| GitHub Copilot | Microsoft / GitHub | ~4.7M paid, ~$1.1B ARR | Consolidated |
| Cursor | Anysphere | ~7M MAU, ~$2B ARR | ~$20B private |
| Claude Code | Anthropic | ~1M+ active (estimated), monetised in Claude tiers | Part of Anthropic |
| Devin (Cognition) | Cognition Labs | ~150k seats (estimated) | $25B valuation talks |
| Windsurf (Cognition) | Cognition (acquired) | ~$82M ARR at acquisition | Now part of Cognition |
| Cline | Cline (open core) | ~600k installs | Open weights friendly |
| Aider | Aider OSS | ~800k installs | Open source |
| Continue.dev | Continue | ~400k installs | Open source |
| Tabnine | Tabnine | ~1M paid (est) | Independent |
| Codeium / Windsurf legacy | Now Cognition | Part of Cognition | Merged |
| Amazon Q Developer | AWS | ~600k seats (estimated) | Consolidated in AWS |
| Replit Agent | Replit | ~250k paid | ~$3.5B private |
| Sourcegraph Cody | Sourcegraph | ~150k paid (estimated) | Independent |
| JetBrains AI Assistant | JetBrains | Bundled in IDE | Independent |
JetBrains Workplace Adoption (January 2026)
| Tool | Workplace Adoption Share |
|---|---|
| GitHub Copilot | ~24% |
| Cursor | ~18% |
| Claude Code | ~18% |
| ChatGPT (general use for code) | ~16% |
| JetBrains AI Assistant (bundled) | ~11% |
| Devin / Cognition | ~3% |
| Other | ~10% |
Capability Benchmark Comparison (SWE-Bench Verified, May 2026)
| System | SWE-Bench Verified Score |
|---|---|
| Claude 4.7 Opus + Claude Code | ~82% |
| OpenAI o5-codex + agentic tooling | ~78% |
| Gemini 3.1 Pro Code Mode | ~73% |
| Cognition Devin | ~62% |
| Cursor agent mode (Claude backend) | ~80% |
| GitHub Copilot Workspace | ~64% |
| Aider (Claude backend) | ~71% |
| Best open-weight (Qwen 3.5 + agentic tooling) | ~58% |
Pricing Landscape
| Product | Per-User Pricing | Notes |
|---|---|---|
| GitHub Copilot | $10 individual / $19 Business / $39 Enterprise | Bundled with GitHub plans for some tiers |
| Cursor | $20 Pro / $40 Business per month | Usage-based for high-end models |
| Claude Code | Bundled in Claude Pro $20 / Max $100 / Team $30 per seat | Direct usage by Claude subscribers |
| Devin (Cognition) | $500 / month entry seat | Higher-tier custom pricing |
| Amazon Q Developer | $19 Pro | Bundled with AWS |
| Tabnine | $12 Pro | Privacy and on-prem options |
| Replit Agent | $25 Core / $40 Teams plus usage | Bundled with Replit hosting |
| Sourcegraph Cody | $9 Pro / $19 Enterprise | Codebase-aware |
Strategic Context
Three structural patterns shape the 2026 coding agent market. First, the model-backend competition is fierce: Cursor reportedly routes a majority of agent-mode traffic to Claude backends, making Anthropic the unseen winner in Cursor adoption while Anysphere captures the application-layer economics. Second, the differentiation has moved from completion-quality to agent-loop quality: the ability to plan multi-file changes, run tests, iterate on failures, and maintain context across long agentic sessions is the new battleground. Third, the consolidation pattern: Cognition\u2019s acquisition of Windsurf is one of multiple consolidation events, and the market is likely to settle to three to five major application-layer products plus IDE-integrated alternatives over the next 18 months.
Brand Visibility Implications
AI coding agent coverage is one of the highest-traffic technology categories in 2026. AI assistant queries about coding tool selection, Cursor vs Copilot comparisons, Claude Code adoption, and adjacent topics drive sustained traffic from developer audiences. Brands selling developer tools, code review, observability, AI evaluation, and adjacent products face strong AI-mediated discovery surface for this category.
Methodology
User and revenue figures sourced from primary company disclosures, the January 2026 JetBrains workplace developer survey, and reporting from The Information, Bloomberg, and TechCrunch. Benchmark scores from public SWE-Bench leaderboards. Updated quarterly.
How Presenc AI Helps
Presenc AI monitors brand visibility on AI coding agent queries across ChatGPT, Claude, Gemini, and Perplexity. For developer-tool brands, the platform identifies the prompts driving adoption decisions and the gaps where new content unlocks share of voice in a fast-moving category.