Cognition
AI software-engineering agent platform — Devin, the autonomous AI engineer that writes, tests, debugs and deploys production code; sold to enterprise engineering organisations including Goldman Sachs, Mercedes-Benz, Dell, NASA, Santander, the US Army and the US Navy; ~$492M annualised revenue and $26B valuation in May 2026.
The Business
Cognition builds Devin, an autonomous AI software-engineering agent that writes, tests, debugs and deploys production code with minimal human intervention. The product line is anchored on the Devin runtime (long-horizon planning, code generation, sandboxed execution, test-and-deploy automation), a multi-provider foundation-model stack with Anthropic Claude as the principal upstream model at time of writing, and an enterprise deployment surface with named customer references including Goldman Sachs, Mercedes-Benz, Dell Technologies, NASA, Santander, the US Army and the US Navy. The company is privately held — founded 2023 by Scott Wu, Steven Hao and Walden Yan — and has raised approximately $1.6B+ of external capital through a May 2026 round at a $25B pre-money / $26B post-money valuation, co-led by Lux Capital, General Catalyst and 8VC with Ribbit Capital, Atreides Management and Peter Thiel’s Founders Fund participating.
Customers and Distribution
Cognition disclosed annualised revenue of approximately $492M as of May 2026, up from $37M in May 2025 — a 13x increase in 12 months, with enterprise usage growing 50% month-over-month for the prior six months per the company’s funding announcement and corroborating press cycle. Named customer disclosures include Goldman Sachs, Mercedes-Benz, Dell Technologies, NASA, Santander, the US Army and the US Navy. Distribution sits across two principal motions: direct enterprise sales driven by the founder team’s network and product-led-growth motion via Devin’s developer interface and Cognition-hosted runtime. The company stated a target of crossing $1B annualised revenue by end-2026; the most distinctive internal-validation signal disclosed is that 89% of Cognition’s own code commits are written by Devin itself — the data point that anchors the autonomous-agent disruption thesis in named-press coverage.
Model Strategy
Cognition is a Frontier-capability-first play under the IM Framework eight-trajectories taxonomy as it applies to coding AI: the strategic bet is that an autonomous AI engineering agent capable of closing long-horizon software-engineering tasks (multi-file changes, test-driven iteration, deployment workflows) beats human-in-the-loop copilots on production-engineering throughput inside enterprise procurement. The foundation-model stack is third-party multi-provider — Cognition discloses Anthropic Claude as the principal upstream model layer with additional model-routing across providers — which positions the company as an agent-platform play rather than a foundation-model lab. Above the foundation-model layer, the Devin runtime is the autonomous-agent surface; the long-horizon planning and sandboxed-execution system is the structural differentiator; enterprise outcome-based pricing on Devin tasks is the monetisation surface. The thesis does not depend on Cognition winning at the model layer — it depends on Devin sustaining a defensible lead on autonomous-agent capability against GitHub Copilot, Cursor, Claude Code and Codex CLI as the foundation-model providers race the frontier upstream.
At A Glance
The Numbers
Annualised revenue
Headcount (FTE)
Funding to date
Leadership Team
Cognition is founder-led with all three founders remaining in operating roles through the May 2026 funding round. Senior recruiting has drawn from the competitive-programming cohort, OpenAI, Scale AI, Tesla Autopilot and Anthropic alumni networks. CFO, CRO and CTO roles are not separately publicly named at time of writing; the company has not disclosed precise headcount in primary sources, though press coverage placed it in the low-hundreds range as of mid-2026.
IM Framework Scoring
IM’s structured assessment of Cognition’s competitive position. The summary below is the headline; expand “Show the full analyst-grade analysis” near the bottom for the per-dimension reasoning and evidence. Methodology →
Funding History
| Date | Round | Raised | Post-money | Lead investor(s) |
|---|---|---|---|---|
| May 2026 | Late-stage round | $1.0B+ | $26B | Lux Capital, General Catalyst, 8VC (co-led) |
| Sep 2025 | Late-stage round | $400M | $10.2B | Founders Fund (reported) |
| Mar 2024 | Series A | $175M | ~$2B | Founders Fund |
| 2023 | Seed | $21M | — | Founders Fund |
Cumulative external capital is approximately $1.6B+ disclosed through the May 2026 round at a $25B pre-money / $26B post-money valuation, co-led by Lux Capital, General Catalyst and 8VC with Ribbit Capital, Atreides Management and Peter Thiel’s Founders Fund participating. Round-by-round figures from Cognition’s own blog and named-press coverage at TechCrunch, eWeek, TheNextWeb and Winbuzzer. The May 2026 round was a ~2.5x mark-up from the September 2025 $10.2B post-money valuation reported eight months earlier. Earlier round figures (Series A, Seed) from Cognition’s own blog and the founding cycle press.
Competitive Landscape
| Competitor | Positioning | Distribution edge | Threat profile |
|---|---|---|---|
| GitHub Copilot ((Microsoft)) |
The incumbent AI coding assistant inside the GitHub and VS Code surface, offering chat, edit, agent and review modes across multiple model backends (OpenAI, Anthropic, Google); positioned as the default coding-AI for the 100M+ GitHub developer base. | Direct GitHub seat-based pricing bundled into GitHub Enterprise and Microsoft 365 enterprise agreements; the GitHub and VS Code installed base plus Microsoft enterprise procurement gravity are the principal moats. | High — the broadest distribution surface in coding AI via GitHub’s installed base of 100M+ developers; competing on copilot-to-agent transition with Copilot Workspace and the agent-mode roadmap. |
| Cursor ((Anysphere)) |
Standalone VS Code-fork AI code editor from Anysphere positioned as the developer-loved alternative to Copilot, with deep agentic-edit features, multi-model routing (Claude, GPT, Gemini) and a strong inline-completion experience. | Direct self-serve developer subscription plus an enterprise tier; viral developer-led growth and a fast release cadence drive the funnel, with seat-based monetisation as the channel. | High — the closest pure-play coding-agent rival with comparable ARR trajectory and developer mindshare; head-to-head on the autonomous-agent capability axis. |
| Claude Code ((Anthropic)) |
Anthropic’s first-party agentic coding CLI and terminal-native agent positioned as Claude’s reference implementation for autonomous software-engineering workflows on real repositories; direct competitor to Devin on the autonomous-engineer surface. | Direct distribution through claude.ai accounts, the Anthropic API and the Claude Pro / Max / Team / Enterprise plans; Anthropic’s model-provider distribution gravity and bundled inclusion in Claude subscriptions are the channel and moat. | High and asymmetric — foundation-model provider competing at the agent-workflow layer; the same indirect-competitor risk that defines the coding-AI cohort. |
| Codex CLI ((OpenAI)) |
OpenAI’s terminal-native coding agent and successor to the original Codex line, positioned as the first-party autonomous-coding surface for the GPT-5 / o-series model family inside the ChatGPT and OpenAI Platform ecosystem. | Direct distribution via the OpenAI API, ChatGPT Plus / Pro / Business / Enterprise subscriptions and the OpenAI developer platform; OpenAI’s enterprise customer base and model-provider gravity are the principal channels. | Medium-high — OpenAI’s first-party autonomous-coding interface; structurally symmetric foundation-model-provider play on the agent surface. |
| Windsurf ((Codeium)) |
AI-native code editor (formerly Codeium) positioned as an agentic IDE for enterprise developers with cascading multi-file edits, an autonomous-agent Cascade mode and on-prem / VPC deployment for regulated customers. | Direct self-serve developer plans and a direct enterprise sales channel; OpenAI’s 2025 acquisition cycle reshaped the distribution position — enterprise procurement and the on-prem deployment posture are the principal moats. | Medium-high — AI-native IDE with a strong enterprise channel; competes on the developer workflow surface adjacent to Devin’s autonomous-agent positioning. |
Potential Risks
Foundation-model supplier dependence
Devin runs on third-party foundation models (disclosed multi-provider mix with Anthropic Claude as the principal upstream model layer at time of writing). Capability shifts at the model-provider tier — including foundation-model-provider direct agent products like Claude Code and Codex CLI — propagate directly into Cognition’s agent capability and competitive position.
Symmetric competitor cadence on autonomous-agent capability
Cursor / Anysphere and the Codeium / Windsurf cohort are structurally symmetric pure-play competitors with comparable ARR trajectories and overlapping developer mindshare; GitHub Copilot inside Microsoft layers the broadest distribution surface on top. The symmetric-competitor cadence compresses outcome-based pricing and the gross-margin shape; whether Devin’s long-horizon agent capability sustains a defensible lead through 2026 is the most-watched competitive variable.
Valuation-to-ARR multiple at the $26B mark
The May 2026 $26B post-money valuation against ~$492M annualised revenue implies a ~53x ARR multiple — well above the public SaaS comp range of 8x to 15x. The bull case is that the ARR ramp continues at the disclosed 13x-in-12-months pace and the multiple resolves through growth; the bear case is that autonomous-agent unit economics or competitive substitution compress the trajectory before the multiple resolves.
Enterprise procurement and reliability bar at agent-of-record scale
Devin is positioned as the autonomous engineering agent of record for production code at Goldman Sachs, Mercedes-Benz, Dell, NASA, Santander, the US Army and the US Navy. The reliability bar for autonomous-agent code commits inside regulated and high-stakes enterprise environments is materially higher than for IDE-embedded copilots; any high-profile reliability incident at a marquee customer is a brand-defining event. The defensibility composite 5.75 prices in this load-bearing risk.
Founder-team concentration and executive-bench depth
Cognition is founder-led with no separately publicly named CFO, CRO or CTO at the $492M annualised revenue scale; senior recruiting has drawn from a focused competitive-programming and AI-engineering alumni pool. The bench-depth build-out over the next 12 months is a material watch-item as the company scales against Fortune 500 enterprise procurement and a fast-moving competitive set.
Recent IM Coverage
- Coding AI — sector landing May 2026.
- AI Tracker — methodology and universe May 2026.
Show recent press coverage of Cognition
- May 2026 — AI coding startup Cognition raises $1B at $25B pre-money valuation (TechCrunch)
- May 2026 — Funding, growth, and the next frontier of AI coding agents (Cognition Blog)
- May 2026 — AI Coding Startup Cognition Hits $26B Valuation After Massive $1B Raise (eWeek)
- May 2026 — Cognition Raises $1B as AI Coding Agent Devin Revenue Nears $492M (Winbuzzer)
- May 2026 — Cognition raises $1B at $26B valuation for AI coding agent (TheNextWeb)
- Mar 2024 — Introducing Devin, the first AI software engineer (Cognition Blog)
Show the source register for the figures on this page
IM operates a primary-source-where-possible discipline. The figures above come from:
- Revenue: Cognition disclosed annualised revenue of approximately $492M as of May 2026, up from $37M in May 2025 — a 13x growth in 12 months — per the company’s May 2026 funding blog post and corroborating TechCrunch coverage. Enterprise usage was disclosed at 50% month-over-month growth for the prior six months.
- Customer accounts: Cognition’s enterprise customer base includes Goldman Sachs, Mercedes-Benz, Dell Technologies, NASA, Santander, the US Army and the US Navy among others disclosed in the May 2026 funding announcement. Precise paid-account count is not disclosed in primary sources; we decline-to-publish a specific figure pending company disclosure.
- Headcount: Cognition does not publicly disclose precise headcount. Named-press coverage and the careers page indicate the company is in the low-hundreds-employee range as of mid-2026; we decline-to-publish a precise figure and reference the careers page as the canonical entry point.
- Funding to date: Cumulative external capital is approximately $1.6B+ through the May 2026 $1B round at a $25B pre-money / $26B post-money valuation, co-led by Lux Capital, General Catalyst and 8VC with Ribbit Capital, Atreides Management and Founders Fund participating.
Methodology & Disclaimer
For metric definitions, source-tier hierarchy, and decline-to-publish rules, see the tracker methodology. Confidence dots (• green / • amber / • red) follow the same convention as the AI Tracker.
Spotted a figure you believe is wrong? Send corrections to info@informationmatters.net.
Information Matters Framework scores are the considered opinion of the IM team — human and AI — applied to publicly-available evidence under a disclosed methodology. They are not statements of fact about the companies scored and they are not investment advice.
