• Skip to main content
  • Skip to secondary menu
  • Skip to footer
information matters logo

Information Matters - Agentic AI News and Market Forecasts

The Agentic AI Revolution: what it means for business and the rules of competition

  • Home
  • About
    • The Team
    • About Us
    • Our Methodology
  • Contact
  • Subscribe
  • Downloads
  • Agentic AI Company Tracker
  • Agentic AI Sector Analysis

Cognition

COMPANY PAGE

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.

Founded 2023
Private — Series funding
Coding AI
cognition.ai

Last Updated: 28 May 2026
Fact-checked: 2 June 2026
Coverage: Tracker · Category Report (Coding AI, forthcoming)
← Back to AI Tracker

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

Annualised revenue
$250M ●
2026-04-30 as-of

2024-09-302026-04-30

Enterprise customer accounts
500 ●
2026-01-31 as-of

Headcount
400 ●
2026-04-30 as-of

2024-12-312026-04-30

Funding to date
$696M
2025-12-31 as-of

2024-12-312025-12-31

The Numbers

Annualised revenue

$250M $1M 2024-09-30 — 1 2024-12-31 — 4 2025-03-31 — 20 2025-06-30 — 73 2025-09-30 — 155 2025-12-31 — 200 2026-04-30 — 250 2024-09-30 2026-04-30

Headcount (FTE)

400 30 2024-12-31 — 30 2025-06-30 — 60 2025-09-30 — 300 2025-12-31 — 350 2026-04-30 — 400 2024-12-31 2026-04-30

Funding to date

$696M $196M 2024-12-31 — 196 2025-06-30 — 296 2025-09-08 — 696 2025-09-30 — 696 2025-12-31 — 696 2024-12-31 2025-12-31

Leadership Team

Co-founder & CEO
Scott Wu
Co-founded Cognition in 2023. Former competitive-programming gold medalist (International Olympiad in Informatics) and engineer at Lunchclub. Public-facing across the funding cycle including the March 2024 launch, the September 2025 $400M round and the May 2026 $1B raise; framed the autonomous-agent thesis on the Lex Fridman podcast and across press cycles.

Co-founder & President
Steven Hao
Co-founded Cognition in 2023. Former competitive programmer and engineer at Scale AI; technical lead on the Devin agent architecture and the long-horizon planning system that anchors the autonomous-engineering thesis.

Co-founder
Walden Yan
Co-founded Cognition in 2023. Competitive programmer and engineering operator across the Devin product line, agent runtime and the enterprise deployment surface.

Engineering Lead
Russell Kaplan
Senior engineering leader at Cognition; previously at Tesla on Autopilot and at Scale AI. Joined the Cognition founding team in 2024; technical lead on the agent runtime and the production-deployment architecture for enterprise customers.

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 →

Competitive Position
Disruptive Challenger
Coding AI sector

The Information Matters Compass

5 7.5 10 5 7.5 10 Defensibility → Disruption Potential →Disruptive Challengers Dominant InnovatorsEmerging Players Established Incumbents Cognition © Information Matters

Strategic Bet
Frontier-capability wins — an autonomous AI engineering agent (Devin) that closes long-horizon coding tasks beats human-in-the-loop copilots on enterprise software-engineering throughput
Plus: Plus: rewire — enterprise engineering organisations restructure around AI agents inside three to five years, with Cognition as the agent-of-record for production code

Watch: The pace of the $1B annualised-revenue target Cognition stated for end-2026; the Cursor / Anysphere and GitHub Copilot competitive cadence on autonomous-agent capability; OpenAI Codex CLI and Anthropic Claude Code on the foundation-model-provider-direct substitution; the Windsurf / Codeium acquisition by Cognition’s competitors; and the gross-margin shape as Devin throughput scales against the underlying inference cost stack.

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.

Footer

  • LinkedIn
  • YouTube

Copyright © 2026 · Information Matters

Manage Consent
To provide the best experiences, we use technologies like cookies to store and/or access device information. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
Functional Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes. The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
  • Manage options
  • Manage services
  • Manage {vendor_count} vendors
  • Read more about these purposes
View preferences
  • {title}
  • {title}
  • {title}