Cognition

Sacra estimates that Cognition hit $155M in annual recurring revenue (ARR) in July 2025 following its acquisition of Windsurf, which was contributing $82M ARR across 350+ enterprise accounts at the time of the deal.

Prior to the acquisition, Cognition’s Devin agent had scaled to $73M ARR in June 2025, up from $1M in September 2024, growing primarily through self-serve adoption by individual developers and small teams.

The addition of Windsurf expanded Cognition’s customer mix from bottoms-up developer subscriptions into larger enterprise contracts, with revenue now split between per-seat and usage-based subscriptions for Devin and multi-seat enterprise deployments of Windsurf’s AI coding platform.

Growth has been driven by rapid adoption of Devin in the developer community, where tens of thousands of users subscribe for autonomous coding assistance, alongside Windsurf’s top-down enterprise sales motion.

In September 2025, Cognition closed a $400M funding round led by Founders Fund, bringing its valuation to $10.2B.

Previously, Cognition raised approximately $500 million in a Series C round in August 2025 led by Founders Fund, valuing the company at $9.8–10 billion. This followed a $21 million Series A in March 2024 and a $175 million round one month later, both also led by Founders Fund.

Total funding raised amounts to approximately $696 million across all rounds. Investors include Founders Fund, 8VC, Khosla Ventures, and Pear VC.

Cognition is an autonomous AI software engineer, referred to as Devin, that processes natural language tickets to execute complete development tasks from start to finish. Unlike code completion tools, Devin analyzes entire repositories, builds contextual understanding, and operates independently within containerized development environments to write, test, and deploy code.

The platform processes English-language requirements such as "Add OAuth2 login flow to our Flask app" and autonomously plans the implementation. Devin initiates a development environment using the recently acquired Windsurf IDE, writes or modifies code, runs tests, iterates until tests pass, and opens pull requests with confidence indicators. The Devin 2.1 release introduced a confidence meter that quantifies the probability of task success, enabling teams to allocate resources more efficiently.

Core features include large codebase analysis through vectorized project graphs, the DeepWiki server for repository comprehension, and BlockDiff snapshotting for rapid rollback during failed experiments. Devin integrates with GitHub, GitLab, Bitbucket, Jira, Linear, and Slack, and offers a REST API for integration into CI/CD pipelines. The platform can also remain available to address code review feedback and support ongoing development discussions.

In October 2025, Cognition launched SWE‑1.5, a frontier‑size coding model served via Cerebras at up to 950 tok/s—about 6x faster than Haiku 4.5 and 13x faster than Sonnet 4.5—and made it available in Windsurf; the release pairs model, inference, and agent harness to improve real‑world speed and reliability for Devin/Windsurf users.

In November 2025, Cognition introduced Windsurf Codemaps, AI‑annotated structured maps of codebases powered by SWE‑1.5 and Claude Sonnet 4.5. Codemaps accelerates onboarding and debugging by grounding navigation to exact lines and visual node graphs, with “trace guide” expansions for deeper context. Codemaps also plugs into Cascade prompts via @{codemap}, improving Devin’s task performance with precise, just‑in‑time context.

Cognition operates a B2B SaaS model with usage-based pricing through Agent Compute Units (ACUs) that scale with the complexity and duration of development tasks. The company has transitioned from an enterprise-only approach to a tiered structure that includes a $20 monthly Core plan alongside higher-tier enterprise contracts, broadening its customer base while maintaining premium pricing.

The ACU pricing model ties costs directly to value delivery, as customers are charged based on the computational resources and time required for Devin to complete tasks. This structure supports expansion revenue as teams undertake more complex projects and manage larger codebases. The recent acquisition of Windsurf adds a dedicated agentic IDE and development infrastructure, reducing reliance on third-party tools and creating additional integration opportunities.

Cognition's go-to-market strategy addresses both individual developers seeking productivity improvements and enterprise teams aiming to enhance development capacity. The platform's autonomous capabilities appeal to organizations managing engineering talent shortages or aiming to accelerate development timelines.

GitHub Copilot reports 20 million lifetime users and a 90% adoption rate among Fortune 100 companies, supported by its integration with IDEs and Microsoft's cloud ecosystem. Google's Gemini Code Assist and Amazon Q similarly integrate AI coding tools with their respective cloud platforms, creating switching costs tied to infrastructure dependencies. These incumbents leverage established developer relationships and subsidize AI features through revenue from their broader platforms.

Replit Agent provides browser-based development with integrated deployment pipelines and has partnered with Microsoft to function as a lightweight Azure front door. Sourcegraph Cody focuses on code-graph context to minimize hallucinations and has recently secured enterprise funding. JetBrains AI Assistant integrates deeply across the company's development tool suite. These platforms differentiate themselves by offering comprehensive development environments rather than standalone AI tools.

Cursor, Tabnine, and other AI-first coding tools specialize in intelligent code generation and completion. Sweep AI and CodeRabbit address specific workflows such as automated code review and bug fixing. While these tools often provide advanced AI capabilities in niche areas, they lack the broader platform integration increasingly prioritized by enterprise customers. Competition is intensifying as 90% of development teams now experiment with multiple AI coding tools concurrently.

The Windsurf acquisition enables Cognition to develop multi-agent assembly lines capable of handling QA, DevOps, and low-code application generation. This expansion shifts focus from individual coding tasks to managing entire development workflows, with the potential to capture budgets allocated to application lifecycle management and platform engineering. Multi-agent systems may automate testing pipelines, deployment processes, and infrastructure management.

Cognition's REST API and integrations with Jira, Linear, and Slack create opportunities to expand further into enterprise development workflows. The platform could extend its functionality to include project scoping, technical documentation generation, and developer onboarding processes. This would allow Cognition to extract greater value from enterprise relationships while addressing a broader range of software development lifecycle requirements.

The introduction of a $20 monthly Core plan significantly increases Cognition's addressable market by including individual developers, bootcamp students, and small businesses globally. Devin's cloud-based architecture eliminates local hardware dependencies, making the platform accessible in emerging markets where device specifications might otherwise pose a barrier. Early adoption by companies such as Nubank in Latin America indicates potential for international growth, while planned European infrastructure investments are expected to support adoption in regions with data residency requirements.

Model dependency: Cognition's autonomous capabilities depend on large language models managed by third parties, which control their performance, availability, and pricing. Any decline in model quality, substantial price increases, or restrictions on access could materially affect Devin's functionality and Cognition's unit economics.

Security concerns: Autonomous AI systems with repository access and code generation features pose significant security and intellectual property risks for enterprise customers. A single vulnerability or data breach involving Devin could undermine customer trust and constrain enterprise adoption, particularly in regulated sectors.

Talent competition: The autonomous coding market is drawing significant investment and talent from established technology companies and well-funded startups. Cognition must continually enhance its AI capabilities while competing for limited AI engineering talent against organizations with greater financial resources and more established research teams.