Devin: Cognition's AI software engineer from $1M to $73M ARR in nine months
Scott Wu's Cognition launched Devin in 2024. Company posts say Devin ARR grew from $1M (Sep 2024) to $73M (Jun 2025), then Cognition raised at $10.2B and later $48B valuations.
- Product
- Devin (Cognition)
- Founder
- Scott Wu
- Launched
- 2024-03
- Model
- Enterprise subscription for AI coding agents and IDE
Published · 2 min read · 2 sources
Key takeaways
- Cognition launched Devin in March 2024. Company blog: Devin ARR grew from $1M in September 2024 to $73M in June 2025, with total net burn under $20M.
- September 2025: over $400M raised at a $10.2B valuation; Cognition says acquiring Windsurf more than doubled ARR.
- September 2026 Series E: over $2B at a $48B valuation; company says run-rate revenue grew from $492M to almost $900M since the prior round.
What it is
Devin is Cognition's autonomous AI software engineer: agents that take coding tasks while humans act as architects. Cognition also owns the Windsurf AI IDE after acquiring it, selling both agents and an IDE (Cognition).
Timeline
- 2024: Cognition founded; Devin launches in March (Cognition).
- Sep 2024 → Jun 2025: company says Devin ARR grew from $1M to $73M, with total net burn under $20M across company history to that point (Cognition).
- 8 Sep 2025: Cognition raises over $400M at a $10.2B post-money valuation led by Founders Fund; post says acquiring Windsurf more than doubled ARR and combined enterprise ARR rose over 30% in the seven weeks after the deal (Cognition).
- 8 Sep 2026: Series E — over $2B at a $48B valuation led by Andreessen Horowitz and Accel. Company: run-rate revenue grew from $492M (May) to almost $900M (Cognition).
Why it worked
- Agent + IDE bundle. Wu argues engineers want both: an IDE when they decide each step, and agents for async tasks; Windsurf filled the IDE side (Cognition).
- Enterprise logos. Named customers across posts include Goldman Sachs, Citi, Dell, Cisco, Ramp, NVIDIA and others (Cognition, Cognition).
- Capital-efficient early growth. Sub-$20M net burn while scaling Devin ARR into the tens of millions is central to the September 2025 narrative (Cognition).
- Low customer overlap pre-deal. Cognition says less than 5% overlap between Devin and Windsurf enterprise customers before the acquisition (Cognition).
What to learn
- Pair complementary surfaces (agent + IDE) instead of forcing one UX to do both jobs.
- Publish burn alongside ARR if you want a capital-efficiency story investors will repeat (Cognition).
- Acquisitions work when GTM and product overlap are low and the combo is obvious to buyers.
Risks and what didn't work
- Company-reported everything. ARR, burn, run-rate revenue and valuations come from Cognition blog posts, not audited filings (Cognition, Cognition).
- Early product was "junior." Wu writes that at launch Devin could nail some tasks but was "still a very junior engineer" for most work (Cognition).
- Valuation vs durability. A jump to a $48B valuation prices in continued hypergrowth (Cognition); model costs, competition from Cursor/Claude Code/Copilot, and enterprise sales cycles are risks.
FAQ
Short answers drawn from the sections above.
What is Devin (Cognition)?
Devin is Cognition's autonomous AI software engineer: agents that take coding tasks while humans act as architects. Cognition also owns the Windsurf AI IDE after acquiring it, selling both agents and an IDE (Cognition).
Why did Devin (Cognition) work?
- Agent + IDE bundle. Wu argues engineers want both: an IDE when they decide each step, and agents for async tasks; Windsurf filled the IDE side (Cognition).
- Enterprise logos. Named customers across posts include Goldman Sachs, Citi, Dell, Cisco, Ramp, NVIDIA and others (Cognition, Cognition).
- Capital-efficient early growth. Sub-$20M net burn while scaling Devin ARR into the tens of millions is central to the September 2025 narrative (Cognition).
- Low customer overlap pre-deal. Cognition says less than 5% overlap between Devin and Windsurf enterprise customers before the acquisition (Cognition).
What can indie builders learn from Devin (Cognition)?
- Pair complementary surfaces (agent + IDE) instead of forcing one UX to do both jobs.
- Publish burn alongside ARR if you want a capital-efficiency story investors will repeat (Cognition).
- Acquisitions work when GTM and product overlap are low and the combo is obvious to buyers.
What didn't work for Devin (Cognition), and what are the risks?
- Company-reported everything. ARR, burn, run-rate revenue and valuations come from Cognition blog posts, not audited filings (Cognition, Cognition).
- Early product was "junior." Wu writes that at launch Devin could nail some tasks but was "still a very junior engineer" for most work (Cognition).
- Valuation vs durability. A jump to a $48B valuation prices in continued hypergrowth (Cognition); model costs, competition from Cursor/Claude Code/Copilot, and enterprise sales cycles are risks.
Sources
Every link cited above, in order of first use.
Related case studies
More in AI tools →ElevenLabs: voice AI from first product to $330M ARR and an $11B Series D
Mati Staniszewski and Piotr Dabkowski's ElevenLabs closed 2025 above $330M ARR, then raised a $500M Series D at an $11B valuation in February 2026.
Tome: 20 million users of an AI slides app, then a pivot away from slides
Tome rode the ChatGPT wave with an AI presentation tool and tens of millions of mostly free users. It cut staff, moved to sales teams, dropped slides; its founders then launched the AI CRM Lightfield.
Framer: designer-loved site builder hits $50M ARR and a $2B valuation
Amsterdam's Framer raised a $100M Series D at a $2B valuation in August 2025. A spokesperson said $50M ARR that year, break-even for a year, and a goal of $100M ARR next.