Stack L8
Cognition
Devin maker at $10.2B; acquired Windsurf, piloting autonomous SWE agents at Goldman Sachs
HQ US
Fit Score
avg of 4 segmentsRecent Business Signals
Strategic Position
What they do best
Selling autonomous agentic SWE (Devin) into high-value enterprises, evidenced by the Goldman Sachs pilot (signal 17459433) targeting thousands of agents.
Their bet
That latency wins agentic coding — SWE-1.5 on Cerebras at 950 tok/s plus SWE-grep ~20x retrieval (signals ee38c466, 5e8274f1) make speed, not raw accuracy, the moat.
Top risk
The Goldman Sachs pilot (signal 17459433) may not convert to paid hundreds-to-thousands of Devin seats within 6 months, leaving the $10.2B valuation propped up almost entirely by Windsurf's acquired ~$82M ARR.
Compared to peers
Direct competitor
Cursor
Cursor monetizes per-seat AI-native IDE usage; Cognition sells autonomous Devin agents priced on work done, not seats occupied.
Substitute
Cursor
Cursor's IDE autocomplete/chat can substitute for low-complexity Devin tasks, capping Cognition's per-seat value claim where full autonomy isn't needed.
Why someone would join
- 1.Valuation jumped from ~$4B to ~$10.2B with a ~$400M Founders Fund round on 2025-09-12 (signal d83818b9), plus an inherited ~$82M ARR / 350+ enterprise base from the Windsurf buy.
- 2.Goldman Sachs (signal 17459433, 2025-07-11) is piloting Devin toward thousands of agents across ~12,000 developers — and hiring is concentrated in enterprise deployment/platform roles (3 of 5 recent reqs) to service it.
Recent Hiring (60 days)
- enterprise SaaS sales1
- developer tools1
- distributed systems1
- sandboxing1
- cloud infrastructure1
Reverse-Hype Watch
- !Capacity/perf claims unbacked by customer signals: SWE-1.5 'near-frontier coding accuracy at ~13x the speed' / 950 tok/s and SWE-grep '~20x faster' are vendor self-reported with no independent benchmark; flagship Goldman Sachs Devin use is a 'pilot scaling toward hundreds-to-thousands', not booked.
- !Stretched valuation: ~$10.2B post-money (2025-09-12) against only disclosed recurring revenue of ~$82M Windsurf ARR (~124x); Devin ARR undisclosed and Goldman pre-production (note: no layoffs reported, so not a classic up-round-with-burn).
- !Performance claims are vendor self-reported with no independent benchmark: SWE-1.5 'near-frontier coding accuracy at ~13x the speed', SWE-grep '~20x faster context gathering'; flagship Goldman Sachs Devin deployment is explicitly a 'pilot scaling toward' hundreds-to-thousands, so capacity claims are not yet backed by production customer signals.
- !Capacity claims unbacked: SWE-1.5 'near-frontier coding accuracy at ~13x the speed', SWE-grep '~20x faster context gathering' are vendor self-reported with no independent benchmark; flagship Goldman Sachs Devin deployment is still a 'pilot scaling toward' (not booked revenue).
- !Performance claims vendor self-reported with no independent benchmark: SWE-1.5 'near-frontier coding accuracy' at '~13x the speed', SWE-grep 'up to ~20x faster context gathering' — capacity/quality claims unbacked by third-party or customer validation.