Cognition Raises $1 Billion at $26 Billion Valuation

Cognition announced it raised more than $1 billion at a $26 billion post-money valuation, per a company blog post and corroborating coverage by Bloomberg and Axios. The company blog reports a run-rate revenue of $492M and states enterprise usage has increased by more than 10x since the start of the year. Reporting by The Next Web cites CEO Scott Wu saying that Devin now writes over 90% of Cognition's internal code. The financing was led by Lux Capital, General Catalyst, and 8VC, with participation from Founders Fund, Ribbit Capital, Atreides Management and other investors, according to the company and multiple media reports.

Editorial analysis - technical context: Public reporting describes Cognition's flagship product, Devin, as an end-to-end coding agent that plans, writes, debugs, and deploys software across multi-step workflows rather than offering line-by-line code completion. The Next Web and the company note Cognition routes workloads across a mix of proprietary models and third-party models from OpenAI and Anthropic; that multi-model approach is presented in coverage as a way to optimise price and performance for different tasks.

A $1 billion round at a $26 billion valuation places Cognition among the highest-valued startups focused on software engineering automation. Media reporting highlights the company's rapid reported revenue growth, from roughly $37M to $492M in 12 months, per The Next Web, and customer deployments at large enterprises and government organizations listed in the company blog including Goldman Sachs, Mercedes-Benz, Citi, Santander, the U.S. Army, and the U.S. Navy. Coverage frames the raise as a validation of investor appetite for agentic automation in software development and as part of a broader surge in AI-native developer tooling.

Editorial analysis: Observers will monitor whether Cognition sustains the revenue acceleration the company reports and how effectively multi-model orchestration translates into predictable unit economics for customers. Industry watchers will also be interested in third-party verification of internal-usage claims such as the 90% figure reported by The Next Web and how large enterprise customers measure reliability and security when an agent carries out end-to-end engineering tasks. Finally, the company's approach to model procurement and cost management is relevant to practitioners evaluating trade-offs between own-model investments and orchestration of external providers.

The figures above come from Cognition's blog post and contemporaneous reporting by Bloomberg, Axios, The Next Web, and other outlets. Where coverage paraphrases executive statements, those paraphrases are attributed to the reporting outlets; the company has posted its own blog announcing the raise and the revenue/run-rate figures.

Editorial analysis: Teams evaluating agentic automation should treat reported customer outcomes and vendor internal-usage metrics as starting points for technical due diligence. Standard checks include red-team exercises, reproducibility of generated artifacts, integration testing across CI/CD pipelines, and economic modelling of token, inference, and orchestration costs compared with developer time saved.