Bolt.new's browser-AI bet: $0 to $40M ARR in five months after a near-death pivot

Bolt.new is a browser-based AI development agent: type a prompt and it generates, runs, edits and deploys full-stack web apps with no local setup, monetized through subscriptions.

AI agents would build, run and deploy full-stack apps inside a browser tab; WebContainers gave Bolt instant local-feel previews with no per-user server cost.

Total ~$105.5M, including a previously unreported $22M Series A and an $83.5M Series B in January 2025 led by Emergence Capital and GV at roughly a $700M valuation, with Madrona Venture Group, Conviction and Mantis.

Eric Simons and Albert Pai founded StackBlitz in 2017 around WebContainers; after seven years and millions of developer sessions the company was on the brink of shutting down. In mid-2024 the latest LLMs proved AI could build quality apps end-to-end, so the team pivoted and relaunched as Bolt in October 2024 with a single tweet.

Within 30 days Bolt went from $0 to $4M ARR; six months later it was at $40M ARR with ~1M monthly active users and an engineering team of about 15. In January 2025 the company raised an $83.5M Series B led by Emergence Capital and GV at roughly a $700M valuation, bringing total funding to about $105.5M; the core reference codebase was released open source, and Bolt later added mobile app building (Bolt + Expo) and enterprise features as vibe-coding rivals crowded in.

Still running and scaling as of September 2026; after a summer 2025 hype peak, Barclays and Google Trends data showed Bolt's traffic down ~27% from June 2025, and CEO Eric Simons admitted churn was high across the category, pushing the company to add stickier features and new subscription tiers.

The bet is explicit and falsifiable: AI agents, not human devs, become the primary interface for building software, with the browser as the runtime.

It is a rare documented near-death pivot: seven years of an infrastructure nobody seemed to want became the moat when the interface shifted.

Growth numbers are dated and sourced: $4M ARR in 30 days, $40M ARR in five months, ~1M monthly users, and an $83.5M round at ~$700M.

The Google Trends angle is documented: searches spiked through summer 2025, then traffic fell 27% from June — a real search-rising story with a visible peak.

It shows the failure mode of the category too: high churn forced Bolt to add stickier features and new subscription tiers.

A seemingly failed infrastructure bet became the moat when AI shifted the interface: WebContainers was the one runtime where an agent could build and run apps instantly, with no per-user server cost.

As of 2026-09-02 Bolt.new is still running and scaling: StackBlitz has added mobile app building (Bolt + Expo), a hosted cloud backend, and enterprise offerings, while keeping the core reference codebase open source. After the summer 2025 hype peak, Barclays and Google Trends data showed traffic down roughly 27% from June; CEO Eric Simons publicly admitted churn was high across vibe-coding platforms and said services must build more retentive products, so Bolt added new features and an updated subscription in August 2025. No newer funding round was found in this pass.