Open-source cloud runtime that gives AI agents secure, isolated sandboxed environments — with browser, filesystem and compute — to execute code and use real tools.
The bet
Every AI agent gets its own instant-on, sandboxed cloud computer — and that open-source runtime becomes the default layer under agentic apps.
$32M total by July 2025: $11.5M seed led by Decibel Partners (Oct 2024) and $21M Series A led by Insight Partners (Jul 2025).
Czech childhood friends Vasek Mlejnsky and Tomas Valenta built developer tools together for years, including Devbook; they pivoted to E2B in early 2023 and released the open-source sandbox runtime in March 2023, betting agents needed cloud infrastructure designed for them rather than for humans.
E2B's repo hit Python #1 on GitHub Trending in May 2023 and grew to 13.6k stars by Sep 2026. Sandbox launches scaled from 20,000 in April 2024 to 2 million in September 2024, then past 6 million cumulative since January 2024. Customers include Hugging Face, LMArena, Perplexity, Manus, Groq, Genspark and Lindy. Decibel led a $11.5M seed in Oct 2024; Insight Partners led a $21M Series A in Jul 2025, bringing total funding to $32M, with the company claiming 88% of the Fortune 100 signed up.
Still private and scaling as of 2026-09-01: E2B is building a standardized 'agent cloud' with secrets vault, observability, shared context and fleet orchestration, aiming to make its open-source sandbox protocol the universal runtime for AI agents.
The founders bet on infrastructure beneath the wave, not another agent product, so they rode every customer's agent as it shipped.
Open source removed trust barriers: enterprises could inspect and self-host the sandbox before paying.
Traction compounded visibly: 20k to 2M monthly sandboxes in five months made the enterprise claims credible to investors.
Sandboxes with browser, files and compute covered agent needs that plain model APIs never could, so the wedge stayed defensible.
Bet on the missing layer, not the app: as agents arrived faster than infrastructure for them, E2B built the runtime beneath every agent instead of another agent product.
As of 2026-09-01, E2B is scaling engineering and go-to-market from San Francisco with $32M raised. Its open-source sandbox runtime is used by Hugging Face, LMArena, Groq, Perplexity, Genspark, Lindy and Manus, and the company reports hundreds of millions of sandboxes launched since October 2024, with 13.6k GitHub stars. The roadmap is a standardized agent-cloud layer — secrets vault, sandbox observability, shared context and single-console fleet orchestration — deployable on AWS, Azure, GCP, Kubernetes, Cloudflare or a company's own VPC.
FOLLOW THE EVIDENCE
The sources
- We Raised $21M to Give Fortune 100 Cloud for AI Agents E2B (official blog) · e2b.dev ·
- E2B shares its vision of sandboxed, cloud environments for every AI agent after raising $21M in funding SiliconANGLE · siliconangle.com ·
- Czech Startup E2B Secures $11.5M to Develop Cloud Infrastructure for AI Agents The AI Insider · theaiinsider.tech ·
- E2B: The AI Agents Cloud Decibel (investor interview) · decibel.vc
- e2b-dev/E2B star history star-history.com · star-history.com ·