E2B Raises $21M to Build the Cloud Designed for AI Agents

E2B, the company pioneering open-source cloud infrastructure built specifically for AI agents, has raised a $21 million Series A funding round. The round was led by Insight Partners and included participation from Decibel, Sunflower Capital, Kaya, and angel investor Scott Johnston, former CEO of Docker. The investment reflects growing industry momentum around the need for a new kind of infrastructure—one designed not for humans, but for autonomous AI systems operating independently at scale.

The Origin of E2B

The idea for E2B was born when co-founders Vasek Mlejnsky and Tomas Jasovsky were developing their own AI agent and encountered a hard limitation: today’s infrastructure simply wasn’t built for software that thinks and acts on its own. Most cloud environments were made with humans in mind—interfaces, file systems, and workflows all designed around human interaction.

This insight led them to a simple but powerful idea: if humans use computers to do work, AI agents should have access to computers too—virtualized environments equipped with browsers, file storage, execution capabilities, and access to essential tools. The result was E2B: an open-source, sandboxed runtime that gives AI agents their own ephemeral, isolated “computers” in the cloud. These sandboxes can be launched in milliseconds, run securely, and scale up or down to meet the demands of any workflow.

Serving Fortune 100s and AI’s Fastest Innovators

E2B’s technology has already reached remarkable adoption. According to the company, 88% of Fortune 100 companies have used its sandbox infrastructure in some capacity. Its customer base includes foundational model labs and AI-native startups such as Hugging Face, Perplexity, Groq, and Manus.

These companies use E2B to support a range of high-value agentic workflows, including:

• Autonomous research agents that browse the web, summarize findings, and generate analyst-grade reports
• Agents that write and deploy full-stack web applications from a single prompt
• Secure execution environments for reinforcement learning loops and LLM evaluations
• Data analysis agents that interpret datasets and generate visual insights in seconds

These use cases span across finance, logistics, software development, enterprise support, and other sectors where speed, automation, and secure execution are critical.

Infrastructure for the Agent Era

The company’s core product—a lightweight, open-source sandbox—acts as a runtime standard for how AI agents access external tools, interact with data, and complete complex multi-step tasks. Unlike traditional Docker-based approaches, which are often too heavy or complex for fast-scaling agents, E2B’s solution is optimized for high-speed, short-lived, and secure deployments.

Each sandbox provides a clean, self-contained environment with access to a browser, file system, code execution tools, and optional plugins. Enterprises can run thousands—or millions—of these agent computers concurrently. This plug-and-play model allows organizations to focus on improving agent performance rather than building and securing infrastructure from scratch.

Because of its open-source architecture and focus on enterprise deployment, E2B appeals to organizations that want maximum control and customization. The platform can be deployed inside a company’s own cloud, supporting strict data governance and regulatory compliance needs.

A Glimpse into the Future of AI Infrastructure

E2B’s funding marks more than a company milestone—it points to a broader shift happening across the AI landscape. As AI agents become more autonomous and capable, the surrounding infrastructure must evolve. The current cloud paradigm, designed around human users and predictable applications, is poorly suited for the demands of agent-based workflows that are ephemeral, compute-intensive, and dynamic by design.

This transformation is already raising industry-wide questions: How should autonomous agents be audited and monitored? What security standards will govern agent-to-agent interaction? How do organizations control access, memory, and behavior in rapidly spun-up compute environments?

E2B represents one of the first serious attempts to answer these questions at scale. By offering a clean standard for running agents safely and effectively, the company is helping shape a future where AI agents are no longer isolated in R&D environments—but fully integrated into core business processes.

Just as Kubernetes became the orchestration layer for microservices, platforms like E2B may become the default execution layer for intelligent software agents. If this happens, we may look back on infrastructure not just as a technical layer, but as the primary enabler—or blocker—of the AI-native enterprise.