Daytona, a sandbox for AI agents, becomes unicorn with $60 million round

Daytona, a start-up that rents out secure, disposable computers for AI agents to work in, has raised $60 million in a Series B funding round that values it at $1 billion, the company announced on Tuesday, October 20, 2026. The round was led by FirstMark Growth, and existing investors supplied about $50 million of it, according to reports of the deal. The new money brings Daytona's total funding to $91 million. The company says it is building what it calls an "agent cloud": computing infrastructure designed for software agents rather than for human programmers.

AI agents are programs that do not just answer questions but carry out tasks: they write and run code, browse websites, fill in forms and test software. To do that safely, an agent needs somewhere to work that is cut off from the company's real systems, so that a mistake or a malicious instruction cannot damage files or leak data. In computing, that walled-off space is called a sandbox. Daytona sells sandboxes on demand. An agent can start one in moments, do its work, pause it, copy it and throw it away when finished. The company says its sandboxes can keep their state when paused and can be duplicated, so an agent can try several approaches to a problem from the same starting point and keep the one that works. It offers environments running Linux, Windows and macOS, as well as machines with graphics processors.

The business is a bet on a shift in who uses cloud computing. For two decades, cloud services were built around people: developers logging in, running servers and writing code. If AI agents become common in companies, they will need computers too, and in far greater numbers, because a single person might direct dozens of agents at once. Daytona changed course to serve that market. Reports say it previously sold development environments for human programmers and rebuilt its product as a runtime for AI agents.

The valuation shows how much investors will pay for picks-and-shovels businesses in AI, the companies that supply infrastructure rather than building the AI models themselves. Daytona's round comes in the same week that another agent-focused company, Town, reached unicorn status. Reaching a $1 billion valuation with $91 million raised in total is a sign that investors expect fast growth, although the company has not disclosed how much revenue it actually earns. The 25-fold growth figure is the company's own and has not been independently verified.

For Indian founders and developers, the deal is relevant in two ways. Many Indian start-ups are building AI agents for customer service, banking and software testing, and they face the same need for safe environments in which agents can act. And the list of backers includes Aravind Srinivas, the Indian-born co-founder of Perplexity, one of several founders of Indian origin now investing in the AI infrastructure layer.

Short-term: Daytona will use the money to expand capacity and compete for companies deploying AI agents at scale. Competition among agent-infrastructure providers is likely to push prices down for developers.

Long-term: If AI agents become a routine part of office work, sandbox services could become as basic a part of the cloud as storage and servers are today. That would also raise new questions for security teams about how agents are monitored and controlled.

Who is affected: Software companies building AI agents, including Indian start-ups; enterprise security teams; cloud providers that may face new competition; and investors tracking AI infrastructure valuations.

Questions and Answers