Daytona agent cloud hits $1B after $60M Series B

Daytona Platforms has raised 60 million dollars in a Series B funding round at a valuation of 1 billion dollars, according to reporting by TechCrunch on October 23, 2026. The round was led by FirstMark Growth and brings the company's total funding to 91 million dollars.

The New York-based startup is building cloud infrastructure designed specifically for AI agents. The funding will expand the Daytona agent cloud, and its pitch is simple: agents do not behave like traditional software, so the cloud that runs them should not look like traditional cloud infrastructure either.

FirstMark also led Daytona's 24 million dollar Series A round in February 2026. The latest round drew a long list of well-known angel investors, including Stripe chief executive Patrick Collison, Shopify chief executive Tobi Lutke, Perplexity chief executive Aravind Srinivas, Zoom founder Eric Yuan, Datadog chief executive Olivier Pomel, Quora chief executive Adam D'Angelo, and Mercor chief executive Brendan Foody.

Institutional participants named in the reports include Qualcomm Ventures, Samsung Next, and Okta Ventures. Existing investors also increased their commitments in the round, a signal that backers see the agent infrastructure market expanding rather than cooling.

Why agents need a different kind of cloud

Traditional cloud platforms were designed for long-lived services that humans plan in advance. A web server or a database runs continuously, scales on a predictable schedule, and stays put. AI agents work differently.

According to Daytona founder Ivan Burazin, agents spin up thousands of short, stateful jobs that nobody planned in advance. An agent might open a browser, run code, test a change, and then disappear, only for another agent to start a similar job seconds later.

That pattern creates a mismatch. Spinning up a conventional virtual machine can take minutes, which is far too slow when an agent needs a workspace right now. Leaving machines running idle between jobs wastes money. Sharing one environment across many agents risks leaking data between tasks.

Daytona's answer is to give every agent its own computer. Each agent gets an isolated sandbox with its own operating system, storage, and tools. The company says those sandboxes can be configured in about 60 milliseconds by pre-loading snapshots onto server drives, a sharp contrast with the minutes a conventional setup can take.

The sandboxes support Linux, Windows, and macOS, along with graphics processing units for heavier workloads. That range matters because agents are increasingly asked to do real desktop work, from testing native Mac applications to operating Windows software and working with three-dimensional design tools.

A sandbox that pauses, forks, and disappears

Daytona's platform is built around the lifecycle of an agent job rather than the lifecycle of a server. Environments start quickly, keep their state when paused, fork into parallel copies when an agent wants to try several approaches at once, and disappear when the task is finished.

That design fits how agent development actually works. A coding agent may need to test a change, roll it back, and try a different approach without losing progress. A research agent may need to run many searches in parallel. A customer-service agent may need a clean, isolated workspace for every conversation.

Customers named in coverage of the round include Anthropic, OpenAI, Nvidia, Meta, Cursor, Cognition, and Cohere. Those companies use Daytona sandboxes to train and test models and to run agent workloads, according to Dealroom. Data company Snorkel alone has provisioned more than one million sandboxes on the platform, the same report noted.

Daytona says demand has accelerated sharply in 2026. Reporting on the round put revenue run-rate growth at 25-fold between January and September, with average monthly growth of about 51 percent. The company did not disclose absolute revenue figures, so the size of the business behind the growth rate remains unclear.

What the unicorn valuation says about agent infrastructure

A 1 billion dollar valuation at the Series B stage is a bet on growth rather than on current scale. Dealroom noted that the valuation sits in the top tier of United States enterprise-software Series B rounds, which means investors are paying for where they expect the market to go.

That expectation is easy to understand. The agent wave has moved from chatbots that answer questions to software that takes actions: writing code, browsing the web, operating applications, and carrying out multi-step tasks with limited human supervision. Every one of those actions needs a safe place to happen.

The same shift is driving other parts of the agent stack. Google Cloud recently introduced its Agent Compute Engine, a purpose-built service for agent workloads, as reported in coverage of Google's agent compute launch. Trust and identity layers are also emerging, with platforms such as WISeQey's agent trust platform aiming to verify which agent is acting and on whose authority.

The Daytona agent cloud sits in the middle of that stack, at the compute layer. If agents become the main way work gets done, the company argues, the cloud that hosts them will look less like a server farm and more like a fleet of disposable computers, created on demand and discarded when the job ends.

Sceptics have an obvious question, and Burazin has acknowledged it: why does the market need Daytona when large cloud providers already sell virtual machines? The company's answer is speed, isolation, and a design built around short-lived, stateful agent jobs rather than long-lived services.

Investors in the Series B appear to accept that argument, at least for now. Whether Daytona can hold its lead as the large cloud providers add their own agent features will be the test of the next two years. For the moment, the funding gives the startup room to scale its platform while agent workloads keep growing.

Sources for this report include TechCrunch's report on the Daytona funding round, Dealroom's coverage of the Series B, and Pulse 2.0's report on the agent cloud expansion.