Daytona Hits a 1 Billion Dollar Valuation With a 60 Million Dollar Series B to Build the Cloud Where AI Agents Work

Daytona, a startup building what it calls the agentic cloud — cloud infrastructure purpose-built for AI agents rather than human developers — said Tuesday it has raised a 60 million dollar Series B at a 1 billion dollar valuation. Fortune first reported the round, which was led by FirstMark and brings total funding to roughly 91 million dollars, about 50 million of it from existing investors doubling down.

The cap table reads like a founder referral list. Alongside FirstMark, the round includes Stripe co-founder Patrick Collison, Shopify CEO Tobi Lütke, Perplexity CEO Aravind Srinivas, Quora founder Adam D'Angelo, Zoom founder Eric Yuan, Mercor co-founder Brendan Foody and Datadog CEO Olivier Pomel, plus strategic backers Qualcomm Ventures, Samsung Next and Okta Ventures. Existing investors increased their commitments too: Pace Capital doubled, Upfront Ventures tripled and E2VC quadrupled. FirstMark also led the 24 million dollar Series A in February.

The pitch is a timing argument. An AI agent that writes code, browses the web or operates business software needs a computer to do it on, and virtual machines from AWS, Azure or Google Cloud take anywhere from 30 seconds to several minutes to configure — far too slow, Daytona argues, for a world where companies run thousands of agents that each need a different software stack per task. Daytona pre-distributes snapshots of configured machines onto server drives and attaches compute on demand, which lets it configure a new sandbox in about 60 milliseconds, roughly the time a blink takes. Agents can also clone a running machine to try two approaches in parallel, then keep forking until one works.

The customer list suggests the infrastructure layer is real. Daytona serves Anthropic, OpenAI, NVIDIA, Cursor and Cognition directly, with Meta, Cohere, Mercor, Surge AI, Bolt.new and n8n also on the books. AI data company Snorkel, which benchmarks models for customers, has provisioned more than one million sandboxes on the platform. Sandboxes double as a safety mechanism: an agent running amok inside an isolated VM damages only that VM, which matters as enterprises worry about agents taking unauthorized actions or leaking data.

The commercial numbers are company-reported and unaudited. Daytona says its revenue run rate grew 25-fold between January and September, averaging 50.9 percent month over month, without disclosing absolute figures. The company got here by pivoting hard: founded in 2023 by Croatian tech veterans Ivan Burazin, Goran Draganić and Vedran Jukić to automate development environments for enterprises, it gave away its original customer base and rebuilt for agents with about three months of cash left. With the new round it plans to expand beyond sandboxes into a broader agent cloud platform and hire across San Francisco, Croatia and Singapore. The round also makes Daytona the third unicorn from the Croatian tech ecosystem, after Infobip and Rimac.

The broader signal is that agent infrastructure is becoming its own market, distinct from the model layer above it and the hyperscalers below it. Startups are carving out pieces — E2B for code sandboxes, browser-isolation vendors for web agents — and the hyperscalers will not stand still. But if agentic workloads really do demand millisecond spin-up, per-task tool stacks and fork-and-branch experimentation, that is a workload profile the existing clouds were never designed around. Burazin frames the demand side bluntly: companies are going to manage thousands of digital employees, and each one needs a safe place to fail.