Daytona, a San Francisco startup building cloud infrastructure for AI agents, announced Tuesday that it has closed a $60 million Daytona Series B at a $1 billion valuation, turning the three-year-old company into the newest unicorn in the agent economy. The round was led by FirstMark, according to Tech.eu, and brings Daytona's total funding to about $91 million.

The Daytona Series B is a bet on a simple thesis: AI agents need a place to work, not just a place to think. Daytona provides secure, isolated computing environments — sandboxes — where agents can execute code, call software tools, and complete tasks without touching a user's laptop or sensitive company data. Developers launch those environments through a single API across Linux, Windows, and macOS, and each agent gets compute that expands or contracts with the task at hand. The platform also lets agents pause, resume, or try a different approach without losing progress, according to FinSMEs, combining startup speed with flexible compute, persistent state, and granular controls for training, evaluation, and production workloads.

Why agents need their own cloud

Daytona co-founder and CEO Ivan Burazin frames the product with a sharp division of labor. As reported by Nile1, Burazin describes the inference cloud as where the agent thinks, and Daytona's cloud as where the agent works. Every autonomous agent — one that writes code, browses the web, or operates business software — needs a computer to act on, and existing hyperscalers take 30 seconds to several minutes to spin up an isolated machine. Daytona does it in about 60 milliseconds by keeping pre-configured machine snapshots on physical drives and attaching compute and memory on demand, according to Nile1's analysis.

That speed gap matters because of where the agent trend is heading — and it is part of what the Daytona Series B is pricing in: infrastructure that fails safe by design, even as agent fleets scale into the thousands.

Burazin told Tech.eu that companies are going to manage thousands of digital employees, and each one needs a safe place to fail — Daytona's job is to make sure one rule-breaking agent doesn't spill over into company-wide chaos. The sandbox model gives each agent an isolated room: even if it malfunctions, the blast radius stops at the sandbox walls.

The news continues a fast run of infrastructure plays for the agent economy. Earlier coverage on this site looked at live-web access layers for agents, such as the Context.dev raise for agents' live web access, and at collaboration frameworks like the Wave AI framework for agent collaboration — all of them pointing at the same shift: the industry is moving from asking how to build agents to asking how to run them safely at scale.

The cap table is a who's who of agent builders

The investor roster for the Daytona Series B reads like a directory of the people building agentic products. Founder-CEOs in the round include Patrick Collison of Stripe, Tobi Lütke of Shopify, Aravind Srinivas of Perplexity, Adam D'Angelo of Quora, and Eric Yuan of Zoom, alongside Mercor's Brendan Foody and Datadog's Olivier Pomel, according to Tech.eu.

Strategic capital joined too, with Qualcomm Ventures, Samsung Next, and Okta Ventures taking part, plus Alumni Ventures, Evergreen Fund, and Torch Capital. FirstMark is no newcomer: as reported by Crypto Briefing, the firm also led Daytona's $24 million Series A, making this at least its second time leading a priced round. Burazin told Fortune that roughly $50 million of the round came from existing investors, as reported by JBizNews — and those backers backed the claim with larger checks: Pace Capital doubled its investment, Upfront Ventures tripled, and E2VC quadrupled theirs, according to Tech.eu.

The customer list helps explain the enthusiasm. Daytona names Anthropic, OpenAI, NVIDIA, Cursor, and Cognition among its customers, and says its base also includes Meta, Cohere, Mercor, Handshake, Surge AI, SambaNova, Postman, Bolt.new, n8n, and Clay, according to Tech.eu. For AI training and evaluation workloads, customers including Snorkel and Trajectory use the platform to cut processing times. Daytona also reported a 25-fold increase in its revenue run rate between January and September — the kind of growth line that turns a sandbox product into a billion-dollar story.

From sandbox tool to agentic cloud

Daytona's climb has been fast. According to Crypto Briefing, the company raised a pre-seed of about $2 million in 2023 and a $5 million seed round in 2024, then the $24 million Series A — and now the $60 million Daytona Series B takes the total to roughly $91 million. Going from a $2 million pre-seed to a $1 billion valuation took only a few years.

With the new funding, Daytona plans to expand its agent cloud platform beyond its initial sandbox product, scale its computing infrastructure, and grow its international team, according to FinSMEs — hiring across San Francisco, Croatia, and Singapore. The expansion suggests the company sees the sandbox as just the starting wedge: the bigger prize is the full operating layer where companies run fleets of agents in production, governed and isolated by default.

Investors keep funding that vision because the pain it solves is visible to anyone running agents today. Every agent that can act — writing code, moving money, touching customer data — is an agent that needs guardrails around where it runs. The Daytona Series B puts a billion-dollar price tag on the idea that the cloud those agents work in should be designed for them from the start.

Sources

Related coverage: Continue with the latest GenZ NewZ headlines, more ai news coverage, Accused OpenAI Spy to Stay Jailed Before Trial, Anthropic in Talks for Second Massive Compute Deal, and DeepSeek Federal Ban: DHS Orders Agencies to Drop It.