Daytona Hits US$1B Valuation in Series B for the Agent Execution Cloud
On October 20, Daytona Platforms announced a $60 million Series B round at a $1 billion valuation, led by FirstMark Growth. The three-year-old startup, based in San Francisco with operations in New York, has now raised a total of $91 million since its founding in 2023. But the number that draws the most attention isn't the money itself — it's what it signals: the market believes there's an entirely new infrastructure layer that needs building around autonomous AI agents, and Daytona is one of the few companies with real traction in this space.
Why Daytona Exists
Daytona's thesis is straightforward but ambitious. According to CEO and co-founder Ivan Burazin, AI agents don't work like the traditional services the cloud was designed for. "Agents don't work like the long-lived services today's cloud was designed for — they spin up thousands of short, stateful jobs nobody planned in advance, and they need a cloud built for that," Burazin wrote in his announcement post.
The analogy the company uses is direct: each agent needs its own computer. Not an EC2 instance that takes minutes to boot and stays running until manually shut down, but machines that start in milliseconds, pause while keeping state intact, fork into parallel copies, and disappear when the job is done. In practical terms, that means full sandboxes with Linux, Windows, macOS and GPUs on a single platform.
"Think of it as an Agent Execution Cloud. It isn't inference, observability, or the dashboards people use to manage agents. It's the execution layer: computers, browsers, storage, and anything else an agent needs to do its job," the company explained in its blog.
Daytona was founded in 2023 by Burazin, Goran Draganić and Vedran Mudronja — veterans of developer tooling. Burazin had previously founded Codeanywhere (acquired by JetBrains) and established the Shift conference, now one of the leading global developer events. The prior experience with dev tools wasn't incidental: Daytona began as a sandbox for AI agents and evolved into a full execution platform.
Who's Betting
The Series B investor panel reads like a Silicon Valley directory. In addition to FirstMark (which also led its $24 million Series A in February 2026), participants included Patrick Collison (Stripe), Tobi Lütke (Shopify), Aravind Srinivas (Perplexity), Eric Yuan (Zoom), Olivier Pomel (Datadog), Adam D'Angelo (Quora) and Brendan Foody (Mercor). Strategic investors include Qualcomm Ventures, Samsung Next and Okta Ventures. Existing investors E2VC, Upfront Ventures, FirstMark and Pace Capital have participated in their second, third or fourth consecutive rounds.
Matt Turck, partner at FirstMark, said in a statement: "We believe the next infrastructure shift is from human-centric cloud primitives to agent-native ones. Daytona's breakthrough in making 'a computer for every agent' practical — instant startup, persistent state and tooling to write code, use Git and execute safely at scale — is a foundational building block for the agentic economy."
Notably, about $50 million of the round came from existing investors, meaning early backers are doubling down on the same bet. Burazin revealed that investors repeatedly asked why the market would need Daytona when "AWS can do this thing" — and Daytona's response is that existing infrastructure is fundamentally unsuited to agent workloads.
Real Traction, Not Just Theory
What sets Daytona apart from many infrastructure startups is that it already has real customers — and first-line customers at that. The company provides sandboxes to Anthropic, OpenAI, Nvidia, Cursor, Cognition, Meta and Cohere. Customers use the virtual machines to let their models practice tasks, execute AI-generated code in safe environments and perform reinforcement learning. The data company Snorkel has already provisioned over one million sandboxes on the platform.
The growth numbers are striking. Burazin told Fortune that revenue run rate grew 25-fold between January and September, averaging 50.9% monthly growth. By late October 2026, the revenue run rate was approximately $2.2 million ARR (annualized revenue based on the last 12 months).
Daytona claims to be the only provider with Linux, Windows, macOS and GPU sandboxes on a single platform. An agent can optimize a CUDA kernel, test a native Mac app, drive a Windows desktop or work in 3D CAD without leaving the platform.
The Agent Cloud Market
Daytona's $1 billion valuation is consistent — but not isolated — in the current landscape. It sits in the top 8% of US enterprise software Series B rounds, but the sector as a whole is attracting enormous capital.
Modal, a direct competitor, raised $355 million at a $4.65 billion valuation in May 2026 and, by October 2026, had already climbed to $750 million at a $15.75 billion valuation, with a run rate above $1 billion. Its Sandboxes product alone generates $300 million per year. Sail Research emerged from stealth in June 2026 with $80 million at a $450 million valuation, led by Kleiner Perkins with Sequoia participation.
The comparison between Daytona and Modal reveals two different approaches to the same problem. Modal grew rapidly as a platform for AI inference and model deployment, expanding into sandboxes as a complementary product. Daytona, by contrast, was built specifically as execution infrastructure for agents — which, according to Burazin, made a difference from the system design: Daytona can configure a sandbox in 60 milliseconds by pre-loading snapshots onto server drives, while Modal takes several seconds to boot a machine.
Sail Research, meanwhile, bets on cost efficiency: its Sailboxes are designed for long-horizon agents, priced over 70% lower than competitors, but with higher p99 latency, trading performance for economy. The trade-off is deliberate: agents working on multi-hour tasks won't notice the difference, but the savings accumulate at scale.
What to Watch
The question isn't whether an "agent cloud" will exist — it already does, and billions of dollars are already invested in it. The question is which of these approaches will dominate. Daytona bets on speed and platform flexibility (multiple OSes, GPUs, full sandboxes). Modal scales with volume and per-capita revenue already far higher. Sail Research pursues cost efficiency for long-duration workloads.
The timing matters: most of these AI agents that will need this infrastructure are still internal products or niche. If production adoption scales as investors expect, these companies are at the forefront. If the forecast proves overly optimistic, the agent cloud market could contract — and Daytona's (and Modal's and Sail's) investors will need to convince the market that the bet is worth the risk. What we're observing today is less a race for infrastructure and more a race to define the standard — and in platform markets, the standard that wins tends to stick.
Who knows: two years from now, when an automatic agent schedules a meeting, writes code, debugs a data pipeline and optimizes a CUDA kernel without human intervention, will it matter whether those virtual machines ran on Daytona, Modal or Sail? Probably not. But the fact that someone needs to decide how they're provisioned, scaled and billed is already defining the architecture of the next generation of autonomous systems — and Daytona is competing in that space with $91 million in the bank and investors who are, literally, the founders of the world's most valuable software companies.
Sources: TechCrunch, Fortune, Dealroom
✓ Independent sources cross-checked and verified before publishing
