Daytona sells computers to software. Not inference, not dashboards: the execution layer underneath AI agents, where an agent gets its own sandboxed machine to write code, run it, browse and fail safely. The company says an agent's computer has almost nothing in common with a cloud instance. It has to start in milliseconds, pause with its state intact, fork into parallel copies and disappear when the job ends. TechCrunch reported on October 23 that investors had priced that idea at a reported $1 billion: a $60 million Series B led by FirstMark Growth, which also led the $24 million Series A in February, taking total funding to $91 million. Daytona's own announcement post is dated October 20.
The investor list is part of the pitch, and it is a strong one. The angels run companies that operate agents at scale: Patrick Collison of Stripe, Tobi Lütke (Shopify), Aravind Srinivas (Perplexity), Eric Yuan (Zoom), Olivier Pomel (Datadog), Adam D'Angelo (Quora) and Brendan Foody (Mercor). Qualcomm Ventures, Samsung Next and Okta Ventures joined as strategic investors.
The growth, in multiples
Daytona's announcement gives two growth figures. Revenue has grown approximately 25 times since January. Average month-over-month net revenue retention is approximately 135%. Both are the company's own numbers, unaudited, and neither comes with the figure it is a multiple of. Twenty-five times a small January base and twenty-five times a large one describe entirely different companies, and the announcement does not say which one this is.
The retention figure deserves a slower read, because Daytona frames it against a famous benchmark. Daytona compares its approximately 135% average monthly net revenue retention with Snowflake's roughly 160% annual retention at IPO. The company explicitly identifies the different time periods, but placing them together still invites a misleading comparison. A monthly average and an annual cohort measure cannot be ranked directly without consistent definitions and time horizons.
The same caution applies inside Daytona's own numbers. If a single cohort retained 135% of its revenue in each of six consecutive months, its spending would grow about sixfold. Daytona, however, reports an average monthly retention rate and separately says its customers from six months ago spend nearly three times as much today. Those are not necessarily equivalent measures. Without cohort definitions, monthly rates and underlying revenue figures, readers cannot reconstruct either claim or directly compare them.
None of this makes the growth invented. The named customers are checkable, and one detail has the texture of an operational fact rather than a slogan. Daytona says Writer moved to its platform after operating more than 20,000 sandboxes a day internally, and subsequently provisioned 300,000 sandboxes in a month through Daytona. The company provides no independently audited usage data for that example.
What the round actually prices
Strip the multiples away and the concrete assets are a customer list that spans the agent stack (Meta and Cohere building models, Handshake and Surge AI building training data, Writer and n8n shipping agent products), a lead investor that has now led two consecutive rounds, and a product category that every serious agent deployment needs in some form. Agents that act on real systems need somewhere isolated to act, and the recent run of agent incidents has made the cost of a bad sandbox concrete rather than theoretical.
What readers cannot test the $1 billion against is revenue. The figure is a reported financing valuation: the investors who set it may have reviewed private financial data that readers cannot see. Daytona has disclosed neither a revenue figure nor any profitability measure, so outside readers cannot assess the valuation against either. For readers, multiples without a base are a shape, not a measurement.
Sources
- [1] Daytona, "Daytona Raises $60M Series B to Build the Agent Cloud" (announcement post dated 20 October 2026) - primary source for the round terms and $91M total, the investor and customer lists, the product claims (the only provider with Linux, Windows, macOS and GPU sandboxes on one platform; computers that start in milliseconds, pause with state intact and fork), the approximately 25x revenue growth and approximately 135% average monthly net revenue retention figures, the nearly 3x six-month cohort claim, the Snowflake comparison in Daytona's own words ("Snowflake's famous NRR at IPO was around 160% - per year. Daytona is averaging close to that every month"), and the Writer figures (20,000 sandboxes a day in-house; 300,000 provisioned in a single month on Daytona); read first-hand 23 October 2026Read source
- [2] TechCrunch, Julie Bort, "Agent cloud startup Daytona hits $1B valuation backed by a who's who of angels" (23 October 2026, 09:26 PDT) - source for the 23 October reporting of the $60M Series B at a $1 billion valuation led by FirstMark Growth, the $91M total, FirstMark's $24M Series A lead in February, the New York base and the company's three-year age; read first-hand 23 October 2026Read source