The global race to build infrastructure for autonomous AI systems is intensifying, and there is new unicorn joining the race.
Croatian Daytona announced today that it has raised a $60 million (€56 million) Series B venture capital round. This fresh injection of capital values the enterprise at $1 billion, officially making it the third unicorn to emerge from the Croatian tech ecosystem, joining the ranks of Infobip and Rimac Automobili.
FirstMark doubles down on the agentic cloud
New York-based FirstMark led the latest round, just as it spearheaded Daytona's $24 million (€22 million) Series A earlier this year in February. The venture firm is clearly convinced by the company's trajectory, as well as the accompanying investors are – approximately $50 million (€46.5 million) of the new funding coming from existing ones. This brings the total amount Daytona has raised to date to around €85 million.
A roster of prominent tech founders also joined the cap table, including Stripe's Patrick Collison, Shopify's Tobi Lütke, Perplexity's Aravind Srinivas, Quora's Adam D’Angelo, and Zoom's Eric Yuan, alongside strategic backers like Qualcomm Ventures, Okta Ventures and Samsung Next.
The core problem Daytona solves revolves around the need for autonomous AI agents to have their own secure, rapid computing environments. According to the company, traditional cloud infrastructure from giants like AWS or Microsoft Azure can take up to several minutes to spin up virtual machines. In contrast, Daytona claims it can configure a new sandbox for an AI agent in just 60 milliseconds. This speed allows agents to launch, fork into parallel branches, and safely execute code at a scale that human developers never required.
"The biggest change is the shift from Chatbot AI to Agentic AI," Burazin explained to The Recursive in February. "While earlier models just generated text, modern AI agents now attempt to perform tasks. However, this is where we encounter a key obstacle: for an agent to truly work as a digital expert, they need a 'workspace': a computer, an operating system, and a secure environment."
A computer for every agent
Founded in 2023 by Croatian IT veterans Ivan Burazin, Goran Draganić, and Vedran Jukić, the startup initially focused on automating software development environments. Daytona started as an open-source development environment manager, attempting to solve a persistent problem where developers lose a massive chunk of their productive time to inefficient environments. However, the shift to agent infrastructure came as generative AI evolved from chatbots to autonomous systems attempting real tasks.
"We basically gave our competitors all our customers, rebuilt everything, had like three months left of revenue in the bank, and then relaunched this thing," Ivan Burazin, CEO of Daytona, told Fortune.
The pivot paid off remarkably fast. Daytona's platform now serves a string of high-profile AI companies, counting Anthropic, OpenAI, Meta, and Nvidia among its customers. Between January and September of this year, the company's revenue run rate grew 25-fold. "We've been growing on average every single month 50.9%," Burazin notes.
This explosive growth relies on providing what the firm calls programmatic, composable computers, or sandboxes. These are virtual machines where CPU, memory, and operating systems are allocated on demand, giving AI models the space to practice tasks repeatedly during training and testing. It also ensures safety, minimising the chance that an agent will run amok or that an outside party could hijack sensitive data from the host machine.
Lessons from a 15-year journey
Burazin's mission to redefine developer productivity has been years in the making. Daytona is partially based on the work he did with his previous startup, Codeanywhere, a cloud-based development environment. His subsequent roles, including managing Croatian unicorn Infobip's efforts to establish itself as a dev-first company and running the Shift conference, taught him crucial lessons about selling to both end-users and enterprises.
The company's rapid funding history reflects this deep industry trust. Before the massive Series B, Daytona closed a €1.8 million pre-seed round in November 2023, followed by a €4.6 million seed led by Upfront Ventures in June 2024. The subsequent Series A brought their total raised to €28.8 million before this latest round.
While Daytona operates from its New York headquarters and serves key customers in San Francisco, its core engineering power remains in Europe. The company runs its 20-person technical team primarily from offices in Split and Zagreb, Croatia, maintaining strong ties to the local tech talent pool as it scales its operations globally.
The impending CPU crunch
While AI models traditionally run on GPUs to handle complex computations, Daytona's approach highlights a different requirement: to do actual work, agents need CPUs. The company distinguishes its offering from inference clouds, pointing out that while models think on GPUs, agents work on CPUs.
Burazin predicts this dynamic will create a significant shift in hardware demand. The CEO believes that the amount of money spent on CPUs could soon outpace GPU spending, which seems counterintuitive in today's market. He warns that the sheer volume of CPU machines needed to sustain this growth trajectory does not currently exist, signalling a potential hardware bottleneck.
Daytona plans to deploy the new funds to expand its go-to-market sales team and hire more engineers. As AI agents increasingly transition from experimental concepts into daily production tools, the demand for specialised infrastructure will only grow. The real test for Daytona will be scaling its operations fast enough to meet a potential CPU shortage, all while maintaining its rapid 60-millisecond execution times.