Inside Vilnius's Cyber City campus on the grounds of a former sock factory, Source has joined Lithuania's unicorn club as number six. The group debuted this week as the umbrella over Oxylabs and four other web-intelligence businesses, and it kept Oxylabs as a standalone brand. CEO Vytautas Savickas, who stepped into the role this year, said a summer deal with Warburg Pincus pegged Source's value at $3.6 billion, placing it behind only Vinted among the country's private tech companies.
The neighborhood packs in local heavy hitters. Next door is Nord Security, with Vinted located across the street. Even the nearby bus stop nods to the scene, carrying the name Vienaragių, which means unicorn in Lithuanian. The single city block has become shorthand for Lithuania's evolution since the 1990s from low-cost manufacturing to a tech hub where digital services are taking a larger share of exports.
Source grew out of Oxylabs, founded in 2015, which helps Fortune 500 firms in e-commerce, travel and cybersecurity gather and process public web data at scale. Today the company counts about 350,000 tech teams globally, from five-person startups to 10,000-employee giants. It is running at a $350 million annual recurring revenue pace, with roughly half of that coming from the US. To stay close to priority customers such as tech companies, leading retailers and AI labs, Source will open a San Francisco office.
AI model builders are a big slice of that demand. "Basically, eight out of 10 biggest LLM labs leverage our infrastructure to collect that data real time," Savickas said, while declining to identify them because of non-disclosure agreements.
Unicorn valuations in smaller markets signal where talent is concentrating. Market Briefs covers that growth free every morning.
As AI's appetite grows, Source has been racing to keep up. Savickas said the company boosted data-processing capacity by about tenfold in the first six months of this year, letting it work through "petabytes and petabytes" of data daily. That scale requires constant resource juggling.
"You always feel behind," he said. "It brings a lot of anxiety." To handle the engineering load, Source added hundreds of hires and today has 800 employees worldwide, of whom more than 350 are developers and engineers. Its closest rival is Israel's Bright Data, with both among the largest providers of web-data infrastructure.
Next on the roadmap is a web API purpose-built for AI agents. It will merge the company's various data-collection tools into one service, backed by Source's own index of the internet, so agents can quickly find and pull fresh public information. The product is in early beta, slated to launch within weeks, and was built entirely in-house by roughly 100 engineers and developers. As Savickas put it, "They need the freshest data to ground their answers, otherwise they can hallucinate."
There is a catch. AI-driven scraping has ignited criticism and lawsuits over uncompensated use and copyright. Savickas frames Source as backing an open web, while also arguing that companies have the right to safeguard private data and intellectual property.
He says Source follows compliance and ethics rules to respect those lines. In his view, restricting access to public information would tilt the field toward the biggest platforms and away from the many businesses building products, services and AI tools. "Big tech, big players want to monopolize" data, he said.
That would be "benefiting the core big tech, but then not the whole economy."
For your wallet, this is the picks-and-shovels story inside AI's boom: a private infrastructure player scaling like crazy, winning usage-based revenue and shipping tools aimed at the next wave of AI agents. If public data access tightens, growth could whipsaw. If it stays open, the companies that move bits efficiently may keep compounding.
Data infrastructure is quietly one of the better businesses in tech. Get the free Market Briefs daily newsletter and follow it.
