Oxylabs takes first outside round: $130M at $3.6B from Warburg

A Lithuanian proxy shop founded in 2015 just became the highest-priced pure-play in a category most AI investors barely think about, and that says something about where agentic AI is actually leaning. Oxylabs UAB has raised $130 million from Warburg Pincus, its first outside capital since it was founded, at a $3.6 billion valuation, SiliconANGLE reports. The check comes out of Warburg's $4 billion Capital Solutions Founders Fund.

What makes the profile unusual for a first outside round is the reported scale underneath the raise. Oxylabs says it already runs at more than $350 million in annual recurring revenue, serves over 350,000 customers globally, and routes billions of requests a day through a pool of more than 175 million consumer-device IPs. That is not a company taking growth capital to find product-market fit. It is a company taking growth capital because a private equity firm decided the pipe it owns is strategic.

The pitch CEO Vytautas Savickas is making is worth taking seriously even if you discount his numbers. "The next generation of AI won't be powered by static indexes," he told the outlet, arguing that agents need "a live feed of what is out there, not a stale index." If that framing is right, the interesting infrastructure question for the next couple of years is not who owns the biggest index, it is who can reliably fetch the live page when an agent asks for it, at scale, without getting blocked.

The honest caveat is that this is also a legally exposed business. Reddit sued Oxylabs, Perplexity AI, and other scraping providers last year, comparing them to "bank robbers" who will do almost anything to reach the data except pay for it. What the reporting doesn't give you is the dilution attached to Warburg's check, Oxylabs' profitability or growth rate, or how the "ethically sourced" consumer IPs are actually obtained, which is a live compliance question and probably the most important one still hanging over the category.

Watch this less as a fundraising story and more as a signal. If the agentic AI thesis holds, the picks-and-shovels layer includes not just GPUs and vector databases but the contested plumbing that turns the open web into a queryable surface for agents. Oxylabs is now the priced comp for that layer, and the buyers most likely to benefit are the AI labs and agent startups that would rather rent the pipe than build and defend it themselves.