On July 9, 2026, Oxylabs — one of the world's largest providers of proxies and web data — announced it had raised $130 million from Warburg Pincus. The deal values the company at $3.6 billion and marks its first external investment in 11 years of existence. For a market that was recently associated with "gray" tools and semi-anonymous traffic sellers, this is a watershed moment: institutional capital has officially recognized proxies as infrastructure for the AI era. Let's explore what this means and how it changes the landscape for proxy buyers.
What Happened
The Lithuanian company Oxylabs, founded in 2015 and previously developed solely through its own funds (bootstrapped), agreed to accept $130 million from Warburg Pincus Capital Solutions Founders Fund. The post-money valuation reached $3.6 billion — making Oxylabs the second most valuable private company in Lithuania's history and the second "unicorn" to emerge from the Tesonet accelerator.
The figures disclosed by the company alongside the deal explain the interest from large capital:
- $350 million in annual recurring revenue (ARR) — a level comparable to mature SaaS companies;
- over 350,000 technical teams as clients worldwide;
- more than 160 patents in web data collection and delivery;
- two acquisitions in recent years — Webshare (2022) and ScrapingBee (2025).
The rhetoric surrounding the deal is telling. CEO Vytautas Savickas frames the mission not as "selling proxies," but as building a foundation for machines: "As AI agents begin to navigate the web far more actively than humans ever did, the future belongs to the data infrastructure that grounds these systems in real, continuous knowledge." From the investor's side, Allison Ross, principal at Warburg Pincus, emphasizes the "complex, resilient, and compliant technology" — the term "compliance" is not accidental, and we will return to it.
Why Now
Just three or four years ago, proxy providers primarily sold bypassing bans and price scraping services. Today, they are being repackaged as "data infrastructure for AI" — and funding is flowing precisely under this thesis. The reason is simple: large language models and autonomous AI agents have developed an insatiable appetite for fresh web data, while traditional access channels are collapsing.
Official APIs of social networks and platforms have become either expensive or closed over the past year and a half: X has eliminated free read access (the basic rate is $200/month), Reddit charges about $0.24 for 1,000 calls, and the Instagram Basic Display API was disabled back in late 2024. When the legal "front door" to data closes, demand shifts to proxies and scraping infrastructure — the only remaining way to collect public data at scale.
The market scale confirms the trend. Analysts estimate that the web scraping market grew from approximately $1.03 billion in 2025 to $1.17 billion in 2026 and, with an annual growth rate of about 13-18%, is heading towards $2.2-2.3 billion by the early 2030s. Broader estimates, including AI scraping and data processing, yield significantly larger numbers. It is into this growing stream that private capital is flowing.
The value of public web data is underscored by the legal battles surrounding it. Reddit filed a lawsuit against Perplexity and three companies, accusing the AI startup of collecting data from the platform through intermediaries — the lawsuit mentions SerpApi, Oxylabs, and AWMProxy, and Reddit unfavorably compared Perplexity to a "North Korean hacker" willing to do anything for data for its "answer engine." According to the plaintiff, the defendants accessed nearly 3 billion pages in just two weeks. Perplexity denies the allegations. For us, the outcome of the dispute is less important than the fact: data has become an asset worth fighting over, and the infrastructure for its extraction is valued in the billions.
Oxylabs Is Not Alone: The Industry Is Consolidating
The Oxylabs deal is not an isolated event but part of a broader wave. Direct competitor, Israeli Bright Data, according to CEO Oren Lenchner, has surpassed $300 million ARR and aims to reach $400 million by mid-2026 — a growth of over 50% year-on-year. The company claims to serve 14 of the top 20 global LLM laboratories and processes over 100 million AI agent interactions daily for more than 20,000 clients, including Fortune 500 companies.
Simultaneously, brand mergers and asset acquisitions are underway: Smartproxy rebranded to Decodo in 2026, Oxylabs is acquiring niche players (Webshare, ScrapingBee), and Bright Data is strengthening its position through acquisitions like Market Beyond. The market is clearly stratifying into three price tiers:
- Enterprise — Bright Data, Oxylabs: premium rates for residential traffic;
- Mid-market — Decodo (ex-Smartproxy), SOAX, NetNut: mid-tier pricing;
- Budget — IPRoyal, Webshare, and dozens of smaller players: bottom-end price dumping.
Since 2025, more than 50 new proxy vendors have entered the market, and competition in the lower segment is driving prices to the limit — largely due to services built on dubious traffic sources. And this is where the second, less glamorous side of the story begins.
The Dark Side: The Market Is Splitting in Two
The influx of institutional money and the focus on "compliance" is a reaction not only to demand but also to increasing legal pressure. Just a week before the Oxylabs deal, on July 2-3, 2026, the FBI, in collaboration with Google, dismantled the residential proxy network NetNut, which operated as a botnet of millions of hacked Smart TVs and set-top boxes. We covered this operation in detail in our article on the dismantling of the NetNut proxy botnet — and this contrast explains the logic of Warburg Pincus.
The proxy market is visibly splitting into two poles. On one side is a legitimized infrastructure with external audits, patents, contractual relationships with traffic sources, and a willingness to show the origin of IPs. On the other side are cheap networks of unknown origin, some built on compromised devices, which live until the first visit from law enforcement. Institutional investors are putting $130 million only into the first pole: PE funds do not buy companies that could lose their domains and "millions of IPs" by court order tomorrow.
For buyers, this is a key signal. Cheap residential traffic at suspiciously low prices is almost always a compromise on legality and stability. A network that gets taken down takes all your workflows with it: accounts, sessions, scraping tasks. The origin of IPs has transformed from a technical detail into a selection criterion for providers, on par with price and speed.
What This Means for Proxy Buyers
The Oxylabs deal is good news for the industry as a whole, but it has practical implications that should be considered now.
- Prices in the upper segment will rise, but not dramatically. Institutional capital expects returns, which means pressure on enterprise rate margins will increase. Meanwhile, competition in the mid-market and budget tiers remains fierce — buyers still have options.
- The focus on "AI infrastructure" is shifting products. Major players are increasingly selling not raw traffic but ready-made scraping APIs and "unlockers." This is convenient for some scenarios but removes control in others — for example, in multi-accounting and anti-detect tasks, where a clean managed IP is needed, not someone else's black box.
- Compliance is becoming a competitive advantage. After NetNut, the question "where do your IPs come from" is no longer theoretical. A provider that can clearly answer this question is worth their price.
- Consolidation will narrow choices at the bottom. Some small vendors will be absorbed or exit the market — relying on an anonymous seller from a forum is becoming increasingly risky.
The practical takeaway is simple: choose a provider not based on a single line "price per gigabyte," but on a combination — transparency of traffic source, type of proxy for a specific task, and predictability of the network. For data collection and working with reputation-sensitive platforms, it makes sense to rely on residential proxies with clear origins, while for high-load scraping of open sources, where speed and volume are more critical, data center proxies are preferable. A tool tailored to the task is almost always more cost-effective than an "all-purpose" expensive package.
Conclusion
$130 million from Warburg Pincus and a valuation of $3.6 billion is not just a significant deal for one vendor. It marks a moment when the proxy industry has officially matured: transitioning from a niche of "bypass tools" to an infrastructural layer upon which the entire AI data economy relies. This has two sides. The upper segment of the market gains capital, legitimacy, and a focus on compliance — while simultaneously, the market sharply separates "clean" networks from criminal botnets, which are being dismantled one by one. For proxy buyers, the main lesson of 2026 can be summed up in one phrase: the origin and legality of your IP are now just as important as its price.