Signals Inbox·July 19, 2026·AI Infrastructure

Is Oxylabs really worth $3.6B today?

Oxylabs’ $3.6 billion valuation is aggressive but defensible: the company combines extraordinary growth and profitability with real AI demand, although its opaque group revenue, strong competition from Bright Data and rising restrictions on web scraping leave little room for disappointment.

Q1What happened to Oxylabs’ valuation?

Oxylabs became a $3.6 billion company in one transaction. Warburg Pincus invested $130 million on July 9, 2026, ending an 11-year period during which Oxylabs had taken no outside capital.

There was no previous funding valuation, so Oxylabs did not suddenly jump from $1 billion to $3.6 billion. The company simply received its first external price after building a large and profitable business privately.

The deal still leaves one important unknown. Warburg invested through a fund that can finance growth, acquisitions, balance-sheet changes or shareholder liquidity. Oxylabs never disclosed how much of the $130 million entered the company and how much, if any, went to existing shareholders.

Oxylabs reached the valuation through operating growth and two acquisitions rather than repeated fundraising. Webshare brought more than 10,000 active customers in 2022, while ScrapingBee expanded the developer-facing API business in 2025.

Oxylabs’ valuation and funding history

Q2Is Oxylabs really doing $350M in ARR now?

Probably, although Oxylabs has provided surprisingly little detail behind the $350 million ARR claim. The figure appeared in the company’s financing announcement alongside a claim that the platform serves more than 350,000 technology teams.

Warburg Pincus confirmed the valuation, investment and customer scale in its own announcement, but did not repeat the ARR figure. Oxylabs has published no ARR history, retention rate, organic growth rate or breakdown between fixed subscriptions and usage-based spending.

Usage-based revenue can be highly recurring without being contractually locked in. It can also produce an inflated annualized figure when a recent month benefited from unusually heavy consumption.

We still use management’s number as our base case. Warburg had access to information unavailable to the public, and attaching an invented run rate to such a visible transaction would carry obvious reputational risk. The figure looks credible. Its quality is simply harder to judge than audited revenue.

Q3Why is Oxylabs’ filed revenue so much lower?

The revenue gap is the biggest weakness in the valuation case. According to data submitted to Lithuania’s Centre of Registers, Oxylabs UAB generated €139.5 million of sales in 2025. The group subsequently announced a run rate more than twice as large after currency conversion.

Some of the difference has a straightforward explanation. The filing covers one Lithuanian company, whereas the headline figure covers the entire group. Webshare, ScrapingBee and other international entities may sit outside the UAB accounts. ARR also reflects current spending, while annual revenue includes slower periods from earlier in the year.

Growth could explain another part. If the business continued expanding rapidly after the end of 2025, the latest monthly run rate would sit well above the previous year’s average.

Still, a gap of this size deserves an actual reconciliation. Oxylabs has not disclosed how much revenue comes from each entity, how acquisitions are consolidated or how the ARR calculation treats variable usage. We can defend the group figure, but we cannot independently rebuild it.

Q4Is 10x revenue expensive for Oxylabs today?

At 10.3 times announced ARR, Oxylabs looks expensive but reasonably priced. At roughly 22 times the latest filed revenue, it looks too expensive.

The first multiple can work for a company combining rapid growth, high profit margins and strong exposure to AI infrastructure. The second requires several more years of exceptional execution and assumes that the Lithuanian filing represents only a minority of the real group business.

The valuation therefore comes down to one choice. Investors who trust the consolidated run rate can defend $3.6 billion. Investors who rely mainly on the filed accounts cannot.

Warburg saw the consolidated numbers before investing, so the first interpretation deserves more weight. Public readers still have to accept a large information gap that Warburg did not face.

Q5Is Oxylabs still growing this fast?

Oxylabs’ filed business is currently compounding at an exceptional speed. Revenue rose from €44.9 million in 2023 to €80.9 million in 2024 and €139.5 million in 2025.

That represents 80% growth followed by 72% growth. Across the two years, revenue more than tripled, producing a compound annual growth rate of about 76%.

The longer trajectory is nearly as impressive. Revenue has increased 8.3 times since 2021, equivalent to roughly 70% annual growth over four years. Few businesses maintain that pace while remaining profitable.

The pattern is useful too. Growth slowed to 19% in 2023 before accelerating sharply over the following two years. Oxylabs was already an established proxy provider, so this was not an early-stage company growing from almost nothing. The acceleration arrived alongside the wider shift toward live web data, scraping APIs and AI-related demand.

We do not know whether the whole group is still expanding above 70%. Even a slowdown toward 30% would leave Oxylabs growing much faster than most listed infrastructure companies.

Q6Did acquisitions inflate Oxylabs’ growth?

Acquisitions helped, but they cannot explain most of Oxylabs’ recent growth. The filed numbers are unusually clear on this point.

ScrapingBee reportedly had about $5 million in ARR when Oxylabs acquired it in 2025. During that same year, Oxylabs UAB added €58.6 million of revenue. Even if we unrealistically attributed all of ScrapingBee’s annual revenue to the Lithuanian entity, it would explain less than 9% of the increase.

The acquisition also closed during the year, meaning its actual recognized contribution was probably smaller. ScrapingBee was strategically useful because it brought more than 2,500 developers and an efficient API product, but it was too small to create the financial acceleration by itself.

Webshare is harder to measure because its revenue was never disclosed. Oxylabs acquired it in 2022, however, well before revenue accelerated in 2024 and 2025. A business purchased two years earlier cannot fully explain a sudden second acceleration unless it also grew extremely quickly after the deal.

The group run rate includes acquired revenue, so we cannot call the growth entirely organic. The available numbers still suggest Oxylabs built most of its recent expansion inside the existing business.

Q7Is Oxylabs actually profitable today?

Oxylabs is exceptionally profitable. Its Lithuanian entity reported €66.4 million of pre-tax profit and €62.7 million of net profit in 2025.

That gives the company a 47.6% pre-tax margin and a 44.9% net margin. Those numbers would be strong for a mature software company growing 10%. Oxylabs produced them while revenue increased 72%.

The improvement happened quickly. The pre-tax margin fell to 15.9% in 2023, recovered to 35.8% in 2024 and reached 47.6% in 2025. Revenue tripled during those two years, while pre-tax profit increased more than ninefold.

Group margins may be lower because the filing does not capture every subsidiary and acquisition. Even cutting the reported profitability by one-third would leave Oxylabs with economics that most high-growth infrastructure companies cannot match.

Profitability is currently the best argument for the valuation. Oxylabs does not need investors to finance routine losses, so the new capital can go toward products, network expansion and acquisitions.

Q8How does Oxylabs make so much money with fewer than 300 employees?

Oxylabs has built an unusually efficient operation. Lithuanian employment records currently show about 285 employees, while the company’s financial data works out to approximately €548,000 of annual sales and €246,000 of net profit per employee.

The direction matters more than the exact ratio. Sales per employee increased by 54% in 2025, while profit per employee more than doubled. Oxylabs generated substantially more revenue without matching that increase with equivalent hiring.

Its balance sheet also looks relatively light. Non-current assets rose from €3.8 million to €22.3 million in 2025, yet remained small compared with annual revenue. The company operates a global data network without carrying the infrastructure base we would expect from a hardware-heavy provider.

This efficiency helps explain how Oxylabs remained bootstrapped for so long. Customer spending appears to scale faster than payroll and fixed assets, allowing the business to fund product development and acquisitions internally.

It also makes the valuation less dependent on perfect revenue comparables. A company converting almost half of filed revenue into profit deserves a much higher multiple than an infrastructure provider producing similar growth with heavy losses.

Q9Does Oxylabs look expensive next to public companies?

Oxylabs currently sits below the most highly valued AI infrastructure platforms and well above mature software companies. That is roughly where it belongs.

Based on the latest market values and reported financial results, Cloudflare trades near 38 times annualized quarterly revenue and Snowflake near 17 times. Both provide broader platforms and disclose far more information about customer expansion, backlog and retention.

Elastic, Akamai and Similarweb sit between roughly two and four times revenue. Their growth is much slower, ranging from about 6% to 17%, although Akamai’s smaller cloud infrastructure unit recently grew by 40%.

Oxylabs deserves a premium to that slower group if consolidated growth remains above 30%. It would struggle to justify Cloudflare’s multiple because Cloudflare controls a much wider part of internet infrastructure and has stronger platform-level distribution.

Oxylabs compared with public internet and data platforms

Q10Is Bright Data already ahead of Oxylabs?

Bright Data currently looks at least as strong as Oxylabs on the public numbers. It reported more than $300 million of annualized revenue in late 2025, growth above 50% and high profitability.

The company expected to reach $400 million by the middle of 2026. There is no fresh confirmation that it reached that target, but the earlier figure already placed it close to Oxylabs in scale.

Bright Data also claims to serve 14 of the 20 largest LLM laboratories, seven of the ten largest AI-first companies and more than 100 million AI-agent interactions each day. Its advertised residential network contains more than 400 million monthly IPs, compared with more than 175 million at Oxylabs.

Network figures can use different definitions, so they should not be treated like audited market share. The broader comparison remains pretty clear: Bright Data matches Oxylabs across revenue scale, growth, profitability, products and AI customer exposure.

Oxylabs therefore deserves no monopoly premium today. The market appears to have two major leaders, followed by cheaper providers such as Decodo, SOAX and NetNut and newer API companies attacking individual parts of the stack.

Q11Do recent AI deals make Oxylabs look cheap?

Oxylabs looks sensibly priced beside the most aggressive recent AI data deals. It looks less cheap beside companies with similarly proven revenue.

AlphaSense raised at a $7.5 billion valuation after passing $600 million in ARR, giving it a multiple of about 12.5 times. Databricks reached $134 billion with more than $4.8 billion of revenue run rate, close to 28 times.

Exa and Parallel reached valuations of $2.2 billion and $2 billion without disclosing comparable revenue. Their prices rely heavily on the idea that AI agents will require a new search layer built specifically for machines.

Oxylabs receives a lower multiple than AlphaSense and a fraction of Databricks’ multiple. The discount makes sense. AlphaSense is deeply embedded in finance and research workflows, while Databricks controls a much larger share of enterprise data infrastructure.

Oxylabs has something those younger search startups do not: hundreds of millions in claimed recurring revenue, filed profits and a decade of operating history. Compared with Exa and Parallel, its valuation looks conservative. Compared with mature public infrastructure companies, it remains demanding.

Q12Is Oxylabs really an AI company now?

Oxylabs currently looks more like an established web-data company benefiting from AI than a newly created AI software company. The valuation partly depends on investors giving old infrastructure a new AI multiple.

The company began as a premium proxy provider. Its website still leads with residential proxies, datacenter proxies and scraping APIs, products that existed well before the current agent boom.

The product direction has clearly changed. Oxylabs launched OxyCopilot in 2024, followed by AI Studio, an AI Crawler and a Browser Agent in 2025. These tools let customers describe scraping and browser tasks in natural language rather than building every workflow manually.

Warburg’s announcement describes the same evolution: Oxylabs is moving from premium proxies toward a full-stack platform that helps AI systems access and interpret the live web.

Oxylabs has never disclosed how much revenue comes from AI companies or its newer AI-branded products. Given their recent launch dates, most current revenue probably still comes from the underlying proxy and scraping infrastructure.

The real bet is broader than the success of AI Studio. Investors expect AI agents to increase consumption of the infrastructure Oxylabs already sells. That thesis is much more credible than pretending a few new AI products suddenly created a multibillion-dollar company.

Q13Does Oxylabs really have 350,000 customers?

The 350,000 figure probably measures teams, accounts and developers rather than 350,000 conventional paying companies. Dividing the disclosed revenue base by that total produces average annual revenue of roughly $1,000 per account.

The source wording confirms the ambiguity. Warburg describes more than 350,000 customers, Oxylabs calls them technology teams, while the company’s impact reporting previously referred to reliably serving more than 15,000 customers.

The figures may cover different populations. A developer can register a team without becoming a meaningful customer, while one enterprise contract may support hundreds of users.

Broad reach still has value. Webshare and ScrapingBee give Oxylabs access to thousands of smaller developers who may later need higher-volume enterprise products.

For valuation, the missing data is more important: customers spending above $100,000, customer concentration, churn and net revenue retention. Snowflake reports 779 customers producing more than $1 million each. Similarweb reports its large-customer count and retention every quarter. Oxylabs reports neither.

Q14Are AI agents actually creating real demand for Oxylabs?

Yes. Recent spending shows that live web data is moving from experimental AI budgets into large and sometimes multi-year contracts.

Bright Data passed a $300 million annualized run rate while growing above 50%. That gives us evidence from Oxylabs’ closest private competitor.

Similarweb then announced two multi-year enterprise contracts worth about $47 million over three years. Both carried seven-figure ARR commitments, involved AI-driven companies or large enterprises and pushed Similarweb above $300 million in ARR. The combined contract value represents roughly $15.7 million per year.

Akamai provided an even larger infrastructure example in its latest results. A frontier-model provider committed $1.8 billion over seven years to its cloud infrastructure services, equivalent to about $257 million annually.

These businesses operate at different layers: web collection, proprietary digital data and cloud capacity. All three are seeing large AI-related spending. That pattern carries more weight than one provider launching a new product or one startup raising money.

Oxylabs has not revealed its own AI revenue or major AI contracts. Category demand is real. The remaining question is how much of it Oxylabs is capturing.

Q15What does Oxylabs have that rivals cannot easily copy?

Oxylabs has a real moat built from infrastructure, operational knowledge and distribution accumulated over a decade. A new company can launch a scraping API quickly. Recreating the whole system would take years.

The company manages more than 175 million residential IPs and two million datacenter IPs across hundreds of locations. It also maintains the routing, fraud controls, sourcing relationships and website-specific logic required to keep those connections useful.

Its products now cover raw proxy access, automated scraping, browser automation, pre-collected datasets and natural-language data workflows. Customers can buy basic connectivity or outsource most of the collection process.

Oxylabs reached 150 patents in 2025, up from 100 one year earlier. The raw number does not prove every patent is commercially valuable, though a 50% expansion in one year shows sustained technical investment.

Webshare and ScrapingBee add another layer. They let Oxylabs reach self-service customers and developers without forcing every account through an enterprise sales process.

Bright Data has built much of the same machinery and may have greater network scale. Oxylabs’ moat protects it from new entrants far more effectively than from its closest rival.

Q16Can Cloudflare and publishers squeeze Oxylabs now?

Yes, and the change is already happening. Cloudflare now gives website owners more control over whether AI crawlers can access content, what purpose they declare and whether they must pay.

Pay Per Crawl lets participating publishers set prices for crawler access. Cloudflare is also pushing AI companies to separate bots used for search, model training and agent activity. Crawlers that fail to identify their purpose may face default blocking across participating sites.

This changes the economics of web access. Oxylabs may have to pay publishers, pass higher costs to customers or negotiate licenses for important sources. Any of those routes can reduce margins or make some datasets less attractive.

The pressure may eliminate weaker competitors too. Large customers will need providers capable of handling permissions, payments, provenance and changing rules across thousands of websites. Oxylabs has the scale and compliance teams to play that role.

Its future position depends on what customers are paying for. A trusted access and licensing layer could become more valuable as the web closes. A business relying mainly on cheaper ways around technical restrictions has a much harder future.

Q17Could Reddit’s lawsuit seriously hurt Oxylabs?

The Reddit lawsuit could hurt Oxylabs if it creates a playbook that other platforms follow. Reddit sued Oxylabs, Perplexity, SerpApi and AWMProxy in October 2025, alleging that the companies circumvented technical protections to collect and resell Reddit content.

Oxylabs rejects the allegations, and no final judgment has been reported. The immediate financial impact remains unknown.

The wider pattern matters more than this individual case. Reddit has signed data-licensing agreements with companies such as Google and OpenAI. It now has a direct financial reason to challenge third parties accessing similar information without paying.

European rules are also becoming more specific. The European Data Protection Board’s latest guidelines cover companies scraping personal data themselves and those hiring third parties to do it. They focus on lawful basis, data minimization, sensitive information and accountability.

Oxylabs can absorb higher legal and compliance costs better than a small scraper. Repeated lawsuits and licensing requirements could still reduce the supply of cheap, valuable data across the whole industry.

Q18How much revenue does Oxylabs need to grow into $3.6B?

Oxylabs already supports the valuation at a ten-times multiple. It needs to almost double revenue if the market eventually values it like a mature infrastructure company at five times.

At 30% annual growth, the disclosed revenue base would reach about $592 million after two years. At 40%, it would reach roughly $686 million. Either path would bring the valuation close to five or six times revenue.

Those targets look achievable beside the recent filed growth. They become much harder if consolidated growth has already slowed toward 15%.

Revenue required to support Oxylabs’ $3.6 billion valuation

Q19What has to go right for Oxylabs to look cheap?

Oxylabs will look cheap if group growth stays above 30%, consolidated margins remain above 25% and AI agents keep increasing the number of web requests made by machines.

Thirty percent growth for two years would take revenue close to $600 million. A six-times multiple would then support nearly the entire current valuation, while anything above that would create upside.

The company also needs its developer distribution to feed larger enterprise accounts. Webshare and ScrapingBee can acquire smaller users, while Oxylabs sells the successful ones premium proxies, automated scraping, browser agents and managed infrastructure.

Legal changes must favor scale rather than destroy margins. If publishers increasingly demand permission and payment, Oxylabs needs to become the intermediary managing those relationships.

None of these conditions requires another 70% growth year. The bull case works with a controlled slowdown, provided the revenue is genuinely recurring and the margins survive.

Q20What would break Oxylabs’ valuation?

A slowdown below 20% would quickly expose the price. At 15% annual growth, revenue would reach roughly $463 million after two years. Applying a four-times mature infrastructure multiple would produce a valuation near $1.85 billion.

Weak ARR quality would cause the same problem. Heavy usage can disappear faster than contracted subscriptions, particularly if a few AI companies account for a large share of consumption.

Bright Data could also take more of the AI opportunity. It already reports similar scale, fast growth and strong relationships with leading model developers.

Publisher fees, stricter crawler controls and adverse court rulings would pressure both costs and accessible supply. Oxylabs could continue growing while becoming materially less profitable.

Warburg is paying for Oxylabs to remain exceptional. Merely becoming a decent, normally growing infrastructure company would probably cut the valuation by a third or more.

Q21So is Oxylabs really worth $3.6B today?

Yes, but $3.6 billion sits near the top of the defensible range.

At the disclosed group run rate, investors are paying about 10.3 times revenue for a business with extraordinary filed growth, high profitability and strong exposure to a real AI infrastructure spending wave. That price remains below premium companies such as Snowflake, AlphaSense and Databricks.

The filed accounts make the case unusually strong. Revenue tripled in two years, profit grew even faster, acquisitions explain only a small part of the latest increase and the Lithuanian operation produces remarkable revenue and profit per employee.

Three issues stop us from calling Oxylabs cheap. The company has not reconciled group ARR with filed revenue, Bright Data appears at least as strong, and publishers are gaining more power to restrict or charge for automated access.

Our verdict is aggressive but justified. We would accept the $3.6 billion valuation, but we would not pay much more today. The current price already assumes consolidated growth stays near 30%, margins remain excellent and Oxylabs becomes a central access layer for AI agents rather than a premium proxy provider facing higher content costs.