Every proxy network on your shortlist now has a parent company

On 9 July 2026 Warburg Pincus published a press release headlined "Oxylabs Announces $130 Million Investment from Warburg Pincus." The release puts the investment at $130 million, values Oxylabs at approximately $3.6 billion, and calls it the company's first outside investment since it was founded in 2015. The capital comes through the Warburg Pincus Capital Solutions Founders Fund and is described as a growth investment. Oxylabs' own announcement adds a company-stated figure of $350 million in annual recurring revenue and more than 350,000 customers.

Neither release says what share of the company changed hands. That omission is worth sitting with, because $130 million against a $3.6 billion valuation is under four percent. Whatever this deal is, it is not a change of control. Private equity has arrived at the top of the proxy market, and it has arrived holding a small stake.

The headline on this post is an overstatement. The rest of it is about how much of one.

The deals underneath the headline

The investment is the visible layer. The more interesting part is the two acquisitions it sits on top of, both listed in Oxylabs' own announcement: Webshare Software in 2022 and ScrapingBee in 2025. Those come from the acquirer's primary announcement rather than from a news roundup, which matters in this category, where most coverage is single-sourced and rarely revisited.

Read the group as a portfolio and the shape is obvious. Oxylabs is the enterprise proxy network. Webshare is the cheap self-serve datacenter pool a solo developer signs up for with a card. ScrapingBee is the managed scraping API that abstracts the network away entirely. One owner now covers three price points and three buying motions, from a $3 hobby plan to a negotiated enterprise contract.

Bright Data, the other incumbent at that tier, has assembled a comparable spread of proxy types and scraper products, which is why we keep a dedicated Bright Data vs Oxylabs breakdown. The difference now is that one of the two has an institutional investor on its cap table and the other, as far as any announcement we can find says, does not.

Elsewhere the deal flow is harder to pin down. Proxyway's news index has covered smaller pools being absorbed by mid-size networks, including acquisitions attributed to SOAX. We could not find matching announcements on SOAX's own site, so treat those as reported rather than established, and do not attach a date or a price to them in a procurement document.

What the pessimist expects

The standard buy-side worry about consolidation runs like this. A network buys a smaller product for its customer list. Within a year the self-serve tier is deprecated, the entry plan gains a minimum commitment, and the product is repriced toward the parent's enterprise motion. Support moves to a shared queue. The roadmap goes quiet.

It is a reasonable model. It is also testable, because both Oxylabs acquisitions are now old enough to test it against. So we read the pricing pages on 27 July 2026.

Webshare, four years after the acquisition, still lists a free tier of 10 datacenter proxies with no card required. Paid datacenter starts at $0.0299 per proxy, shown as $2.99 billed monthly for 100 proxies. Rotating residential starts at $3.50 per GB for a single gigabyte. There is a custom plan behind a contact form, which is normal, and nothing on the page forces a new buyer into it.

ScrapingBee, roughly a year after the acquisition, still offers 1,000 free API credits without a card. The entry Freelance plan is $49 per month for 250,000 credits, rising through Startup at $99, Business at $249 and Business+ at $599, with a custom tier above that. The pricing page carries no Oxylabs branding at all.

Those are the figures the vendor pages showed on 27 July 2026, and proxy pricing moves faster than editorial does, so check them before quoting them at anyone. What they show is that in both cases the deprecation story failed to arrive on the schedule it predicts.

That is worth saying plainly, because the opposite claim circulates as received wisdom. Consolidation in this market has not reliably killed self-serve pricing. It may still change other things.

What actually changes, and how you would notice

The risks that survive the evidence are quieter than a price hike, and most of them never appear on an invoice.

Start with pool identity. A residential proxy pool has a specific IP source and a specific consent mechanism behind it, and it behaves in its own way against your targets. If two pools merge after an acquisition, the addresses behind your endpoint can change while your contract, your price and your dashboard all stay identical. You find out through success rates on one stubborn domain, not through an email. Which proxy type your target actually requires is a per-target question, covered in choosing proxy type by target, and the answer does not survive a silent pool change.

Sourcing diligence is the second one, and it does not transfer. Where consumer IPs come from is now a procurement item rather than a footnote, which we have written about in where your residential IPs come from and in the accompanying sourcing audit guide. When a pool changes hands, the diligence you did on the seller does not automatically carry to the buyer. Proxyway reported on 3 July 2026 that the FBI and Google had disrupted NetNut's residential proxy network, with the provider said to have lost domains and a large number of IPs. We have not verified that against a court filing or a Google security post, so treat it as a reported event rather than a settled one. It still illustrates the shape of the risk: a proxy supplier can lose capacity overnight for reasons that have nothing to do with your account.

Renewal is where the third one shows up. A vendor that owns three products has an incentive to move you off a single line item and onto a platform agreement, and platform agreements are where minimum commitments and annual terms live. You will see it in contract language rather than in a list price.

A renewal checklist

Four questions, asked in writing, before you sign the next term.

Who owns this pool today, and did that change during our current term? Ask for the legal entity, not the brand. If the answer is a group structure, ask which other brands sit inside it. Two subsidiaries of one parent is not dual-sourcing.

Has the IP source or pool composition changed since we onboarded? If the answer is vague, stop asking and re-run your own tests against the same target set you used at purchase.

Is the self-serve tier we are on still on the roadmap for the next twelve months? Ask for a horizon in months. A vendor that intends to keep the tier will say so. A vendor that will not put a number on it has answered.

What is the exit ramp? Notice period, configuration export, whether unused prepaid balance survives cancellation, and how long credentials keep working after termination. If the honest internal answer is three weeks of engineering, that is a switching cost the incumbent can price against you.

None of that is adversarial. It is the same diligence you would apply to any web data provider, and a vendor with nothing planned answers it in a paragraph.

Dual-sourcing is the boring hedge

The structural answer to ownership uncertainty is not to avoid the consolidators. Their pools are large for a reason and they are frequently the best-performing option on a hard target. The answer is to make switching a config change rather than a project.

Run one incumbent and one smaller network behind a single gateway, splitting traffic by target and by cost. The routing, health-checking and cutover logic has to exist before you need it, which is the argument for treating failover across proxy vendors as part of the stack rather than a break-glass script.

For the second source, the directory has options outside the deal flow described above. Evomi markets itself as Swiss and listed core residential at $0.49 per GB with a 100 GB monthly minimum at $49.99, as of 27 July 2026. Rayobyte is a US operator with no acquisition we can document. Decodo has kept shipping: Proxyway reported on 30 June 2026 that city and ASN targeting were added to its ISP and datacenter products at no extra cost. That is a statement about the public record, not a prediction. Any of them could be acquired next, which is exactly why ownership belongs on the renewal checklist rather than in a one-time selection document.

Comparing those quotes is its own problem, because per-GB, per-IP and per-request pricing are not the same unit. The proxy pricing guide covers the normalization, and per-GB pricing stopped describing what you actually pay for covers why the headline rate keeps drifting away from the bill.

What this means for a team collecting web data

Put an ownership column in whatever holds your vendor list and fill it with the parent entity rather than the brand. Re-check it at every renewal. It takes an afternoon and it is the input to every other decision here.

Keep a fixed set of target URLs you re-test quarterly against each vendor. Pool merges are invisible in billing and obvious in success rates, but only if you have a baseline from before the merge.

Get a second network to something like ten percent of production traffic and leave it there. Not as a disaster plan you have never run, but as a live path with real credentials and real numbers, so that a repricing, a merge or an enforcement action against your primary costs you an afternoon of routing changes instead of a quarter of migration work.

The market got more concentrated. On the evidence available today it did not get more expensive at the entry level, and the self-serve tiers people expected to disappear are still on sale. Both of those can change without anyone announcing it, so re-check them at renewal instead of assuming.