Companies Using Vercel

Vercel is the second-largest platform in the TechnologyChecker.io PaaS category and the first one where the detection method is genuinely uncomplicated. That is worth saying plainly, because the two profiles either side of it are not. This page publishes 389,653 domains, and I want to explain what that number contains before anyone builds a prospecting list on it.

What actually triggers a Vercel detection

Three signals, and they are not equal. The first is a response header on the page itself: x-vercel-id, x-vercel-cache or the legacy x-now-trace. The second is DNS, where the domain's start-of-authority record points at vercel-dns.com. The third is a script loaded from a vercel.app hostname.

Only the third one can lie, because a vercel.app URL might belong to a vendor whose widget you embed rather than to you. We spent real effort on the AWS profile establishing that its equivalent clause accounts for about a third of its count, so the same question had to be asked here. The answer is that it barely matters: the busiest shared vercel.app host in our data appears on 766 domains, and the next eleven together add roughly 1,900 more. Against 389,653 that is noise.

The sample, and what the failures mean

We drew 100 active Vercel domains at random and checked each one by hand, resolving the start-of-authority record and reading the live response headers. Eighty-five carried a first-party signal. Headers did almost all of the work at 84%, with Vercel DNS at 10% and an overlap of 9%.

The fifteen that did not confirm are the interesting part. Five returned a 403 of about 4.5KB from Cloudflare, which is a bot block and tells us nothing either way. Two were unreachable. The remaining eight were serving from somewhere else entirely, including Wix, Framer and IIS, which means they had moved on since we last saw them. None of them looked like a third party's vercel.app widget sitting on an unrelated site.

A CDN in front of an origin replaces the headers we read. That makes a Cloudflare-fronted Vercel site invisible to us, so every number on this page is a floor, and the enterprise end of the list is where the undercount will be worst.

Why the company list is shorter than it could be

We tested the seventy largest enriched domains and confirmed sixty-two. The eight we dropped include some names that would have looked good on this page, and the reason we dropped them is the paragraph above: they sit behind Cloudflare. I would rather publish a list where every row was re-checked at the exact hostname than a longer one padded with entries I cannot verify today.

Vercel is a small-company platform, and the comparison is fair

89.73% of enriched Vercel domains are companies under 50 people, and 0.89% are above 1,000. The equivalent figure is 7.49% on Azure and 2.56% on AWS. That comparison holds up because all three are detected through response headers and DNS records, in the same category, in the same month, so we are not comparing a header rule against a JavaScript pixel and calling the difference a market insight.

Company age says the same thing from another angle. 47.04% of Vercel companies were founded in the 2020s and 81.85% since 2010. On AWS the 2020s figure is 23.27% and on Azure it is 11.48%, where 44.44% predate 2000. These three platforms are selling to three different generations of company.

The company behind the number

Vercel raised $300 million at a $9.3 billion valuation in September 2025, a Series F co-led by Accel and GIC, having started out as ZEIT in 2015 under Guillermo Rauch. It also owns Next.js, which was downloaded more than 500 million times in the trailing twelve months, beating the framework's entire 2016 to 2024 total. Its AI SDK went from 446,012 weekly downloads to roughly 3.2 million over a single year.

That is not trivia on a detection page, it is the explanation for the stack table below. A hosting company does not usually get to ship the framework its customers build in. Vercel does, and the consequence is measurable here as 67.77% Next.js and 72.95% Tailwind against 4.15% Bootstrap. A valuation is a claim about the future. The co-occurrence is a measurement of the present, and of the two it is the one I would build a prospecting list on.

The stack signal is the most useful thing on this page

Tailwind CSS appears on 72.95% of Vercel domains and Next.js on 67.77%. I have not seen co-occurrence that high on any other infrastructure profile at TechnologyChecker.io. The control that makes it meaningful is Bootstrap, which sits on 5.7 million domains across our corpus and reaches only 4.15% here. A platform that were merely popular would inherit the web's baseline distribution of CSS frameworks. Vercel does not, which tells you its adoption is driven by a specific toolchain rather than by general hosting demand.

Churn, and the one thing it does not prove

Vercel is net positive against every competitor we tested, which is rare enough that I checked it twice. It gained 2,100 domains from AWS while losing 634, and beat Netlify, Firebase and Supabase on the same basis. The caveat is that our churn window opened in May 2026, so this is a snapshot of recent movement rather than a full history, and a domain running both Vercel and AWS can appear in these corridors without anybody having cancelled anything.

Read 389,653 as a floor on domains that Vercel serves directly and that are not hidden behind a CDN. It is not a count of Vercel accounts, and one agency account can sit behind a great many of these domains.

What this page is a floor on, and what it is not

Two limits are worth stating together, because they pull the number in the same direction. A CDN in front of an origin hides the headers we read, and we only crawl hostnames we know about, so a Vercel deployment on an uncrawled subdomain is simply absent. Both of those cause undercounting, and neither has a correction factor I would trust enough to apply.

The opposite error, counting somebody who is not a customer, is the one we can bound. It would require a third party's vercel.app script on an unrelated site, and we measured that population directly at well under 1%. So 389,653 is a conservative figure. When this page and a vendor's own announcement disagree, the gap is far more likely to be sites we cannot see than sites we counted wrongly.

One last framing point for anyone using this commercially. A domain is not an account. Agencies deploy client work under their own Vercel team, and agencies are 6.89% of the enriched list, so the count of paying Vercel customers behind these 389,653 domains is smaller than the domain count implies. We publish domains because that is what we observe.

Sales Signals

• Next.js plus Tailwind on a domain we do not yet see on Vercel is the highest-intent migration target in this dataset, because the stack is already built for it.
• Supabase or Firebase alongside Vercel (1.92% and 2.3%) marks a team running a serverless backend with no traditional cloud contract, which is a different buyer from an AWS account.
• Vercel Analytics at 19.01% separates paid-tier behaviour from hobby deployments better than domain count does.
• An agency in the list (Advertising, Marketing and Design Services are 6.89% together) is a channel, not one account. Sell once, land many domains.
• India at 9.45% is double its share of the AWS list, so territory planning built on AWS data will under-resource it.
• An enterprise row is almost always a marketing or careers property. Pitch the web team, not central IT.

Everything above comes from the TechnologyChecker.io August 2026 crawl, and the hand-checks were run against live sites the same week. Where a number rests on a sample rather than the full corpus, the sample size is stated next to it.