Intelligence · State of Web · 2026
1 in 5 New European Websites Shows Node.js Evidence
Does Node.js need specialized hosting, and how many customers want it?
ShareShift telemetry reveals that nearly one in five new sites now carries Node ecosystem signals. And the trend holds in actual deployments: CloudLinux has confirmed that hosting accounts with Node.js apps have doubled in two years, and 2/3 run on end-of-life versions.
Rather than treating Node as an edge case, providers need to actively size this hidden demand, update security baselines, and align infrastructure and pricing around memory-intensive runtimes.
Target
Hosting product & commercial leaders
Publication date
November 2026
Authors
Nadya Frost (ShareShift) with Nikolay Galov (CloudLinux)
Key Insights
Node.js is growing fast in new sites, mostly outside traditional hosting
In ShareShift data, nearly 1 in 5 new sites shows evidence of the Node.js ecosystem, through frameworks like Next.js, tell-tale digital fingerprints, or deploying on neocloud hosting like Vercel or Netlify. Adoption is accelerating throughout 2026.
Traditional hosting also carries a real, growing, and undercounted Node workload. Nearly 80% of Node.js apps on CloudLinux servers run under LiteSpeed or OpenLiteSpeed, which do not send the Passenger header the scan looks for, making it easy to underestimate the growing adoption. However, the vendor has confirmed that European hosting accounts running Node.js apps on CloudLinux servers have doubled since 2024, reaching 74,000 by October 2026. Of those apps, 65% run on a Node.js version that no longer receives security fixes.
Strategic Insight
Node is a real workload in a minority of sites today and a growing share of new ones, and most of it is not on traditional hosting at all. For a hosting operator, the question is less whether the servers can cope than whether the next generation of customers has any reason to arrive. Operators that treat Node as a distinct segment should plan on memory, patch more aggressively, and defend the domain-first customer relationship around workloads they don’t host.
Part 1: Why We Looked
Does Node Need Its Own Kind of Hosting?
The question keeps coming back in hosting forums, vendor documentation and developer communities: do Node.js sites need hosting built for them?
The case for yes is mechanical. PHP sites are typically served by pooled workers that pick up a request and release it, so one server can carry many sites inside a fixed memory budget. A Node application is a long-running process. It keeps its memory resident whether or not anyone is visiting, so each application takes its share up front and the number of sites a server can carry falls.
So, how should hosting operators adapt conventional infrastructure to "Node.js hosting"?
Real Field Experiment
Shared-Hosting Density and Latency Benchmark
CloudLinux ran 100 production-like Node.js apps on one shared-hosting server through its Node.js Selector. Keeping every app running used about 5.9 GB of memory, with 0.1% of requests taking longer than 100 ms. Stopping apps after 60 seconds without requests cut that to 2.2 to 2.8 GB, but 5.1% of requests then took longer than 100 ms and 1.5% longer than a second, because a stopped app has to restart on its next visit. Test design and limits are in the Methodology section.
The takeaway: in this test, memory use and speed pulled in opposite directions. That makes memory, rather than site count, the fairer unit for pricing Node plans.
Part 2: How Big Is the Market?
How Many European Sites Run Node.js?
A Floor and a Ceiling
Unlike WordPress, which announces itself at the source, a Node application can sit behind proxies or CDNs that strip identifying headers. Single figures hide counting choices, so we report two. The gap between new and existing sites holds on both definitions, and the quarterly cohorts show that adoption is accelerating.
Caution: Node apps on shared hosting often hide their server headers (see the Passenger gap below), so the installed base is probably undercounted more than new sites, which skew towards platforms that expose theirs. Read the new-vs-installed ratio as directional, not exact.
The Ceiling
Active sites with any Node ecosystem signal
This definition includes framework asset signatures and deployments on Node-native platforms (Vercel, Netlify). Growing from 5.9% of existing, up to 18.6% of new sites.
The Floor
Server-Side Node Signal
The definition includes sites on neoclouds, so not all of them need a resident process on a hoster’s server. This segment is 1.9% of all existing sites, and 4.4% of new sites in 2026.
Among domains first seen in 2026 that passed our content filter, the share showing a Node server header rose every quarter, from 2.9% in Q1 to 4.3% in Q3. Next.js in any form expanded from 4.7% to 7.7%.
Data table
| Name | Server Floor (%) | Any Signal (%) |
|---|---|---|
| Installed base (12M domains) | 1.6 | 5.9 |
| New sites (Q3 cohorts) | 4.4 | 18.6 |
Data table
| Name | Node server header (%) | Next.js in any form (%) |
|---|---|---|
| Q1 2026 | 2.9 | 4.7 |
| Q2 2026 | 3.5 | 6.5 |
| Q3 2026 | 4.3 | 7.7 |
Vendor Telemetry
CloudLinux Node.js Fleet Growth
According to CloudLinux, 74,000+ hosting accounts run Node.js apps in Europe (+54% YoY in Oct '26). Globally, 217,000+ accounts run Node (+50% YoY). Telemetry counts apps configured through the CloudLinux Node.js Selector.
Data table
| Year | Europe | Other Regions |
|---|---|---|
| 2024 | 32,000 | 90,000 |
| 2025 | 48,000 | 97,000 |
| 2026 | 74,000 | 143,000 |
| Metric | 2024 | 2025 | 2026 |
|---|---|---|---|
| Europe: accounts with a Node.js app | 0.34% | 0.49% | 0.72% |
| Europe: domains with a Node.js app | 0.17% | 0.21% | 0.30% |
| Europe: Node.js apps | 32,000 | 48,000 | 74,000 |
| All regions: accounts with a Node.js app | 0.50% | 0.62% | 0.93% |
| All regions: domains with a Node.js app | 0.23% | 0.26% | 0.37% |
| All regions: Node.js apps | 122,000 | 145,000 | 217,000 |
Part 3: Which countries?
Which European Markets Are Moving First?
Adoption is uneven across the 18 European TLDs ShareShift scans: a few small-sized outliers lead the pack on share of adoption, while the volume is dominated by larger-market TLDs. For a holistic analysis of differences between markets, refer to our recent State of Europe deep dive.
To illustrate progressive Node adoption, the chart below plots quarterly cohorts of new sites with a share of deployments on Node sever headers, following "the floor" definition. Similarly-performing TLDs are grouped for convenience.
Data table
| Name | Other Tlds | High Adoption | Largest Markets | Mid Size Climbing |
|---|---|---|---|---|
| Q1 | 2.15 | 3.66 | 2.07 | 2.24 |
| Q2 | 2.52 | 3.52 | 2.29 | 2.43 |
| Q3 | 2.1 | 4.85 | 2.26 | 3.19 |
High Adoption
.no, .dk, .pt
The breakaway pack among European ccTLDs; adoption surged to an average of 4.85% in Q3, with Norway crossing the 5% threshold. These smaller, tech-forward registries serve as the clearest early indicators of developer adoption outside the .ai ecosystem.
Mid-size expansion
.es, .be, .fr, .pl, .it, .cz, .nl
Steady, uniform acceleration across mainland Europe; every market in this cluster climbed out of the low 2% range in Q1 to finish above 3.0% by Q3. The trend quantifies Node.js inching toward the mainstream.
Lower tier
.fi, .sk, .ch, .at, .hu, .se
Markets that are yet to build sustained momentum; the group average slipped from 2.52% in Q2 to 2.10% in Q3 with Sweden showing especially high volatility and dropping below 2% in Q3.
The largest markets
.de, .uk
Large absolute site counts but no meaningful share gains yet: .uk held completely flat at ~2.1% across all three quarters, while .de fluctuated at 2.4-2.5%. It'll take a massive swell to turn the tide.
Part 4: What's being deployed where
Node Sites Living on Specialist Platforms
Across the 700,000+ active European sites that carry at least one signal of the Node ecosystem (our "ceiling" definition), 85% are on Node-native platforms. In that space, Vercel dominates with 380,000+ sites, or 64% platform share; Netlify is at 29.5% with Heroku and Fly far behind. The ratios are broadly the same across all European markets except .uk where Netlify is somewhat strong, but still not #1. The UK is the largest market in absolute volume, with .de and .fr in the top-3 by size.
Across header signals, Next.js is the most visible externally, with nearly 200,000 active sites or 1.7% of all functional European sites that passed our content filter this month. Adoption of Next.js is also growing in 2026, while Nuxt and Express remain niche.
Data table
| Name | Value |
|---|---|
| Vercel | 389,883 |
| Netlify | 180,070 |
| Heroku + Fly | 40,609 |
- Vercel63.9%
- Netlify29.5%
- Heroku + Fly6.7%
Data table
| Name | Value |
|---|---|
| Next.js | 198,670 |
| Express | 75,964 |
| Nuxt | 19,262 |
- Next.js67.6%
- Express25.8%
- Nuxt6.6%
| TLD | Express | Next.js | Nuxt | Frameworks, all | Vercel | Netlify | Heroku + Fly | Platforms, all |
|---|---|---|---|---|---|---|---|---|
| .uk | 19674 | 34710 | 1867 | 56251 | 68702 | 42789 | 13378 | 124869 |
| .de | 13178 | 33641 | 3499 | 50318 | 57836 | 26473 | 10167 | 94476 |
| .fr | 10794 | 27374 | 2876 | 41044 | 54960 | 24771 | 4471 | 84202 |
| .nl | 5094 | 17946 | 1987 | 25027 | 32224 | 15414 | 5922 | 53560 |
| .it | 2870 | 9603 | 939 | 13412 | 20812 | 10529 | 1260 | 32601 |
| .pl | 3096 | 13653 | 1238 | 17987 | 22366 | 8218 | 316 | 30900 |
| .es | 2684 | 8079 | 684 | 11447 | 19275 | 7558 | 651 | 27484 |
| All | 76109 | 199071 | 19301 | 294481 | 389883 | 180070 | 40609 | 610562 |
Sharing Customers with Neoclouds
Today, neoclouds take the runtime, not the whole relationship: only about 15% of domains on Vercel use its DNS (first reported in State of AI 2026 and tracked in our Q3 2026 operator rankings), so the domain, and often TLS and email, stays with a traditional provider.
Passenger Telemetry Gap
Server headers are often stripped by default when Node.js is deployed via a platform selector. That's why ShareShift only finds 2,121 domains running Node under Passenger, and why vendor data like CloudLinux paints a fuller picture of the Node market.
Vendor Telemetry
Express Beats Next.js on Shared Hosting
Inside shared hosting servers, the framework hierarchy flips. CloudLinux security agents report 108,000 apps using a server framework, where Express leads at 60% and Next.js sits at 31%.
| Server framework | Share of apps |
|---|---|
| Express | 60% |
| Next.js | 31% |
| NestJS | 3% |
| Nuxt | 3% |
| Fastify, Koa, and Hapi together | 2% |
| SvelteKit | 0.4% |
Security Risk
End-of-Life Node Versions
Apps split evenly between legacy versions (Node 18 and older) and Node 20, which reached EOL on 30 April 2026. Despite EOL, Node 20 apps grew 35% between April and October 2026.
Resource Limits
Memory Caps Are Tight
Half of Node.js apps sit on servers with a 1 GB account cap, and 35% on 2 GB caps. Comparing April to October, adding Node accounts produced no disproportionate spike in resource-limit hits.
| Servers that added | Hitting limit (April) | Hitting limit (October) |
|---|---|---|
| No new Node.js accounts | 13.8% | 16.6% |
| 3 to 10 Node.js accounts | 11.4% | 14.0% |
| 11 or more | 10.0% | 11.9% |
Part 5: Strategic Playbook
What Hosting Operators Should Do About Node in 2026
Five concrete commercial and architectural moves for hosting providers looking to capture next-generation web workloads.
Size the segment in your own base
The market averages are the starting point: a floor of 1.9% and ceiling of 6.9% for existing sites, 4.4% and 18.6% for new ones. Run both measures on your customer base to establish your specific workload vs ecosystem ratio.
Split the segment before you price it
Over half of Next.js sites on new domains are static exports with no Node server signal, which ordinary hosting can serve efficiently. Only the server-side minority (c. 4% of new sites) can need a dedicated resident process, and some of it runs on neoclouds rather than on a hoster’s servers.
Price the process-hosted tier on memory, not on sites
A resident process costs RAM regardless of traffic. Selling Node plans on disk and domain counts leads to negative margins or heavy throttling that harms customer experience.
Start where new-site adoption is highest
Consider your geographic footprint: markets like .uk and .fr are seeing the highest absolute volume of Node deployments, while Norway, Denmark and Portugal are showing the fastest growth in percentage points. The rest of Europe is holding relatively steady.
Keep the relationship around the runtime
Only about 15% of domains on Vercel use its DNS. Domains, TLS certificates, and email stay with traditional providers. Operators unable to host the runtime can still maintain the core customer relationship around it.
Compare your Node.js base to the Market
Receive a custom analysis and an export of the domains worth targeting for upgrades. Learn how much of your base carries Node signals, what else is attached to those domains, and how you compare to your market.
Appendix
Methodology & Scope
ShareShift weekly DNS, HTTP, and TLS scans across European ccTLDs. Full attribution rules and detectability limits are available at shareshift.io/methodology. This section highlights specifics relevant to this report.
This research was produced in partnership with CloudLinux, which develops the Node.js Selector used in the field test and has a commercial interest in Node.js hosting. ShareShift did not run or independently verify the test.
Timeframe
Latest telemetry scans delivered 19 October 2026. Cohort analysis covering 3, 6, or 9 months refers to domain intake first seen during that month.
Exclusions
Full dataset spans 60M+ domains. The "enriched" active content set of 12M is used as the denominator; it excludes parked, unused, and empty domains.
CloudLinux Data
Scope: analysis considers 14.5M hosting accounts and 55M domains worldwide (6.3M accounts and 22.9M domains in EU/EEA/UK/CH). Counts apps configured via Node.js Selector.
Experiment setup: one server (CloudLinux 9.8 with cPanel, 12 vCPU, 23 GB RAM) running 100 apps (40% Express storefronts, 30% Next.js 15 shops, 30% NestJS APIs), each with its own MySQL database. Memory is measured as PSS, so shared memory is counted once. Reference points: a minimal Express app uses 18 MB and an Express storefront with 34 dependencies uses 56 MB; a container-per-app deployment of the same apps needed 10.4 to 10.9 GB.
Experiment Limits: the test mix differs from the fleet composition reported in this paper (Express 60%, Next.js 31%, NestJS 3%), there was no PHP baseline, and latency was measured in a single run. Treat the figures as an indication of the trade-off, not as a sizing guide.