Real Market Share of JavaScript Frameworks (npm 2026

TL;DR

React controls ~60% of new JavaScript project starts in 2026; Vue and Angular each hold ~12-15%; everyone else fights for the rest. But these numbers obscure a more interesting story: React's dominance is concentrated in new projects and startups, while Angular dominates enterprise; Vue is disproportionately popular in Asia; and Svelte/Astro are capturing the developer satisfaction demographic. Framework "market share" is less useful than framework-for-your-context share.

Key Takeaways

• React: ~60% of new projects — down from 75% peak, but still dominant
• Angular: ~15% — enterprise-heavy; looks smaller in surveys, larger in enterprise
• Vue: ~12% — global reach; Asia representation makes surveys undercount it
• Svelte: ~8% — growing fastest in satisfaction scores; punching above its download weight
• Meta-frameworks: Next.js, Nuxt, SvelteKit tell a cleaner adoption story

How to Measure Framework Market Share (And Why It's Hard)

The measurement problem:

npm downloads:
→ Counts CI installs, monorepo deps, legacy apps
→ React "installs" include every project that installs create-react-app
→ Angular "installs" include enterprise apps installed in CI 100x/day
→ Not representative of NEW project selection

Developer surveys (State of JS, Stack Overflow):
→ Biased toward English-speaking respondents
→ Vue significantly underrepresented (strong Asia userbase)
→ Surveys favor indie developers, not enterprise devs
→ Sample: 35K devs ≠ all 18M JS developers

Job postings:
→ Lags adoption by 12-18 months (enterprise hiring, old tech stacks)
→ React jobs: still 80%+ of "framework required" postings
→ Angular: 15% of enterprise-specific postings
→ Vue: 10%, growing in startup postings

GitHub dependency graph:
→ Best signal: counts repos that actually DEPEND on each framework
→ Not publicly aggregated, but approximated by Sourcegraph/GitHub

The most honest answer: React is dominant by all measures,
the gap with Vue/Angular is smaller than npm downloads suggest,
and satisfaction metrics predict a future where Svelte/Astro are larger.

The measurement problem is not just a methodological footnote — it fundamentally shapes how framework debates get framed online. npm download counts are the easiest numbers to cite and the least informative for gauging actual active usage. React downloaded 200 million times a week is not 200 million developers using React; it is download events across CI pipelines, Docker builds, monorepo package installs, and corporate artifact mirrors. Vue's 50 million weekly downloads do not represent one-quarter of React's usage — Vue's strongholds are in organizations and geographies where CI pipelines run differently, and where projects may be installed and rebuilt less frequently.

Developer surveys have the opposite problem: they reach self-selecting respondents plugged into the English-language technical newsletter ecosystem, which skews heavily toward React and the Vercel ecosystem. Angular developers at large enterprise organizations do not fill out State of JS surveys. Vue developers at Chinese e-commerce companies are systematically underrepresented in Western survey samples. The satisfaction scores and usage percentages in these surveys are real measurements of a real population — but that population is not representative of professional JavaScript development globally.

The most reliable signal for new-project selection — the decision that actually shapes the ecosystem's future composition — is combining meta-framework download velocity, job posting growth rate, and GitHub repository initialization patterns. None of these is individually clean, but together they filter out the worst noise from each source.

Market Share Estimates (2026)

New project selection (estimated, based on combined signals):

React ecosystem: 58%
├── Next.js:         35% of all new projects
├── Vite + React:    18%
└── Other React:     5%

Vue ecosystem: 13%
├── Nuxt:            7%
└── Vite + Vue:      6%

Angular ecosystem: 12%
├── @angular/cli:    12% (monolithic, hard to split)

Svelte ecosystem: 7%
├── SvelteKit:       5%
└── Svelte standalone: 2%

Astro: 5%
(framework-agnostic, but counts as separate framework choice)

Solid.js: 2%
Qwik: 1%
Other: 2%

The meta-framework column is more informative than the raw framework number for new-project selection. "React" is too broad — it includes developers choosing Next.js for a full-stack SaaS, Vite plus React for a purely client-side SPA, and Remix for a server-rendered application with progressive enhancement. These are meaningfully different architectural choices even though they share a component library. Next.js alone at 35% of new projects is larger than Vue's entire ecosystem share — that concentration matters for ecosystem investment, hiring, and tooling support.

---

By company size:
Startups (1-10 engineers): React 70%, Svelte/Astro 15%, Vue 10%, Other 5%
Mid-size (11-100): React 55%, Vue 15%, Angular 20%, Others 10%
Enterprise (100+): Angular 35%, React 45%, Vue 10%, Others 10%

By geography (estimated):
North America: React 65%, Angular 15%, Vue 10%, Others 10%
Europe: React 55%, Angular 20%, Vue 15%, Others 10%
Asia: Vue 30%, React 40%, Angular 20%, Others 10%

The company-size breakdown is where the aggregate numbers become most misleading. React at 60% of all new projects includes React at 70% of startup projects, which pulls the aggregate up substantially. Enterprise organizations — the ones with large engineering teams, long-lived codebases, and high job posting volumes — have a very different distribution. Angular at 35% of enterprise new projects is not visible in the aggregate number, but it creates a durable hiring market for Angular developers that is invisible to developers reading startup-ecosystem publications.

The geographic distribution carries equally significant implications. Vue's 30% in Asia is not a historical curiosity — it reflects that PHP-plus-Vue and Laravel-plus-Nuxt became the dominant web application stack across much of East and Southeast Asia, with their own publication networks, conference circuits, and hiring pipelines underrepresented in English-language technical discourse. A developer evaluating Vue for a project serving Japanese or Chinese markets is operating in a context where Vue expertise is abundant and the local talent market favors it.

Why React Still Dominates

React's structural advantages (2026):

1. Ecosystem depth
   → 3M+ npm packages mention react in package.json
   → More shadcn/ui components than any other framework
   → More tutorials, Stack Overflow answers, YouTube videos

2. Job market gravity
   → "React developer" is the most common frontend job title
   → Teams hire for what they know → creates supply of React devs
   → React devs beget more React projects

3. Meta-framework quality
   → Next.js is best-in-class full-stack React
   → Vercel investing heavily in Next.js (direct financial incentive)
   → RSC + Server Actions are real competitive advantages

4. The Facebook brand (mixed blessing)
   → Trust from "built by a large company" crowd
   → But also: React decisions sometimes felt corporate-first
   → Community ambivalence: React is used, not always loved

5. Migration cost
   → Existing React codebases are enormous
   → Migrating to Vue/Svelte: 3-12 months for large apps
   → Most teams choose to continue: "devil you know"

The flywheel dynamics behind React's dominance explain why a tool can hold market share even as developer satisfaction declines. React's ecosystem network effects compound: more React users generates more React library authors, more React tutorials, more React-experienced job candidates, and more React-first tooling decisions at major companies. Each reinforces the others. A developer entering the job market chooses React because that is where the jobs are; companies hire React developers because that is who is available; libraries are built for React because that is where the users are.

This flywheel has been operating for roughly a decade, which is why React's decline from 75% peak new-project share to 58-60% now has been gradual despite lower satisfaction scores than competitors. The developers most likely to switch are those who have achieved career security and can afford to optimize for developer experience rather than employment risk. That group is growing — but it remains a minority of the JavaScript workforce.

The RSC and Server Actions investment by Meta and Vercel represents an attempt to extend React's advantage by solving the server-rendering problem within the React paradigm. If Next.js continues growing, the flywheel gains a new rotation: full-stack React becomes the default for developers who previously needed a separate backend, further entrenching React's installed base.

Where Each Framework Actually Wins

React wins for:
→ Hiring from a large candidate pool
→ Complex SPAs with lots of dynamic state
→ Teams that need a broad ecosystem of compatible libraries
→ Projects where Meta/Vercel's ongoing investment matters

Vue wins for:
→ PHP-heavy backends (template style is familiar to PHP devs)
→ Asia-based companies (Laravel+Vue is the standard stack there)
→ Teams that prefer progressive enhancement (you can sprinkle Vue)
→ Projects where gentle learning curve matters

Angular wins for:
→ Large enterprise teams with strict conventions
→ Projects where dependency injection and strong typing matter from day 1
→ Teams with Java/C# background (class-based, OOP patterns)
→ Long-lived applications where refactoring safety matters

Svelte wins for:
→ Performance-critical applications (least framework overhead)
→ Teams tired of React's complexity (simpler mental model)
→ Small-to-medium apps where DX matters most
→ Developers who've used React/Vue and want something fresher

Astro wins for:
→ Content-heavy sites (blogs, docs, marketing)
→ Teams that want to use React/Vue/Svelte in the same project
→ Projects where Core Web Vitals are a hard requirement
→ Static site generation with optional interactivity

Solid.js wins for:
→ Maximum performance applications
→ Teams that like React's model but want real reactivity
→ When you need the best possible performance per KB of JavaScript

The "wins for" framing in each category is not an endorsement — it is an acknowledgment that decision constraints vary enough across teams that no framework is universally superior. The meta-framework layer has more meaningful differentiation than the component-library layer. The real question in 2026 is rarely "React or Vue?" and more often "Next.js or Nuxt?" or "Next.js or SvelteKit?" — those comparisons package opinionated routing, rendering strategy, and deployment assumptions alongside the component framework.

The convergence toward meta-framework-first thinking is itself a signal of ecosystem maturity. When a framework ecosystem develops a dominant meta-framework — Next.js for React, Nuxt for Vue, SvelteKit for Svelte — it signals the community has solved the undifferentiated infrastructure problems: routing, server-side rendering, API routes, image optimization. Teams no longer assemble these from scratch; they choose a meta-framework and inherit reasonable defaults. Comparing meta-framework quality and developer experience is often more decision-relevant than comparing the underlying component libraries.

The Satisfaction Gap (Why Survey Data Predicts the Future)

Framework satisfaction scores (% of users who would use again):
(State of JS 2025 approximation)

Svelte:          74% would use again  ← highest
Astro:           73% would use again
Solid.js:        71% would use again
Vue:             66% would use again
React:           62% would use again
Qwik:            58% would use again
Angular:         42% would use again  ← lowest

Pattern: satisfaction scores predict adoption 2-3 years out.
The developers who are most satisfied share their enthusiasm.
The developers who are least satisfied look for alternatives.

This data suggests:
→ Svelte and Astro will continue growing
→ React's growth rate will continue slowing (62% satisfaction is low for dominant tool)
→ Angular faces demographic challenge: new developers choose React/Vue/Svelte
→ Solid might break mainstream threshold in next 2-3 years

Counterpoint: Satisfaction ≠ adoption
→ React's 62% satisfaction on 60% market share = massive absolute number
→ Even declining satisfaction won't shrink React fast (migration cost)

The satisfaction-to-adoption lag follows a consistent pattern in software ecosystems. High satisfaction signals developer enthusiasm that generates blog posts, conference talks, and organic recommendations — the community marketing that drives future adoption. Low satisfaction in a dominant tool signals developers actively seeking alternatives and willing to recommend switching to colleagues. The lag between satisfaction signal and adoption change is typically two to four years, filtered through hiring cycles and project starts.

The React satisfaction number requires a nuance worth stating clearly: 62% satisfaction on 60% market share is an enormous absolute count of satisfied React developers. More developers would use React again than there are total developers using any other framework. The satisfaction ratio is low, but the ecosystem implications materialize slowly because the installed base is so large.

Angular's 42% satisfaction is largely driven by developers who used Angular between 2016 and 2020, before Signals and before the zone.js complexity was addressed. Angular from version 16 onward has substantially different ergonomics. Developer satisfaction data typically lags actual product quality improvements by several years — the metric measures remembered pain as much as current experience.

What to Pick for a New Project in 2026

The honest framework recommendation:

Building a startup/SaaS: Next.js (React)
→ Best ecosystem, best hiring, best meta-framework
→ RSC + Server Actions = full-stack in one framework
→ Vercel deployment = zero-config production

Building a content site: Astro
→ Best performance defaults
→ Use any component framework you want inside it
→ Markdown/MDX built-in

Building for an enterprise with 50+ developers: Angular
→ Strict conventions at scale
→ Angular 17+ with Signals improved DX significantly
→ Large team coordination > individual DX

Care deeply about DX and less about hiring: SvelteKit
→ Highest satisfaction scores
→ Simplest mental model
→ You'll be happier, but hiring is harder

Want the best possible performance: Solid.js
→ JSX syntax (React-like) with real reactivity
→ Smallest runtime in category
→ Ecosystem still maturing

The recommendation matrix is useful as a starting point, but the most important input to framework selection is not represented: the existing expertise of the team making the decision. A three-person team with deep Vue expertise building a startup should build in Vue — the productivity penalty of switching frameworks to a theoretically superior option costs more than any framework-level technical advantage could return. The best framework for a team is nearly always the one they can ship in most effectively right now, calibrated by the expected lifetime of the project and its hiring requirements.

The exception is when significant team growth is expected and hiring will become a constraint. For a team planning to scale from five to fifty engineers, React's dominant hiring market share is a compounding advantage. Building in SvelteKit with five engineers who love it can become a significant constraint when you need to hire twenty more — the pool of experienced SvelteKit developers is meaningfully smaller than the React pool, and training React developers to use Svelte takes time that competes with shipping product.

The Job Market Reality (2026 Data)

npm download share and job market demand correlate, but with a significant lag and several important distortions. Enterprise hiring decisions trail adoption trends by three to five years: a large organization that standardized on Angular in 2019 will still be posting Angular developer roles in 2026, regardless of what share Angular commands in new project starts. This makes job posting data a measure of where the industry's installed base is, not where it is headed.

React's dominance in the job market — roughly 65% of postings that specify a framework — reflects this installed base effect as much as it reflects new project selection. That share is not disappearing quickly. The enormous number of existing React codebases, and the React-native hiring pipelines that organizations have built around them, create a flywheel that is largely independent of what individual developers prefer to use.

Angular's 15 to 20% job market share is almost entirely enterprise-driven. Large financial institutions, healthcare organizations, and government contractors that adopted Angular for its strong conventions and dependency injection system continue to hire Angular developers. These organizations do not make framework decisions based on State of JS survey results; they make them based on existing investment, compliance requirements, and team capabilities.

Vue is systematically undercounted in Western job data for the same reason it is undercounted in developer surveys: its strongholds are in Asia, particularly China and Japan, where Laravel-plus-Vue is the dominant web stack for a large number of companies. The job postings that appear on LinkedIn or Indeed in North America and Europe are not representative of Vue's actual global employment footprint.

The practical signal for individual developers: React remains the highest-confidence investment for career employability in most markets. The Stack Overflow Developer Survey and State of JS data roughly confirm the download and job posting signals — React is used most, even if it is not always loved most.

Market Share Projection: 2027-2028

The broad trajectory of the next two years is visible in the current data, with a few important qualifications. React staying at 60 to 65% of new project starts seems likely — it would require a substantial technical or community failure to accelerate decline from that position. The more interesting story is in the meta-framework layer, where the real competitive action is happening.

Next.js retains a commanding lead in the React meta-framework category. Vue's equivalent, Nuxt 3, has matured significantly and is gaining traction with teams outside Asia who previously avoided Vue due to its weaker TypeScript story. SvelteKit is growing fastest in percentage terms but from a smaller installed base — the percentage growth is high because the denominator is small.

Astro's trajectory is the most interesting to watch. Astro occupies a specific niche — server-side rendering for content-heavy sites — that was previously served by Next.js (overkill for pure content), Gatsby (slow build times, effectively deprecated), and classic SSGs like Jekyll or Hugo (limited interactivity). Astro combines the performance defaults of static generation with the ability to embed React, Vue, or Svelte components where interactivity is actually needed. By 2028, Astro is likely to surpass Remix in weekly downloads; it has already caught up on developer satisfaction scores.

Angular's trajectory is stabilization rather than growth. The Signals-based reactivity system introduced in Angular 16 and matured through Angular 17 and 18 addressed the developer experience complaints that drove many teams away. Angular is unlikely to recapture new-project market share from React or SvelteKit, but it is also unlikely to face the existential decline that some predicted during its lowest-satisfaction years.

The metric most developers should track for forward-looking signals is job posting growth rate by framework — not download velocity, which is heavily weighted by existing CI pipelines. Where are new job postings growing as a percentage? That's where the industry is heading, and where hiring market liquidity will exist in three years.

Developer satisfaction scores are the leading indicator most correlated with that job posting growth rate, with a three to four year lag. Svelte's 74% satisfaction and SvelteKit's continued growth are consistent with the prediction that SvelteKit job postings will increase meaningfully from 2026 to 2029. Angular's sustained low satisfaction is consistent with the prediction that Angular's new-project share — already declining — will continue shrinking, even if existing installations remain in production for a decade or more. Tracking satisfaction now is how you anticipate the hiring market three years out, not just the current one.

What "Market Share" Means for Choosing a Framework

Framework market share data is useful background context, not a decision rule. The framework that is "winning" in aggregate may not be the right choice for a specific team, project, or organization. The aggregate hides variations that matter for actual decisions.

For new projects at a startup or small team: React plus Next.js remains the highest-return-on-investment choice for most teams. The ecosystem depth, hiring market, and meta-framework quality are all best-in-class. This is not a revolutionary finding — it is what the data shows when you cut through the noise.

For teams with existing Vue expertise, or teams operating in markets where Vue developers are more available or affordable, Vue plus Nuxt is a practical choice that does not require justification. The technical quality of Vue 3 is not a limiting factor; the ecosystem gap with React has narrowed considerably.

For performance-critical applications where the team is small and framework overhead matters: Svelte. The mental model is simpler, the output is leaner, and the developer experience scores are highest of any framework. The hiring constraint is real and should be factored into the decision, but for a two or three person team building a focused application, it is manageable.

For large enterprise organizations: Angular when the organization already has Angular investment or when the disciplined structure is genuinely needed for a large, complex application with many contributors. Not because Angular is technically superior in those cases, but because a framework that enforces conventions prevents the accumulated inconsistency that kills large codebases over time.

The key insight that market share data obscures: framework choice is mostly about team skills, hiring market, and existing codebase — not about the frameworks' inherent technical qualities. At the meta-framework level, they are all capable of building excellent production applications. Choose the one your team knows, or the one your industry hires for.

What the market share data is genuinely useful for is calibrating expectations: how much Stack Overflow content exists for your framework, how many open-source UI libraries are available, how easy it will be to find senior engineers in two years. React wins all three of those metrics by a wide margin, which is why "just use React" is often practical advice for teams without strong contrary signals — not because React is the best-designed or most elegant framework, but because those ecosystem properties — documentation depth, library availability, and hiring pool — compound into real, durable advantages over the full lifetime of a production application.

Compare framework package health and download trends at PkgPulse.

See also: Angular vs Vue and Angular vs React, JavaScript Framework Adoption by Company Size.