The Official Position vs Reality
Tailwind's documentation says the CDN version is "for development only" and you should "never deploy it in production." That's good default advice — the full CDN bundle is ~400KB and includes every possible utility class. For a high-traffic marketing site, that matters.
But we build a lot of internal tools, admin dashboards, and low-to-medium traffic public sites. For those, the CDN approach with the new Tailwind Play CDN — and a bit of custom config — is perfectly fine and saves significant build complexity.
The Play CDN Difference
The old Tailwind CDN (v2 era) was a static pre-built CSS file. The new Play CDN (Tailwind v3+) is a JavaScript file that detects what Tailwind classes are in your HTML at runtime and generates only those styles on the fly. In practice it's much smaller than the old pre-built bundle for typical pages.
<!-- Drop this in your <head> -->
<script src="https://cdn.tailwindcss.com"></script>
<!-- Custom config (same as tailwind.config.js, inline) -->
<script>
tailwind.config = {
theme: {
extend: {
colors: {
brand: { 500: '#0ea5e9', 900: '#0f172a' }
},
fontFamily: {
sans: ['Inter', 'system-ui', 'sans-serif']
}
}
}
}
</script>Real-World Performance Numbers
We measured on one of our .NET MVC admin dashboards (dark theme, ~30 utility classes per component):
- Play CDN script: ~112KB gzipped (including runtime)
- Build pipeline (PostCSS + purge): ~8KB gzipped
- Difference in LCP on 3G: ~180ms
For an admin tool used by 20 people on a corporate LAN, 180ms is invisible noise. For a public-facing e-commerce page with 10,000 daily visitors, it starts to matter — especially on mobile 4G connections.
When the CDN Approach Is Fine
- Internal admin tools and dashboards
- Prototypes and MVPs that need to move fast
- Low-to-medium traffic public sites (<50k page views/day)
- Projects where adding a Node.js build step is more complexity than it's worth (pure .NET shops, shared hosting)
- When you're using ASP.NET Core MVC and don't already have a JS build pipeline
When You Should Use the Build Pipeline
- High-traffic public pages where Core Web Vitals scores matter
- Mobile-first products where every KB counts
- Projects already using Next.js, Vite, or another JS build tool
- When you need JIT variants like
hover:[&:not(:focus)]:opacity-0that the Play CDN might not handle consistently
The Bottom Line
Tailwind's "never use CDN in production" advice is written for the average case. Know your use case, measure the actual impact, and make the pragmatic call. We run 6 live admin dashboards on the Play CDN with no complaints — and zero Node.js dependencies in the deployment pipeline.