Skip to main content
Streaming allows the server to send parts of a page to the client as soon as they’re ready, rather than waiting for the entire page to render. Users see content sooner, even if parts of the page are still loading.

​
How streaming works

Without streaming, the browser must wait for the server to render the full page before showing anything. With streaming, Next.js sends a static shell immediately and streams in dynamic content as it resolves: 
Without streaming:
  request → [wait for all data] → full HTML → display

With streaming:
  request → static shell → display
             ↓
             dynamic chunk 1 → streamed in
             dynamic chunk 2 → streamed in
Streaming is built on HTTP chunked transfer encoding. Each <Suspense> boundary in your component tree becomes a streaming chunk.

​
Two ways to stream

loading.js

Streams an entire route segment. Wraps page.tsx in a <Suspense> boundary automatically.

React Suspense

Streams specific parts of a page. Gives you granular control over which components stream.

​
With loading.js

Create a loading.js file in the same folder as your page to show a loading state while the page renders: 
export default function Loading() {
  return <div>Loading...</div>
}
On navigation, the user immediately sees the layout and loading state. The new content swaps in once rendering is complete. How it works internally: loading.js is nested inside layout.js and automatically wraps page.js and its children in a <Suspense> boundary. 
app/blog/
├── layout.tsx    ← wraps everything
├── loading.tsx   ← becomes <Suspense fallback={<Loading />}>
└── page.tsx      ← streamed inside the Suspense boundary
A layout that accesses uncached or runtime data (such as cookies(), headers(), or uncached fetches) does not fall back to the same route segment’s loading.js. It blocks navigation until the layout finishes rendering. Move data fetching into page.js where loading.js can cover it, or wrap the uncached access in its own <Suspense> boundary.

​
With <Suspense>

For more granular control, wrap specific components in <Suspense> boundaries: 
import { Suspense } from 'react'
import BlogList from '@/components/BlogList'
import BlogListSkeleton from '@/components/BlogListSkeleton'

export default function BlogPage() {
  return (
    <div>
      {/* Sent to the client immediately */}
      <header>
        <h1>Welcome to the Blog</h1>
        <p>Read the latest posts below.</p>
      </header>
      <main>
        {/* Streams in when BlogList resolves */}
        <Suspense fallback={<BlogListSkeleton />}>
          <BlogList />
        </Suspense>
      </main>
    </div>
  )
}
Content outside the <Suspense> boundary (<header>) is sent immediately. Content inside streams in when the async work completes.

​
Creating meaningful loading states

Design loading states that help users understand the app is responding. Good fallback UI examples:
  • Skeletons: placeholder shapes that match the layout of the final content
  • Spinners: for simple, short-duration loads
  • Partial content: a cover photo or title before body content loads
export default function PostSkeleton() {
  return (
    <div className="animate-pulse">
      <div className="h-6 bg-gray-200 rounded w-3/4 mb-4" />
      <div className="h-4 bg-gray-200 rounded w-full mb-2" />
      <div className="h-4 bg-gray-200 rounded w-5/6" />
    </div>
  )
}

​
Streaming with Server Components

Server Components and <Suspense> work together. An async Server Component inside a <Suspense> boundary streams its content when it resolves: 
import { Suspense } from 'react'

async function SlowComponent() {
  // Simulates a slow data fetch
  const data = await fetch('https://api.example.com/slow-data')
  const json = await data.json()
  return <p>{json.message}</p>
}

export default function Page() {
  return (
    <>
      <h1>Page Title</h1>
      {/* Shown immediately */}
      <p>Static content that renders right away.</p>
      {/* Streams in when SlowComponent resolves */}
      <Suspense fallback={<p>Loading slow content...</p>}>
        <SlowComponent />
      </Suspense>
    </>
  )
}

​
Streaming data from Server to Client

Pass an unawaited promise from a Server Component to a Client Component, then resolve it with the use API: 
import Posts from '@/app/ui/posts'
import { Suspense } from 'react'

async function getPosts() {
  const res = await fetch('https://api.example.com/posts')
  return res.json()
}

export default function Page() {
  const posts = getPosts() // Don't await

  return (
    <Suspense fallback={<div>Loading...</div>}>
      <Posts posts={posts} />
    </Suspense>
  )
}

'use client'
import { use } from 'react'

export default function Posts({
  posts,
}: {
  posts: Promise<{ id: string; title: string }[]>
}) {
  const allPosts = use(posts) // Suspends until the promise resolves

  return (
    <ul>
      {allPosts.map((post) => (
        <li key={post.id}>{post.title}</li>
      ))}
    </ul>
  )
}

​
Streaming with Partial Prerendering

When Cache Components is enabled, Next.js uses Partial Prerendering (PPR) by default. The static shell (including <Suspense> fallbacks) is prerendered at build time. Dynamic content streams in at request time: 
import { Suspense } from 'react'
import { cookies } from 'next/headers'
import { cacheLife } from 'next/cache'

export default function Page() {
  return (
    <>
      {/* Prerendered at build time */}
      <h1>My Blog</h1>

      {/* Cached — included in static shell */}
      <BlogPosts />

      {/* Fallback prerendered; content streams at request time */}
      <Suspense fallback={<p>Loading preferences...</p>}>
        <UserPreferences />
      </Suspense>
    </>
  )
}

async function BlogPosts() {
  'use cache'
  cacheLife('hours')
  const res = await fetch('https://api.vercel.app/blog')
  const posts = await res.json()
  return <ul>{posts.map((p: any) => <li key={p.id}>{p.title}</li>)}</ul>
}

async function UserPreferences() {
  // Reads cookies — dynamic, streams at request time
  const theme = (await cookies()).get('theme')?.value || 'light'
  return <p>Theme: {theme}</p>
}
The <Suspense> fallback (<p>Loading preferences...</p>) is included in the static HTML shell sent on the first request. The UserPreferences content streams in once it resolves.

Build docs developers (and LLMs) love

Get started for freeTalk to us