
With Next.js 16.4, released on October 6, 2026, the Next.js team now recommends Cache Components for every app. New projects from create-next-app get them by default, and the same post says they become the default in Next.js 17. If you maintain an App Router codebase, the switch is coming either way. The only open question is whether you make it on your schedule or on the framework’s.
What 16.4 actually changed
Caching stops being something the framework guesses. You mark what may be cached with ‘use cache’, give it a lifetime, and everything else renders at request time behind a Suspense boundary. Earlier App Router versions cached some fetches and routes implicitly, which is why so many teams have a story about a page that would not update. Why move before 17 forces it? Because the work is mostly mechanical, and mechanical work is cheaper when you pick the week.
Turning it on in an existing app
The caching guide enables the model with two flags. Set both: leaving partialPrefetching unset logs a warning, because the first release of Cache Components shipped without Partial Prefetching. Then pair each cached function with an explicit lifetime.
// next.config.ts
import type { NextConfig } from ‘next’
const nextConfig: NextConfig = {
cacheComponents: true,
partialPrefetching: true,
}
export default nextConfig
// app/lib/products.ts
import { cacheLife } from ‘next/cache’
export async function getProducts() {
‘use cache’
cacheLife(‘hours’)
return db.query(‘SELECT * FROM products’)
}
The guide recommends a cacheLife on every cache directive. Without one, the implicit default profile applies, and you are back to guessing what the framework decided.
What breaks when caching becomes explicit
Uncached reads outside Suspense
Any component that fetches fresh data now has to be either cached or wrapped in Suspense. Miss one and the dev overlay flags the route as blocking, with a suggested fix. On a large app this is most of the work, and it is the part worth doing route by route rather than all at once.
Random values and timestamps
Calls such as Math.random(), Date.now() and crypto.randomUUID() need a decision. Either defer them to request time with connection() inside Suspense, or cache the value so every visitor sees the same one.
Caches that do not survive a deploy
Entries created with ‘use cache’ are scoped to a single deployment. A new deploy starts cold. If your traffic spikes right after releases, test that path before you rely on the cache to absorb it.
Static routes that quietly turn dynamic
One dynamic component can turn an otherwise static page into request-time work. The new ensureStatic export in 16.4 fails the build when that happens, which makes it a cheap guard for marketing pages and blog posts.
An order of work that keeps the app shippable
Teams that try to convert everything in one pull request usually end up reverting it. A sequence that keeps main deployable looks like this.
- Upgrade to 16.4 on its own branch first. If you work with a coding agent, next upgrade –agent prepares the migration guides, codemods and verification steps for it.
- Ship the plain upgrade before touching caching, so a regression has one cause.
- Turn on both flags and run the dev server. Work through the overlay’s insights one route at a time, starting with the most visited.
- Add a cacheLife to every ‘use cache’ you introduce, and write down why each lifetime was chosen.
- Add ensureStatic to the layouts that must never render at request time.
- Deploy to staging and measure the first minutes after a release, when every cache is cold.
When the move needs another team
Most apps can do this in-house. The hard cases are codebases that grew for years under the old caching rules, where nobody remembers which page depended on an implicit cache. When the codebase is large and the release calendar is tight, teams often bring in a partner offering next js development services for the move itself.
Long-lived codebases are where that experience comes from. Pagepro, a Next.js and Sanity migration specialist, has worked with Proofed since 2020 and built the front end of its order-management platform, including a Next.js middleware security layer, for the US Inc.-5000 editorial-services company; the partnership is now in its 7th year. Pagepro has also looked after the Localcoin site since its 2021 migration, and that site passes Core Web Vitals on desktop across 1,500+ URLs in Chrome UX Report data from August 2026.
The trade-off is timing against control. Waiting for Next.js 17 means one big change on the framework’s calendar. Moving now, on a 16.4 release that already ships React 19.3 with stable View Transitions, costs a few focused sprints, but every route you convert is one you understand, and that knowledge is worth more than the flags themselves.