Next.js 16 ships Cache Components as a first-class caching model: a single 'use cache' directive, paired with cacheLife , that you can drop on a route, a component, or a plain data-fetching function. The pitch is appealing: pick your own granularity, mix static and dynamic content on the same page, and let Partial Prerendering figure out the rest.
That pitch also raises a question the docs don't spend much time on: does it actually matter where you put 'use cache' ? If you wrap an entire component versus just the fetch call inside it, what changes, for the reader and for the server doing the work?
To find out, I built a small dashboard with three crypto price cards (Bitcoin, Ethereum, Solana), all pulling the same kind of data from the same public API, each caching it a different way:
Bitcoin caches the fetch and the render together, at the component level.
Ethereum caches only the fetch; the chart underneath is recomputed on every request.
Solana caches nothing at all.
All three run under the exactβ¦