Before the pitchforks come out: Next.js is not bad. It's an incredible framework.
But too many solo devs and small teams choose it by default, then fight against it for months.
The caching mental model
Next.js 13+ introduced Server Components, Client Components, streaming, multiple caching layers.
You need a mental flowchart just to understand why your data isn't updating.
I've spent entire afternoons debugging a simple database update not reflecting in the UI. The answer was usually some caching layer I forgot existed.
The deployment reality
Yes, you can self-host Next.js. Technically.
But there's a gap between "it runs" and "it runs optimally." Features like ISR, image optimization, and edge functions work differently outside Vercel.
What I switched to
React Router v7 (formerly Remix). The difference was immediate:
- No caching by default. I add it when I need it.
- Standard Web APIs. Request, Response, FormData.
- Explicit data flow. When something breaks, the stack trace points to my code.
When Next.js makes sense
- Content-heavy sites with complex caching needs
- Teams with dedicated DevOps
- Projects committed to Vercel ecosystem
When it might not
- Solo devs who debug everything themselves
- MVPs where iteration speed beats optimization
- Projects where deployment flexibility matters
The boring choice isn't the wrong choice.
Anyone else made a similar switch? What was your experience?
Top comments (2)
I shipped with early Remix about 2+ years ago and it was excellent for speed-to-market.
So this is not anti-framework for me. I just think tool choice should match project stage and team context.
For some projects, a smaller setup (github.com/ije/mono-jsx, github.com/melendezgg/microvibe), with clearer internals is a better starting point, especially for first versions where installing a full batteries-included stack feels heavy when you just need a working web app sketch.
Totally agree - tool choice should match the context. For quick sketches and experiments, minimal setups win.
Where I see frameworks like Remix/RR7 shine is when you know you'll need auth, payments, emails, etc. from day one. Starting minimal then adding those later often costs more time than starting with them baked in.
But yeah, for a pure prototype or proof of concept, lighter is faster.