Web development trends 2024: the ones that mattered, the ones that didn't
A year-end look at the web development trends of 2024: React Server Components going mainstream, edge runtimes maturing, AI in the IDE, and which trends quietly died.
The trends of 2024 we'd bet on again
Trend posts age badly because most trends die in 18 months. The interesting test is which ones we're still betting on six months later — and which ones quietly disappeared from production codebases.
1. React Server Components went mainstream
2024 was the year RSC stopped being “the new thing” and started being the default in serious codebases. Streaming SSR, server-only data fetching, and the disappearance of the over-hydrated SPA shell shipped real performance wins. Our web platform engagements are RSC-first by default now.
2. Edge runtimes earned their keep — for the right shape of work
Cloudflare Workers, Vercel Edge, and Deno Deploy matured into something teams use without fear. The trend isn't “everything at the edge”; it's “the right things at the edge.” Read paths, geo-aware routing, and lightweight personalization win. State and writes stay in the regional database.
3. Type-safe full-stack stopped being optional
End-to-end TypeScript with tools like tRPC, Drizzle, and TanStack Query went from “nice to have” to “table stakes.” The cost of typing once and reusing types across server, client, and database has become so low that not doing it looks reckless.
4. AI in the IDE rewrote the keyboard shortcuts of the day
Cursor, Copilot, and Claude in the editor changed how senior engineers spend their time — less typing, more reviewing. The teams who adopted this consciously shipped faster; the teams who pretended nothing happened didn't. Either way, the productivity bar moved.
5. Tailwind + Radix + shadcn became the design-system default
The trio of utility CSS, headless primitives, and a copy-paste component library became the design-system default for product engineering teams that don't have a 20-person design ops team. Reasonable defaults, total override-ability, no runtime tax.
Trends that quietly died in 2024
- CSS-in-JS as the default: the runtime tax stopped being worth it once Tailwind matured.
- Microfrontends as a small-team strategy: still a real pattern at very large orgs, but the “every team owns its slice” pitch fizzled where it didn't fit.
- NoSQL-by-default for new SaaS: see Firebase vs Supabase — Postgres is back to being the right answer for relational data.
- Webpack-as-bundler-default: Turbopack, Vite, and Bun's bundler ate most of the new-project mindshare.
What we're carrying into the next year
Our default new-project stack now: Next.js + TypeScript + Postgres + Tailwind + shadcn, deployed to Vercel, with selective edge usage for read-heavy paths. It's not novel; it's productive. Most of our recent engagements ship on something close to this.
How we'd help
If you're picking a stack for a new product or modernizing an old one, our Web Platforms service and Modernize your legacy system solution cover both ends of that spectrum. Tell us where you are and we'll come back with a stack recommendation we'd defend in a board meeting.