Choosing a stack in 2026: React, Next.js, or Astro for your next project
We pick a stack based on three criteria: project fit, total cost of ownership, and your team's long-term ability to maintain it. Here's how that plays out in practice.
Stack choice is a bet on the next five years. The hot stack today is often the legacy cost tomorrow. We pick by three criteria: project fit, total cost of ownership, and your team’s ability to maintain what we ship. Here is how that plays out for the three stacks we reach for most often.
Astro
Picks for: content-heavy marketing sites, docs, blogs, sites where SEO and Core Web Vitals matter more than client-side interactivity.
Why: near-zero JS by default. Partial hydration only where interactivity earns its cost. First-class MDX. Built-in image optimization, sitemap, i18n. Static by default; server routes where needed.
Avoid when: the site is actually an app with stateful UI on every page. Use Next for that.
Next.js
Picks for: dynamic apps. Auth, per-request server-side data, middleware, ISR, streaming SSR. SaaS products. Internal tools that pretend to be marketing sites.
Why: the most mature full-stack React framework. Vercel’s performance infra is genuine. App Router is powerful once you accept its mental model.
Avoid when: the project is content-first. You pay for flexibility you don’t need.
Plain React + a backend
Picks for: dashboards, back-office tools, embedded apps, anything where SEO is a non-goal.
Why: less magic, explicit routing, smallest vendor surface. Easier to hand over to a team that wants to own the deploy pipeline.
Avoid when: content matters for SEO. Static site generation is not what you’re optimizing for.
The question we actually ask
What does the next team member — who you haven’t hired yet — need to be able to do? If the answer is “edit a page of content,” Astro. If it is “add a feature to a dashboard,” Next or plain React. The stack should favor the work you expect to do most.