Build1 distinct publisher2 min readUpdated
Vite fixes the build times and the config cliff that retired Create React App, and keeps the client-only rendering. Next.js fixes that, and charges for it.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
Both starter commands are one line long and they buy very different amounts of building. `npm create vite@latest my-app -- --template react` hands over a build tool that leans on the browser's own ES module support and compiles only the code you are currently looking at, so the dev server starts almost instantly no matter how large the project gets [5][11]. What it does not hand over is a router or a server. Those stay with you, and the dev.to piece names React Router as the sort of thing you now have to choose yourself [6]. `npx create-next-app@latest` hands over routing, data fetching and rendering with the decisions already made: routes are files, and any given page can be rendered on the server, generated at build time, or left to the client [8][10][11].
Line up the faults that piece pins on Create React App and the arithmetic is unflattering to the easy answer. Webpack-era dev server times [2] and the one-way `eject` cliff [3] are both gone under Vite, whose config file the author calls simpler than what it replaces [14]. The third fault, shipping nothing but a client-rendered SPA [4], survives intact: Vite output is still an SPA that, by the same author's account, is harder for search engines to index without extra work [12]. Two of three [1]. So the description of Vite as CRA's spiritual successor is more literal than it was probably meant to be [13].
That is what makes the choice an architecture call rather than a taste call. Picking Vite is a bet that markup assembled in the browser is acceptable for what you are building, which is the case the piece makes for internal tools and simple dashboards, where it says Next.js is more architecture than the job needs [9]. Picking Next.js buys the rendering flexibility and bills you for it in learning curve: the caching, data fetching and routing conventions are the framework's, not plain React's [9].
Worth remembering which of the two is actually a React project. Vite is framework agnostic, serving Vue, Svelte and vanilla JavaScript as happily as React [7], and that neutrality is precisely why the React-shaped questions stay open in a Vite app. The old command answered them by default, badly and invisibly. Nothing answers them now except a decision, taken by whoever runs the installer, usually before anyone has decided whether the product needs to be indexed.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
The React core team officially removed Create React App from their recommended starting guides.
CRA relied on Webpack, and as a project grew, spinning up the local development server could take a long time.
Tweaking CRA's underlying Webpack configuration required "eject", a one-way operation that exposed large configuration files.
CRA built only client-side rendered single page applications and did not natively support server-side rendering or static site generation.
Vite, created by Evan You (creator of Vue.js), uses native ES modules in the browser and compiles only the code currently being viewed, so the dev server starts almost instantly regardless of project size, with snappy hot module replacement.
Vite is strictly a frontend build tool; developers must supply their own backend and routing solution, such as React Router.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single uncorroborated community post
Everything rests on one dev.to article by one author. The descriptive claims (CRA removed from React's guides, Vite's native-ESM dev server, Next.js rendering modes and file-based routing, the two scaffold commands) are internally consistent and checkable in principle, but the cluster contains no primary documentation, changelog, benchmark, or second publisher. The evaluative claims — dev-server speed, config simplicity, SEO indexing difficulty, learning curve — carry no measurements at all.
No adoption evidence supplied
The cluster contains no releases, version numbers, download or usage figures, deployment disclosures, survey data, or migration reports. Statements that the community 'coalesced around' Vite and Next.js are unquantified narrative, so no adoption level can be measured without inventing facts.
Rhetoric ahead of the supplied proof
The framing ('Create React App is Dead', 'blazing fast', 'true spiritual successor', dev servers taking minutes) is stronger than what the piece actually demonstrates, which is zero measurements. The gap is moderate rather than severe because the underlying factual spine — CRA leaving React's recommended guides, and the real architectural split between a client-only build tool and a rendering framework — is plausible and consistently argued, and the article does state Vite's tradeoffs and Next.js's costs rather than only selling.
Low disclosed pressure, vendor-shaped subject
The cluster shows no sponsorship, affiliate arrangement, or employer relationship for the author, and the post is self-published developer commentary rather than vendor marketing. The one incentive fact actually in evidence is that one of the two recommended options is built and maintained by a commercial vendor, Vercel, whose hosting business benefits from Next.js becoming the default starter; the article discloses the ownership but not that consequence. Nothing in the supplied material establishes a stronger incentive than that.
Direction credible, specifics unverified
Confidence is moderate-low: one publisher, no primary corroboration, and no adoption dimension to anchor against. The structural read — CRA out of the recommended guides, and the successor choice turning on client-only versus server/static rendering — is coherent and consistent with the article's own internal logic, which supports the descriptive claims; the evaluative and superlative material should not be relied on without a second source.
build
Five frameworks, one store, and a benchmark that measures when the button works1 distinct publisher
build
Next.js's architecture is working. Its operating manual isn't keeping up.1 distinct publisher
build
Flue 2 bets that agents are a rendering problem, not an orchestration one1 distinct publisher
build
Next.js 16.3's memory claim didn't reproduce; its TypeScript handoff cut a build by two thirds1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 21, 2026