NuxtLabs is joining Vercel | Hacker News Reader

NuxtLabs is joining Vercel

nuxtlabs.com

76 pointsblinky8879 comments

Pretty incredible that now Next.js, Svelte/kit and Nuxt will be under Vercel.

This could be a great thing as now all of these devs are much better supported in their work, but I’d be lying if I didn’t say that this situation makes me quite nervous.

Embrace, extend, extinguish…

Tale as old as… well, the 60’s… but still, it’s an old tale.

I've been trying to get embraced, extended, and extinguished for years.

Could you expand on this a little? Specifically, how do you think "Embrace, extend, extinguish…" could apply to this situation?

Vercel's playbook with Next so far has been to make convoluted features that exist solely to pad out how much people spend on hosting costs. They also make sure that hosting it anywhere but Vercel comes with footguns, even though theoretically you can host your Next app anywhere you want (and it's gotten better recently solely because of backlash). See https://opennext.js.org/ for example.

They've been so successful with this in fact, that when you check the actual official React documentation, Next is the offering being advertised prominently [1] (I see they've actually restructured this a bit, a few months ago Next was literally the only thing mentioned here, I assume because of backlash they changed it to what it is now).

I have no doubts they'll do the same with Nuxt, demanding they implement features that solely exist to pad out hosting costs while providing next to no actual benefits to end users and devs, and I have no doubts they'll have the same shenanigans as they do with Next where hosting on anything other than Vercel gives you a broken experience. As an actual Extinguish tactic, I can even see them fully shuttering Nuxt in favor of Next. After all, they now have a monopoly over the most popular frontend meta frameworks (Next, Svelte/Kit, Nuxt, Astro) and paying people to maintain all of that is a lot more expensive than just having the 1 (that already has a huge market share to begin with). Thankfully for now it's MIT and fully OSS, so there's a ray of hope here that even if they pull some shenanigans like that, it'll continue living on independently again, but I don't trust anything that has been touched by VC money as far as I can throw it.

https://react.dev/learn/creating-a-react-app

This makes a lot of sense. Thanks for the explanation. I also remember being a bit shocked when I was learning React and was immediately told to use Next.js with it.

> I have no doubts they'll do the same with Nuxt, demanding they implement features that solely exist to pad out hosting costs while providing next to no actual benefits to end users and devs

has that happened with Svelte?

> they now have a monopoly over the most popular frontend meta frameworks (Next, Svelte/Kit, Nuxt, Astro)

this would be a surprise to the Astro team!

> has that happened with Svelte?

Yes. There are features that have been deliberately kept on the platform level to force you to use the platform instead of implementing it on the framework level or guiding you through it with the docs.

https://svelte.dev/docs/kit/glossary#ISR

ISR _can't_ be implemented at a framework level without tying the framework to the platform. The fact that we instead chose to implement it via a platform-agnostic adapter API surely demonstrates the opposite of what you're implying

It's platform agnostic but Vercel is the only one that supports it? At least that's in the docs I linked.

This would be true if you also provided an adapter-node, which would work on most traditional servers, and most non-serverless platforms.

The fact that you prioritize the vercel one and not the node native one proves what OP implies in my opinion.

> this would be a surprise to the Astro team!

That's my bad, seems my memory failed me here!

Here's to holding out hope Tanstack stays independent

Don’t bet on it. Nobody can resist a big pile of money. Everyone has a price.

That’s awful. Vercel has pretty much a monopoly on the hybrid framework space now, and it’s going to ruin everything. NEVER did a monopoly do any good to anyone but its holder.

Props to Vercel, I guess. Enjoy your champagne. Increased the share holder value, job well done! Clap, clap.

Moat's not deep, this doesn't make it much better.

That shallow moat sure made them a few millions richer than you or me.

Nuxt maintainer here.

there's lots to say here, but from my point of view, Vercel's backing Nuxt largely _because_ of our open vision.

our open approach isn't an optional extra. it's a core value we all share on the team - and indeed, I think, is as close to a core value of the web as I know.

we've pioneered cross-framework adapters and the provider pattern in all we build and there is no way we are changing direction or vision.

nuxt remains an independent framework, like svelte. the fact that a number of us on the team are employed to work full time on OSS is _great_ news for OSS sustainability.

Huge fan of Nuxt and your work, excited for this!

I hope so! The Nuxt adapters and provider pattern are amazing! Its truly impressive how well done it all is and how portable Nuxt is and how other projects have been able to use a lot of the same OSS tech you've created to get portability from the jump.

The portability story for Vercel's own Nextjs is a disaster.

That's nice in theory, but what happens when the inevitable comes and Vercel's grubby little VC hands start squeezing their yoke in order to extract money out of the ecosystem? They're on their Series E last time I checked, that's a lot of VCs who are going to want to cash in on things sooner rather than later.

Enshittification is inevitable when VC is involved in any way whatsoever, so this doesn't strike me as a good thing. I can already see the future where we're getting convoluted features no one ever asked for for the sole reason of inflating Vercel's hosting costs, as was and is the case with Next and how they completely took over React to the point that even the official React docs mention Next before any alternative like Vite. Hell, knowing the VC playbook, I wouldn't be surprised if in a year's time they decide to shutter Nuxt completely and force everyone to move into the abomination that is Next.

This isn't a swipe against you guys, I'm thrilled to see OS devs like you who truly deserve all the success in the world get that success, I'm just not convinced that a VC-funded company with a dubious history & track record monopolizing the entire frontend framework landscape is a good thing. Thankfully Evan seems to still be independent and as such Vue will continue being independent, but it's a bit worrying that it's basically the only one.

the best guarantee I have is that Nuxt is still independent - it's (some of) the core team who are employed by Vercel. you can see this is the case for svelte/sveltekit.

plus, as an open source project, no one _can_ shutter Nuxt. as long as we have a community, we are alive.

from another point of view, any open source project is ultimately accountable to the community, and I think that's especially true for nuxt.

at the end of the day, if sebastien isn't happy with my leadership he can replace me. (and that's _good_ - I wrote about it here: https://roe.dev/blog/governance-and-abuse)

and equally if the community aren't happy with the direction of nuxt, they can always fork it. (and that's _good_)

in fact, I think I'd say that accountability is always good.

"Vue is currently the only mainstream framework that remains independent. (i.e. not dominantly owned / backed by a single corporation" Evan You https://x.com/jpschroeder/status/1762764818254016513/photo/1

I know Nuxt is not Vue itself, and I'm not saying Vue is no longer independent — but I do think it's worth remembering that independence is something highly valued in the Vue community.

Wouldn’t Void 0 contradict that statement? It’s more oriented towards lower-level infrastructure though (Vite, Oxc and other tooling), but it extending to support Vue wouldn’t be a surprising move if it became necessary.

As someone that's been saying Vercel should be avoided for years now, it's nice to finally start seeing more hesitation from others.

They now, to varying degrees, directly employ core maintainers for Svelte, SvelteKit, React, Next, and now Nuxt. This is a very clear systematic overtaking of the web ecosystem. They're a private business, so these moves must be in the interest of increasing profits. It's not just out of the goodness of their hearts.

It's somewhat unfortunate that technical and business-savvy people would both, in my experience, disregard a study saying tobacco is good if it's funded by RJR, and be excited about a giant tech company employing core maintainers for the majority of new web-related software projects. Yes, they're open source projects that you can fork. But if Vercel has influence in the direction of these projects (and of course they do) it should give people much more hesitation to use them than it seems to.

At this point, using any of the technologies that Vercel has its hands in tells me that whatever uses it - a business, project, whatever - doesn't plan to function in five years.

Given that Vercel was founded 10 years ago, and they sell services for money, and then take that money and use that to pay people to work on various technologies; why would that make it less likely to be around in 5 years vs a person who isn't getting paid and is doing an open source project with no visual means of support? How do their bills get paid?

> How do their bills get paid?

With Vercel hiring all framework creators, we will never know. Maybe that is what it will take to get hired by tech companies next: create a world-famous open source framework.

I'm not saying Vercel won't exist, they almost certainly will.

What I mean is that the goal seems likely to be to influence how software development is done in every way possible, from deployment to client code. Vendor locking, with even more control. I'd pitch Vercel as "Heroku's growth plan but more control of everything." Maybe with a bit of EEE thrown in, but who knows.

Where's the "even more control" though? A proprietary closed source version of the same would be the other way to do it, but then it would be closed source.

If we want open source to be viable we have to support actually having businesses around it. Vercel making it easy to deploy to their servers seems like a fairly decent business model compared to some of the other options.

I would love if Vercel paid all the people in the teams I mentioned earlier, told them to do whatever they want, and checked in only to make sure they had enough snacks. But that's probably not the case. Vercel employs these people. To some degree, simplistically, the employees are there to further the goals of the company.

I don't claim to know their plans. I've never been in charge of a multibillion dollar company. I just think I have a vague idea of what their general strategy is, and I don't love it.

I'll also say that I definitely want open source projects to succeed. I don't know how they can in a great way in a capitalist system. So maybe this is, from my standpoint, the best of a bad situation. I still think it's worth pointing out and paying attention to from a free software and business position.

Also just want to say thanks for what seems to be a genuine discussion in good faith.

Some more details by Daniel Roe: https://github.com/nuxt/nuxt/discussions/32559

As also seen here: https://news.ycombinator.com/item?id=44500544

Vercel’s announcement: https://news.ycombinator.com/item?id=44499838

This is like Autodesk having 3D Studio Max, buying up Maya and Softimage, leaving Houdini as the only independent 3d package.

blender.org has entered the chat.

I'll admit that I'm skeptical and that I'd generally prefer less centralization. I love what Nuxt has done with Nitro, for example, and it being compatible with "all" hosting providers. Compare that to Next, which is "best on Vercel". Also see TurboPack, which seems to be effectively exclusive to Next.

That said, I can only imagine how incredibly freeing it must be to not have to worry about funding, so I couldn't blame them in the slightest. I really with them the best and hope Nuxt continues to be great. Looking forward to v4 soon!

A lot of comments express concern about their favorite framework losing independency. I always worry about Angular being made independent. lol

Honestly, this is terrible news. A number of the core team of Nuxt.js are joining Vercel and NuxtLabs itself is acquired by Vercel.

I choose Nuxt.js and Nuxt UI Pro specifically because they aren't Vercel products. I built two SAAS MVPs over the last year based on this, now I'll have to wait and see what Vercel (their competitor) wants to do with it.

Vercel isn't a competitor to Nuxt. Vercel benefits from allowing any frontend framework to be deployed on their service.

All you have to look at is Svelte, which Vercel hired all the developer of, and that turned out to be a great thing. Svelte and SvelteKit are better than ever and nothing they have done since has shown to be forced by Vercel.

Vercel benefits from moving as much of the logic to the server (their paid services) as possible.

Vercel benefits from not letting other, better, cheaper architectures like Cloudflare Workers get better treatment than them from popular framework authors.

That holds true now. Tomorrow, when they can steer the direction of all frameworks, they—and their investors by extension—might notice that Next projects generate the most revenue on average. Now they have an incentive to favor Next; not even by outright harming the other frameworks, but maybe by strategic neglect. Maybe a new web platform feature is first supported here, while integrations stall elsewhere; maybe the backlog is sorted differently; it doesn’t matter.

The point is that you’re just seeing business strategy aligned with our interest, but that can, and will, change.

No, I agree in principle. Letting Vercel functionally own the roadmaps for the "best" frameworks in the ecosystem is dangerous bordering on stupid.

You've expressed my thoughts to a tee. I will start exploring Astro and see what the migration process is like. Many devs questioned my usage of Nuxt over Astro to begin with, so this is a good opportunity to learn.

Astro is nice, but its a different model than Nuxt, more suited to static sites.

I'm just reading the docs and it seems more capable than that: https://docs.astro.build/en/concepts/islands/#server-islands

I have also just discovered a dedicated section for the migration from Nuxt to Astro:

>Key Similarities between Nuxt and Astro

Nuxt and Astro share some similarities that will help you migrate your project:

- Astro projects can also be SSG or SSR with page level prerendering.

- Astro uses file-based routing, and allows specially named pages to create dynamic routes.

- Astro is component-based, and your markup structure will be similar before and after your migration.

- Astro has an official integration for using Vue components.

- Astro has support for installing NPM packages, including Vue libraries. You may be able to keep some or all of your existing Vue components and dependencies.

https://docs.astro.build/en/guides/migrate-to-astro/from-nux...

It's only "more capable" in the sense that yes, ultimately, it's JavaScript and you can do a lot with JavaScript.

But Astro's model is "generate markup once, serve statically".

Nuxt's model is "Server-Side JavaScript"

From a cursory reading of Astro's docs their model/philosophy seems to be more like "fast by default."

So their philosophy is to start with zero client-side javascript, even when one is using components from javascript-heavy frameworks like React or Vue. Interactivity is apparently something one has to explicitly opt-in to by adding a client:* directive, rather than it being the default.

The migration guide states, "Astro projects can also be SSG or SSR with page level prerendering," which puts it on par with Nuxt in that regard.

A difference I've noticed regarding server-side features is that Astro allows one to "define and call backend functions with type-safety."[0] which even Nuxt doesnt offer in such a convenient and type-safe manner. I'm pretty happy with what i'm seeing in Astro's docs so far.

[0] https://docs.astro.build/en/guides/actions/

I think you're fundamentally missing the difference between these two frameworks.

I think that you should substantiate your claims instead of making low effort comments with zero elaboration. I've cited relevant passages from Astro's docs and I realized that it offers me the features I'm looking for, so I'm not sure what you're on about. Here are the "Key differences between Nuxt and Astro"[0] according to the docs:

- Nuxt is a Vue-based SPA (single-page application). Astro sites are multi-page apps built using .astro components, but can also support React, Preact, Vue.js, Svelte, SolidJS, AlpineJS, and raw HTML templating.

- Page Routing: Nuxt uses vue-router for SPA routing, and vue-meta for managing <head>. In Astro, you will create separate HTML page routes and control your page <head> directly, or in a layout component.

- content-driven: Astro was designed to showcase your content and to allow you to opt-in to interactivity only as needed. An existing Nuxt app may be built for high client-side interactivity. Astro has built-in capabilities for working with your content, such as page generation, but may require advanced Astro techniques to include items that are more challenging to replicate using .astro components, such as dashboards.

[0] https://docs.astro.build/en/guides/migrate-to-astro/from-nux...

https://nuxt.com/docs/getting-started/server

Nuxt is a server-side JavaScript framework with a complete HTTP application stack, built on top of Nitro (which includes an even larger suite of capabilities) and H3.

https://nitro.build https://v1.h3.dev

Astro is markup generation.

They are only comparable in that "both can print markup".

The idea that Astro and Nuxt are only alike because they "print markup" is factually incorrect. Astro is designed to give you the same flexibility as Nuxt when it comes to building dynamic, server-rendered sites. One can even use React or Vue within Astro. You can create API endpoints, use SSR, and even define backend functions with type-safety. The migration guide from Nuxt to Astro makes it clear that both frameworks support similar workflows, from file-based routing to dynamic data fetching. Astro’s islands architecture[0] just means you get more control over how and when interactivity is delivered, but under the hood, it’s every bit as much a server-side javascript framework as Nuxt.

[0] https://docs.astro.build/en/concepts/islands/#server-islands

If Astro is "designed to give me the same flexibility" it fails to do so in hilarious ways.

Even if you use the Astro view transitions API to make a SPA, its client side router falls short of other frameworks that are more truly SPA oriented.

Nuxt has always been a strange tool that doesn’t really fit well into any good use case, unless your only experience is Vue and Nuxt.

Hard to see the real reason for Vercel to do this. The pessimist in me wonders if perhaps they are hoping to influence how Vue is developed in the same way they now influence how React is developed after hiring several React team members.

But even that doesn’t seem that likely considering the relatively tiny Vue market share and microscopic still Nuxt market share.

I also consider its “community” to be a strange place too. It’s on Discord, and a couple of years ago common internet abbreviations were considered ban worthy rule-breaking offences.

Even the word “lmao” would get you an instant warning from a bot. The framework itself and its oddball community were enough poor experiences for me to stop using it pretty quickly.

It's an aquihire

Get Nuxt people paying to deploy on Vercel and over time force the user base to Next and React would be my guess.

> doesn’t really fit well into any good use case

I agree.

If you want an SPA just use Vue's official router. It's getting better now thanks to Eduardo working on file-based router, data loaders, Pinia colada, etc.

If you want a static site or MPA, Astro really seems like a better choice than Nuxt.

https://uvr.esm.is/

https://pinia-colada.esm.dev/

There is no universe in which Astro is a better choice than Nuxt for Vue-based MPA, SSG or SPA with SSR situations.

No universe, huh?

https://imgur.com/MxSrgj1

I have no idea what you think that graph is showing or how its relevant.

You don't think having a significantly better CWV score is relevant?

I think plain text has the best possible CWV score yet is also deficient as an alternative for Nuxt if you're dealing with Vue-based MPAs, SSG or SPAs with SSR.

Are you saying that Astro can only output plain text?

If that's the case, you're completely missing the point.

There is no universe in which I'm saying Astro can only output plain text.

I am pointing out that CWV scores are irrelevant to the discussion.

You are arguing that a Miata is valid competitor to an F150, while I'm repeatably trying to explain I need to tow a horse trailer.

> Hard to see the real reason for Vercel to do this.

Next supremacy is very obvious

Kills all those Linkedin threads about moving away from Next. Kills the indecision for what employment-seeking devs need to optimize for. Makes those job descriptions less all over the place and even more Next focused

That doesn't make any sense at all. Vercel wins either way if you choose Next, Nuxt, or Svelte. Seems obvious - why would they care if you use Next vs. Nuxt now?

> Vercel wins either way if you choose Next, Nuxt, or Svelte. Seems obvious - why would they care if you use Next vs. Nuxt now?

Because maintaining three projects is more expensive than maintaining one. It's within their best interests to quietly extinguish any alternatives in favor of the one that most closely matches their vision.

What?

Nuxt is the only "out-of-box-everything-works" framework for Vue.

So yes, if you use Vue, you use Nuxt.

I actually love Nuxt. If you want all the fullstackness of Next, but without having to deal with React, it's perfect.

The "without having to deal with React" bit is crazy important.

I also think, even if Nuxt supported React, its overall architecture is just better than Next's.

Everything just seems to plug together better.

That and the dev tools are a massive upgrade.

I'm baffled by the doom-and-gloom reactions here. Nuxt remains what it's always been: the best convention-over-configuration framework in existence. It's built on Vue which is opinionated as hell, and you get all the benefits of that. The "vendor lock-in" concerns are frankly overblown. At the end of the day Nuxt produces artifacts you can deploy anywhere - AWS, Cloudflare, your own infrastructure, or yes, Vercel. The alternatives (underfunded OSS maintainers burning out) are way worse than having a well-funded team with aligned incentives. If anything this validates that Nuxt is valuable enough for a major platform company to invest in. I'll take that over watching great tools die from lack of resources.

Everything you say is now subject to change due to strategic decisions by a single entity that owns almost all horses in the race. The things you take for granted now can disappear tomorrow, no matter how many times they pinky promise not to.

Nuxt is great, but Chromium is great too. Yet, Google has become the driving force behind changes to the web platform, for better or worse. That’s not a desirable situation, and certainly not the only one: it’s not like there’s only a single company out there able to fund open source software. I desperately hope we, collectively, will figure out a better financing model in the future.

Ugh. Can they not? Who do I talk about them not?

NuxtLabs was doing great work building out support for Cloudflare, making it a viable alternative to Vercel.

Now, I'm sure all that work will get dropped and we'll be stuck with only Vercel being a first-class host for Nuxt-based applications.

The announcement on Vercel's side: https://vercel.com/blog/nuxtlabs-joins-vercel

This is awful news unless hopefully this REALLY doesn't influcence the development into a more react-y way away from Vue.

Why would it? That doesn't really make sense?

Or having the framework adopt design patterns optimized for Vercel infra, like Next is (and why I replaced it with Tanstack)

That's a shame.

So now Nuxt joins Next in the never to use pile.

Reply on news.ycombinator.com