Vercel vs Cloudflare: The Honest Case for Switching

Let me start where I should: Vercel is great. The developer experience is best-in-class. You connect a git repo, push, and your site is live with preview deploys, automatic HTTPS, and sensible defaults. For a lot of people it is the single easiest way to get a modern app on the internet, and I'd never tell someone they were wrong to start there.

There's a specific reason so many people do start there. When you open an AI tool and say "I want to make an app," it very often reaches for Vercel by default. It's the path of least resistance, the one in all the tutorials. That default is doing a lot of work. It's not that everyone evaluated the options and chose Vercel. It's that Vercel is what the tooling hands you first.

I just don't think it's the right long-term home for most of what I build. Here's the measured case. Not a takedown, just my reasons.

Reason 1: usage-based pricing that can surprise you

Vercel has a free Hobby plan (for non-commercial use) and a Pro plan that starts at $20 per user a month. The base price is fine. The thing to understand is what sits on top of it. Vercel's model is usage-based. You pay for bandwidth, function invocations, edge requests, and image optimization as you consume them. Pro includes a $20 usage credit. Past that, the meters run. Bandwidth at roughly $0.15 a GB over the included tier, edge requests at about $2 per million, and so on.

None of that is unfair. It's a perfectly reasonable way to price infrastructure. But here's the structural concern. It's fine until you get traffic. A quiet site costs almost nothing. A site that has a good week (a launch, a post that travels, an unexpected spike) runs those meters, and the bill is harder to predict than a flat plan. The pricing is structured so that success costs more, and that's exactly the moment a founder least wants a surprise.

Reason 2: the walled garden

The second reason is lock-in, and it's the one I weigh more heavily. A lot of Vercel's best features are proprietary platform features, and many of them are oriented around Next.js, the framework Vercel makes. Image optimization, edge middleware, incremental static regeneration, their KV and Postgres add-ons. These are excellent, and they're also Vercel's. The more you build on them, the more your app assumes Vercel exists.

To be fair, every platform has some of this, Cloudflare included. And lock-in isn't automatically bad. Sometimes a proprietary feature is so good it's worth the trade. But it is a cost, and it's one you pay later, quietly, the day you want to leave. Building on a walled garden means leaving has switching costs. I'd rather keep that cost low from the start.

The case for Cloudflare

Here's why I land on Cloudflare instead. It's not that it's trendy. It's that the fundamentals line up with how I want to build.

• ~20%of the web runs on Cloudflare
• $0free tier · real one
• 100kfree requests / day
• $5/mo if you outgrow free

Three things make the difference:

• A truly generous free tier. $0, roughly 100,000 requests a day, free SSL, and a free custom domain. Most personal and small-business sites never leave it. If you do outgrow it, the next step is a flat $5 a month, not a meter.
• You get the whole network and the primitives in one place. Workers (compute), Pages (sites), D1 (a database), R2 (file storage), and KV (key-value storage), all in one account. Instead of renting slices from a host and then bolting on a database vendor and a storage vendor, it's already there.
• More open, far cheaper at scale. Cloudflare runs roughly a fifth of the entire web on a global edge network, so it's fast and stays up. And because the pricing stays flatter, the day your site gets popular is a good day, not a billing event.

Vercel vs Cloudflare, side by side

The fair comparison. Vercel wins on first-step polish. Cloudflare wins on cost predictability, openness, and having the whole toolkit in one account.

Notice what I'm not claiming. That Vercel is bad, or that nobody should use it. If your team lives in Next.js and the developer experience is your top priority, Vercel earns its price. My argument is narrower. For a founder who wants predictable cost, an honest free tier, and one platform that does the whole job, Cloudflare is the better home.

What the bill actually feels like

This is the difference in one picture. It's conceptual, illustrative shapes, not your real invoice, but the shape is the honest part. A flat plan stays flat as traffic grows. A usage-based plan rises with it.

How I'd move a site over

If you're on Vercel and curious, you don't have to guess. The cleanest first move is to have your AI audit what you're actually using. That tells you how much is portable and how much is Vercel-specific before you commit to anything.

I have a site on Vercel. Help me move it to Cloudflare.
1. Audit my project and list what I'm actually using (framework, hosting features, data, storage, images).
2. Flag anything that is Vercel-specific (for example Vercel image optimization, edge middleware, ISR, Vercel KV or Postgres) and tell me the Cloudflare equivalent for each.
3. Give me a step-by-step migration plan in plain English, ordered safest-first, and call out anything that's likely to break so I can decide before we start.
Don't change anything yet. Just give me the plan and the risks.


End by asking me one question: "Which parts of this feel most uncertain — want me to explain them, or should I just build it for you?"

For the bigger picture on the stack I land on after a move like this (why Astro for the site itself, how to build the whole thing with AI, and how to measure it with PostHog), the companion pieces go deeper. If you're weighing a move and want a second opinion, reach out. I'm always happy to help.

The bottom line

Vercel is the easy first step, and the default a lot of AI tools hand you. That's a real strength, not a knock. But for what I build, two things tip it. Usage-based pricing that gets less predictable exactly when things go well, and a feature set that quietly assumes you'll stay. Cloudflare gives me a real free tier, the whole toolkit in one account, more openness, and a bill that doesn't punish a good week. That's the honest case, and it's why, when two engineers I trust both said one word, I'd already moved.