Cloud hosting comparison posts in 2026 are mostly written by affiliate marketers with one preferred provider and a referral link. This is the version after running production sites on Vercel, Netlify, Cloudflare Pages, Render, and Fly.io across the last two years — including the personal site you are reading this on (Netlify), the WordPress Stack Advisor (Vercel), HostList at 91,000 pages (Vercel), and a handful of client builds across the rest. Five platforms, real production economics, no affiliate disclosures.
The framing question is the same as for any infrastructure choice in 2026: what is your build doing, and which platform's primitives match? The answer is rarely the one with the loudest marketing. Each of these platforms genuinely wins for a specific shape of project; picking by feature checklist instead of by primitive-fit is how teams end up overpaying for capabilities they do not use.
The five platforms in 60 seconds
- Vercel — purpose-built for Next.js, exceptional DX, Edge Functions, ISR, image optimisation, the Pro BAA at $350/month. The default for Next.js. Pro plan $20/seat/month, Enterprise around $45K/year median.
- Netlify — Jamstack pioneer, broad framework support, build-and-deploy workflow, Edge Functions on Deno. Strong free tier, Pro $19/seat, Enterprise custom. The default for Astro and Hugo at small-to-medium scale.
- Cloudflare Pages / Workers — edge-first, generous free tier, Workers Paid $5/month plus bandwidth. The cheapest at scale, the most powerful primitives, the steepest DX learning curve.
- Render — full-stack hosting with Postgres, web services, static sites, background workers. Static free, Web Service from $7/month. The right choice when you need a database and a runtime alongside the static site.
- Fly.io — pay-per-resource on a global edge network, Docker-shaped, $5+/month minimum. Right for teams that want full control over the runtime and are comfortable in Docker.
Where each platform actually wins
Vercel: Next.js production, healthcare apps, Server Actions
Vercel is purpose-built for Next.js. Image optimisation, ISR, on-demand revalidation, Server Actions, Edge Functions all work with zero configuration. The 2025 HIPAA BAA add-on at $350/month on Pro changed the equation for healthcare apps — covered in the HIPAA cluster. The DX gap to other platforms for Next.js is real and has not closed. Most teams shipping Next.js in production end up on Vercel, even those who started on Cloudflare or Netlify.
- Wins on: Next.js apps, healthcare with HIPAA, AI products needing Edge Functions, teams that prioritise DX over cost.
- Falls short on: cost at scale (build minutes and bandwidth get expensive past 1M monthly visitors), framework choice beyond Next.js (works but is not optimised), self-hosted requirements.
Netlify: Astro at small scale, Jamstack diversity, marketing sites
Netlify supports more frameworks than Vercel — Astro, Hugo, Eleventy, Next.js, Nuxt, Remix, SvelteKit, Gatsby, raw HTML. The free tier is genuinely generous (300 build minutes per month, 100GB bandwidth) which is why most personal sites and small marketing sites end up here. This site (gautamkhorana.com) runs on Netlify. The build pipeline integrates with form handling, identity, and serverless functions out of the box.
- Wins on: Astro at small-to-medium scale, marketing sites, Jamstack pluralism, the friendly free tier.
- Falls short on: Next.js parity with Vercel (works but the integration is shallower), build minutes at scale (Pro tier of 1,000 minutes goes fast on a 5K-page Astro site), enterprise procurement story.
Cloudflare Pages / Workers: edge at scale, programmatic SEO, low cost
Cloudflare's pitch is global edge plus the lowest cost at scale. Workers run at 250+ data centers, KV storage is cheap, R2 object storage has zero egress fees. For a 50,000-page programmatic SEO site, Cloudflare's economics are dramatically better than Vercel or Netlify — the same workload that costs $300+ per month on Vercel can run for $30-60 on Cloudflare. The DX is rougher: deployments are CLI-shaped, Pages has fewer integrations, and debugging Workers is harder than debugging Edge Functions on the more polished platforms.
- Wins on: programmatic SEO at scale, edge-first architectures, cost-sensitive workloads, R2 as a Cloudinary alternative.
- Falls short on: Next.js (works via the Cloudflare adapter but is the most-likely-to-break path), DX polish, integrations with the rest of the modern Jamstack ecosystem.
Render: full-stack apps with Postgres and runtime
Render is the right answer when your project needs a database, a long-running runtime, and a static site all in one place. They host Postgres natively (cheaper than Supabase Pro at small scale), Web Services for the runtime, Static Sites for the front end. Pricing is transparent and predictable — no build-minute surprises like Vercel or Netlify. The trade-off is that Render is not as polished as the Jamstack-native platforms for static deployments.
- Wins on: full-stack apps with Postgres, predictable pricing, simpler than self-hosted on AWS, ~$25-60/month for a typical small app.
- Falls short on: Next.js DX (works but feels like a generic Node platform), edge runtime story (no built-in edge), brand recognition for procurement.
Fly.io: Docker-shaped global deployment
Fly.io runs your Docker container on their global network with auto-deployment, regional failover, and pay-per-resource pricing. The right call for teams that want full control over the runtime — Bun, Deno, custom Node versions, GPU workloads — and are comfortable writing a Dockerfile. The DX is between Render and self-hosted-on-AWS in terms of polish.
- Wins on: full runtime control, GPU workloads, multi-region without enterprise pricing, Docker-comfortable teams.
- Falls short on: friendly DX for non-Docker teams, build pipeline (you ship containers, not git pushes), free tier (no static-site free tier like the others).
The real cost economics for a 5,000-page content site
Anchored to 2026 prices for a hypothetical 5,000-page content site with image-heavy pages, 100K monthly visitors, 5-person editor team, weekly deploys.
- Vercel Pro: $20/seat × 5 = $100/month base. Build minutes typically run 800-1,200 of the included 6,000 minutes — fine. Image optimisation eats bandwidth — typically $20-50/month additional at this scale. Annual: ~$1,500-2,000 platform layer.
- Netlify Pro: $19/seat × 5 = $95/month base. Build minutes are tighter (1,000 included on Pro) — at 5,000 pages with image optimisation you can hit the cap. Bandwidth typically $20-40/month additional. Annual: ~$1,400-1,800.
- Cloudflare Pages + R2: ~$5/month for Workers Paid + ~$10-30/month for R2 storage and Workers requests. Annual: ~$200-450 platform layer. The cheapest at this scale.
- Render: Static Site free + ~$25/month for a small Postgres + ~$25/month for a Web Service if you have one. Annual: ~$600. Cheaper than Vercel/Netlify if you do not need their build features.
- Fly.io: pay-per-resource, typically $25-75/month for a similar workload depending on regions and runtime. Annual: ~$300-900.
At larger scale (1M+ monthly visitors, 50K+ pages), the economics shift dramatically. Vercel/Netlify become $500+/month easily; Cloudflare stays under $200; Render's predictability becomes more attractive than Vercel's metered pricing.
Decision tree — pick by what you ship
You're shipping Next.js production with Server Actions, ISR, healthcare/HIPAA, or AI features
Vercel. The DX gap on Next.js is real, the Edge Functions are cleaner than Cloudflare Workers for this shape, and the Pro BAA at $350/month opened up the healthcare path. Cost is the trade-off; for most production Next.js apps the DX win pays for it.
You're shipping Astro, Hugo, Eleventy, or a static-first site at small to medium scale
Netlify or Cloudflare Pages. Netlify if the team values DX and free tier; Cloudflare if cost matters and the team can absorb the rougher tooling. For an Astro marketing site under 5,000 pages, Netlify's free tier carries you to launch.
You're shipping programmatic SEO at 25,000+ pages
Cloudflare Pages or Workers. The economics at scale are dramatically better. HostList at 91,000 pages runs on Vercel today; if I were starting that project in 2026 I would seriously evaluate Cloudflare Workers with R2 for the asset layer. The cost gap at that scale is roughly 5-10x in Cloudflare's favour.
You need full-stack with Postgres + runtime + static site in one place
Render. The integrated Postgres is the differentiator versus Vercel/Netlify. Predictable pricing, no metered surprises, easier than self-hosted on AWS for the kind of team that does not want to manage infrastructure.
You need full runtime control, GPU, or multi-region without enterprise pricing
Fly.io. Docker-shaped, pay-per-resource, global edge. The right call for teams that have specific runtime requirements and are comfortable writing Dockerfiles.
FAQ
Is Vercel worth the price for Next.js?
For most production Next.js apps, yes. The DX gap on Vercel for Next.js is real — Server Actions, ISR, image optimisation, Edge Functions all work with zero config. Other platforms support Next.js via adapters but the integration is shallower. The Pro plan at $20/seat is competitive with Netlify; the cost only escalates at high traffic and high build-minute usage. For solo developers and small teams shipping Next.js, Vercel is the default for good reason.
Is Cloudflare Pages better than Netlify?
For cost-sensitive workloads and edge-first architectures, yes. For DX and ecosystem polish, Netlify wins. The choice depends on the brief: a marketing site under 1K pages should probably stay on Netlify or Vercel; a programmatic SEO site at 25K+ pages should migrate to Cloudflare for the economics. Most production teams end up on whichever platform their first project landed on; switching mid-product is rarely worth the disruption.
Can I self-host Next.js to avoid Vercel?
Yes, on AWS, Cloudflare Workers, Render, or Fly.io. The DX hit is meaningful — you lose the ISR-with-zero-config and the image-optimisation pipeline that Vercel provides. Self-hosting makes sense for teams with strict data-residency requirements (some EU public sector contracts), enterprises that already have AWS architecture, or teams shipping at scale where Vercel's costs are a real line item. For most small and mid-sized production teams the DX cost of self-hosting outweighs the platform savings.
What about Render vs Railway vs Fly.io?
Render is the most mainstream of the three with the simplest DX. Railway is in the same shape — pay-as-you-go, web services, databases — with slightly different pricing and a more developer-friendly UI. Fly.io is the most flexible (Docker, global edge, multi-region failover) but the steepest learning curve. For a small team that wants 'just deploy this app and the database', Render is the default; Railway is the cheaper alternative; Fly.io is the right choice when you outgrow either.
Does Vercel still have the best HIPAA story for Next.js?
Yes — the Pro BAA at $350/month on Vercel is the cleanest HIPAA path for Next.js as of mid-2026. Cloudflare and Netlify do not have published BAAs at the Pro tier; AWS, Azure, GCP all have HIPAA-eligible infrastructure but require self-hosting Next.js with the associated DX cost. The HIPAA-compliant Supabase + Vercel setup covers the full $700/month combined platform layer including Supabase Team plus the HIPAA add-on.
Related reading
Next.js vs Remix vs Astro in 2026 — the framework choice that determines half of the platform decision.
Web Frameworks Hub — the framework decision tree, with platform recommendations per framework.
HIPAA-compliant Supabase + Vercel: the $700/month setup — if your platform decision involves healthcare compliance.
How I built a 25,000-page directory in Next.js — the production case study at scale, useful when the platform decision is shaped by programmatic SEO.
The platform pick is rarely the bottleneck. The bottleneck is whether the platform's primitives match what your team actually builds. Pick by primitive-fit, not by feature checklist.
Book a 30-minute platform pick call — describe the framework, the workload, the team, the procurement constraints. Walk away with a Vercel-vs-Netlify-vs-Cloudflare-vs-Render decision that survives both the engineering review and the cost projection.