Hosting for FastAPI: what 15 makers' products chose
Among 15 makers' products built with FastAPI that we know the hosting of, Vercel leads: 6 of them (40%). Next come Amazon Web Services (5) and Render (2).
What shape is your app? A frontend or serverless app fits a frontend cloud. Long-running servers, background workers and databases fit a PaaS or container platform. A hyperscaler cloud has every building block but leaves you to assemble them; a platform or framework that deploys into your own cloud account sits between, with platform-style deploys and the resources on your own cloud bill.
What makers on FastAPI chose
| Option | FastAPI products | All makers | Free for | Team of 5, 10,000 users |
|---|---|---|---|---|
| 6 (40%) | 57% | None | $120/moPro + Flat Rate CDN 10M | |
| 5 (33%) | 13% | None | $17/moPay as you go (Lambda + EC2 t4g.small) | |
| 2 (13%) | 4% | None | $61/moPro | |
| 2 (13%) | 5% | — | — | |
| 2 (13%) | 3% | None | $13/moCPX12 | |
| 1 (7%) | 7% | None | $30/moPro 3,000 credits | |
| 1 (7%) | 3% | None | $45/moPro | |
| 1 (7%) | 2% | None | $15/moPay As You Go | |
| 1 (7%) | 2% | — | — | |
| 1 (7%) | 4% | — | — |
Live makers' products with a source for both: that they're built with FastAPI, and what they use for hosting. A product using two options counts in both; not market share. "All makers" is the option's share among every maker's product that made this decision.
Costs are the cheapest plan that fits, from each vendor's pricing page, at the usage our stack planner assumes for a product with that many users. Free for: the users its cheapest plan a business can use stays free for; plans for personal use only, like a non-commercial hobby tier, don't count.
Which to pick, by situation
- You built a Next.js or other frontend-heavy app and want preview URLs on every pull request
Vercel 6 here
Netlify 1 here - You have a long-running server, a background worker and a database to deploy together
Railway 1 here
Render 2 here - Your users are spread across continents and the app runs in Docker
Fly.io 1 here - You want a flat monthly bill and full control, and don't mind running one server
Hetzner Cloud 2 here
Vercel
Frontend cloud6 FastAPI products here (40%). About $120 a month for a team of 5 with 10,000 users (Pro + Flat Rate CDN 10M).
llms.txtMCPCLIIn v0Best forNext.js apps — zero-config deploys and preview URLs.
Trade-offThe Hobby plan is non-commercial; costs grow with usage.
“Our entire frontend is built using NextJS and Vercel has been an incredible platform to build upon. Allowing us to focus on shipping vs. infrastructure in our early days.”
- Pushing to Git deploys automatically with zero configuration. 4 sources
- Every commit or pull request gets a preview URL, so changes are reviewed live before release. 4 sources
- Heavy bot traffic, such as Meta's crawler, can multiply the bill with little protection. 3 sources
- Execution limits and middleware quirks need workarounds. 1 source
Can I run a Python, Go or Docker backend on Vercel, not just JavaScript? Yes. Official function runtimes cover Node.js, Bun, Python (ASGI/WSGI apps such as FastAPI, Flask and Django), Go, Ruby, Rust and Wasm, and container images pushed to Vercel's registry can run as functions (in beta). Vercel docs ↗
Status page: 21 incidents in the last 90 days, 12 major (its own page, 2026-10-19).
“Our entire frontend is built using NextJS and Vercel has been an incredible platform to build upon. Allowing us to focus on shipping vs. infrastructure in our early days.”
HostingMaker says so +2 · source ↗“Makes it really easy to host the Next JS frontend.”
HostingMaker says so · source ↗
Amazon Web Services
Hyperscaler cloud5 FastAPI products here (33%). About $17 a month for a team of 5 with 10,000 users (Pay as you go (Lambda + EC2 t4g.small)).
Best forThe widest catalog of building blocks — Lambda, containers, managed databases, queues — to assemble your own infrastructure.
Trade-offYou wire up networking, permissions and deploys yourself; billing is per service and hard to predict.
“Reliable Hosting made by Amazon AWS. Scaling will be a pleasure.”
- It offers a huge catalog of services in one place, from databases to queues to AI. 4 sources
- It scales up and down with bursty traffic without re-architecting. 3 sources
- Major outages, from long us-east incidents to struck data centers, have taken customer services down. 4 sources
- Its services can be tricky to configure. 1 source
“A great cloud provider!”
HostingMaker says so · source ↗“Reliable Hosting made by Amazon AWS. Scaling will be a pleasure.”
HostingMaker says so · source ↗“Our runtime engine for parallel package analysis runs on Fargate, giving us scalable isolation and predictable performance without infrastructure overhead.”
HostingMaker says so · source ↗All of Amazon Web Services: who uses it, alternatives, pricing →
Render
PaaS2 FastAPI products here (13%). About $61 a month for a team of 5 with 10,000 users (Pro).
llms.txtBest forA Heroku-style platform for web services, cron jobs and Postgres.
Trade-offFree instances sleep when idle and wake slowly; paid instances are billed whether busy or not.
- Connecting a GitHub repo gives automatic deploys without building pipelines. 1 source
- TLS, scaling, health checks and zero-downtime deploys work out of the box. 1 source
- It costs much more than a cheap VPS, and per-container pricing pushes people to pack processes together. 2 sources
Status page: 13 incidents in the last 90 days, 6 major (its own page, 2026-10-19).
“Allows me to easily deploy the backend Python API.”
HostingMaker says so · source ↗
Google Cloud Platform
Hyperscaler cloud2 FastAPI products here (13%).
Best forContainers on Cloud Run that scale to zero, next to BigQuery, Firebase and Google's AI services.
Trade-offMore setup than a PaaS, and IAM and billing take learning.
“Thanks to the amazing team at Google Cloud for supporting us and providing us with the necessary resources to train our model!”
“Gave us $2,000 USD in AI Startup Credits. Was a Godsend 🙏🏼”
HostingMaker says so · source ↗“Thanks to the amazing team at Google Cloud for supporting us and providing us with the necessary resources to train our model!”
HostingMaker says so · source ↗
Hetzner Cloud
Your own server2 FastAPI products here (13%). About $13 a month for a team of 5 with 10,000 users (CPX12).
Best forA lot of CPU and RAM per euro, in European, US and Singapore regions.
Trade-offFewer managed services than big clouds, and account verification can be strict.
- Servers cost a fraction of hyperscaler prices for strong price-to-performance. 3 sources
- Storage Boxes are a cheap offsite target for backups. 4 sources
- Recent price increases have made it less of a bargain. 4 sources
“To run the MySQL db and fastapi”
HostingMaker says so · source ↗Also chosen, and not seen yet
Fewer FastAPI products here: Netlify (1), Railway (1), Fly.io (1), Microsoft Azure (1), DigitalOcean (1).
Real options for hosting not seen among these FastAPI products yet: Cloudflare, Coolify.
Who switches, and to what
Public GitHub pull requests whose title says "X to Y", by developers in general, not only on FastAPI.
- Vercel → Render 123 pull requests · how they moved
- Netlify → Vercel 117 pull requests · how they moved
- Vercel → Railway 94 pull requests · how they moved
- Vercel → Netlify 65 pull requests · how they moved
- Render → Vercel 61 pull requests · how they moved
- Railway → Render 57 pull requests · how they moved
Getting hosting right
Connect the repo so every push deploys and every pull request gets a preview URL. Put secrets in the platform's environment variables, never in the repo.
- Leaving a paid platform on its default autoscaling and spend limits, so one viral day or a bot crawl turns into a surprise invoice.
- Putting a long-running job, websocket or large upload behind a serverless function and hitting its time or size limit in production.
- Keeping the database on a different provider or region from the app, so every query crosses the internet and pages get slow.
The rest of the FastAPI stack
What makers' FastAPI products chose for their other decisions, the most common pick for each:
- Database: PostgreSQL 6 of 13
Questions
What do most FastAPI apps use for hosting?
Among 15 makers' products built with FastAPI that we know the hosting of, Vercel leads: 6 of them (40%). Next come Amazon Web Services (5) and Render (2).
What does hosting cost for a FastAPI app with 10,000 users?
On our usage model for a team of 5 with 10,000 users, Vercel comes to about $120 a month (Pro + Flat Rate CDN 10M); Amazon Web Services comes to about $17 a month (Pay as you go (Lambda + EC2 t4g.small)); Render comes to about $61 a month (Pro); Hetzner Cloud comes to about $13 a month (CPX12). Prices are from each vendor's pricing page.
How this is counted: a product is on FastAPI when a live-product source says so (the maker, its website, a customer story or a subprocessor list), and the same for its hosting. How evidence is collected.