Every layer of a typical MVP stack has a free tier in October 2026, but two of the free tiers you are most likely to start on cannot carry a paying product. Vercel's pricing page says its Hobby plan "is for personal, non-commercial use", and Supabase's says free projects "are paused after 1 week of inactivity". The cheapest commercial footing for the two layers that matter most, Vercel Pro and Supabase Pro, is $45 a month before usage.
This page goes through eight layers, from framework to error tracking, with the free-tier limits and first paid step read from each vendor's own pricing page on 18 October 2026, so every price below is vendor-sourced. Our bias, stated once and up front: we build applications in Next.js, so the framework choice was made before the comparison started.
The recommended stack at a glance
One default per layer, with alternatives where the default has a case for being wrong: Next.js; Vercel, or Netlify or Cloudflare Workers; Supabase, or Xano or Firebase; Clerk or Supabase Auth, or Auth0; Stripe; Resend; PostHog; Sentry. The tables below give each layer's free tier and first paid step.
Framework: Next.js, with our bias showing
We build applications in Next.js, so read this section as a stated preference rather than a neutral review. The case for it at MVP stage is that one codebase holds the marketing pages, the signed-in product and the server code.
The case against is that Next.js asks you to understand its caching model, and that model changes depending on where you host it. The Next.js self-hosting guide, updated 15 October 2026, says a single `next start` instance with a persistent disk works without extra setup. Run several instances and you need a shared cache handler, a consistent Server Functions encryption key, a deployment ID to prevent version skew, and cache tag coordination so that a `revalidateTag()` call on one instance reaches the others. None of it is hard, and none of it is work a two-person team needs in month one, which is the argument for a managed host.
Hosting: Vercel, and when Netlify or Cloudflare fit better
Vercel maintains Next.js and spares you the multi-instance setup above. The price is a per-seat plan with uncapped overage.
| Host | Free tier | First paid plan | At the limit |
|---|---|---|---|
| Vercel | Hobby: 100 GB transfer, 1M function invocations, 1 seat, non-commercial | Pro: $20 per developer seat, $20 credit, 1 TB then from $0.15 a GB | Hobby cannot buy more; Pro is uncapped, $200 default budget, optional pause |
| Netlify | 300 credits a month | Personal $9 for 1,000 credits; Pro $20 for 3,000 credits, unlimited members | One project over its limit pauses every project on the account |
| Cloudflare Workers | 100,000 requests a day, 10 ms CPU per invocation | $5 a month minimum: 10M requests, 30M CPU ms, then $0.30 per million requests | No data transfer charge on Paid |
Source: vercel.com/pricing, netlify.com/pricing and Cloudflare Workers pricing docs (updated 2026-10-02). Vendor-sourced figures.
Vercel. The default for a small Next.js team. Every extra developer seat adds $20 a month, and overage has no ceiling unless you set the pause, so set the budget on day one.
Netlify. Its credits are easy to read: 20 per GB of bandwidth, 15 per production deploy. The trade-off is the pause rule: a free account running a demo and a real product loses both when the demo goes viral.
Cloudflare Workers. The cheapest place to serve a lot of traffic, because bandwidth is not metered on the Paid plan. Workers is not a `next start` server, so a Next.js app runs there through an adapter; the Next.js docs point platforms to a Deployment Adapter API, so check which features your adapter supports before you choose it. We would pick Cloudflare for a high-traffic, mostly static product, not for a first MVP.
Database and backend: Supabase, Xano or Firebase
Our Xano vs Supabase in 2026 post compares those two plan by plan, so we will not repeat its tables here. The short version: Supabase is Postgres with auth, storage, realtime and functions attached. Its Free plan gives 500 MB of database per project, 50,000 monthly active users and two active projects, pausing after a week of inactivity. Pro starts at $25 a month with 100,000 monthly active users, then $0.00325 each, a $10 compute credit and a spend cap on by default. Xano's Free plan is limited to 10 API requests per 20 seconds, so a real product starts on Essential at $85 a month billed yearly.
Why Supabase is our default. For a Next.js team with a developer, the database is plain Postgres, the data leaves with standard tools, and the same service covers auth. That removes one vendor from the stack.
When Xano fits. When the people changing business logic after launch are not backend developers. The cost is that logic lives on Xano, as the comparison post explains.
When Firebase fits. Firebase's Spark plan gives Cloud Firestore 1 GiB of storage, 50,000 document reads and 20,000 writes a day, and authentication free up to 50,000 monthly active users. It suits mobile-first apps that want a document database. Two limits matter: the pricing page marks Cloud Functions and Cloud Storage "Not applicable" on Spark, so server code and uploads need the pay-as-you-go Blaze plan, whose rates it lists only as "Google Cloud pricing".
Auth: Clerk, Supabase Auth or Auth0
All three give a free start; they differ in what they count, what they leave out and how steeply they climb.
| Provider | Free cap | Not on the free plan | First paid plan | Billing unit and overage |
|---|---|---|---|---|
| Clerk | 50,000 per app | MFA, passkeys, SSO, branding removal | Pro: $25 a month, $20 yearly | Monthly retained users; $0.02 each from 50,001 to 100,000 |
| Supabase Auth | 50,000 | Included in the database plan | Supabase Pro from $25 a month, 100,000 included | Monthly active users; $0.00325 each over 100,000 |
| Auth0 | 25,000 | Role management, Pro MFA | B2C $35 or B2B $150 a month at 500 users | Monthly active users, priced in tiers |
Source: clerk.com/pricing, supabase.com/pricing and auth0.com/pricing, read 2026-10-18. Vendor-sourced figures.
Clerk. In our view, the quickest route to finished sign-in screens and organisations in a Next.js app. Clerk counts a user only if they return at least 24 hours after signing up, so one-time sign-ups cost nothing. The trade-off is the price curve: at 60,000 retained users Clerk Pro costs $25 plus 10,000 at $0.02, so $225 a month, while Supabase Pro still covers that number in its $25.
Supabase Auth. The cheapest at scale if Supabase is already the database, and its sessions sit next to Row Level Security policies. You build more of the interface yourself.
Auth0. The enterprise option, with role management and enterprise MFA on its paid plans, but its paid B2B plan starts at $150 a month at only 500 users. Choose it when the first customer's security review asks for it, not before.
Payments, email, analytics and error tracking
| Service | Free tier | First paid step |
|---|---|---|
| Stripe | No setup or monthly fees | 1.5% + 20p UK cards; 2.5% + 20p EEA cards; 3.15% + 20p international; +2% for currency conversion; Billing 0.7% of volume |
| Resend | 3,000 emails a month, 100 a day, 3 domains, 30-day retention | Pro: $20 for 50,000 emails, $0.90 per extra 1,000; no daily limit |
| PostHog | 1M events, 5,000 recordings, 1M feature flag requests, 100,000 exceptions, 1 project | Pay as you go, with a billing limit per product |
| Sentry | 1 user, 5,000 errors, 30-day lookback | Team: $26 a month yearly or $29 monthly, 50,000 errors, unlimited users |
Source: stripe.com/gb/pricing, resend.com/pricing, posthog.com/pricing and sentry.io/pricing, read 2026-10-18. Vendor-sourced figures; Stripe shown in UK pricing.
Stripe. On a £20 monthly subscription paid with a standard UK card, Stripe takes 1.5% plus 20p, which is 50p, and Billing adds 0.7%, another 14p: 64p, or 3.2% of the charge. Stripe Tax Basic adds 0.5% per transaction on the no-code integration. If you sell software internationally and do not want to register for VAT and sales tax yourself, Stripe's Managed Payments acts as merchant of record, with indirect tax compliance and remittance in more than 75 countries, for 3.5% per transaction on top of payment fees.
Resend. A transactional email API with a free tier sized for a beta. Move to Pro before launch day, because 100 emails a day covers a beta and not a Product Hunt morning.
PostHog. Product analytics, recordings and feature flags in one free tier. Our Analytics Stack for a Four-Week Site Build post compares it with GA4 and Clarity for marketing sites; inside a product, we would start with PostHog because one funnel runs from first visit to first action.
Sentry. The specialist for errors, with tracing, logs and session replay on the same plans. PostHog's free tier includes 100,000 exceptions a month, twenty times Sentry's free count, so a solo founder can start there. Once two developers need alerts, Sentry Team is worth $26 a month.
Three choices we would not start with
These are not tools we have dropped; they are starting choices the pages above make expensive.
1. A paying product on a free tier. Vercel Hobby is non-commercial and cannot buy extra usage. Supabase Free pauses after a week without activity. Netlify Free pauses every project on the account when one project hits its limit. Fine for a prototype, a risk once a customer pays. Budget the $45 a month from the first invoice.
2. Self-managed infrastructure before product-market fit. Running Next.js across several containers means the cache, key and deployment ID work described above. Our MVP Development Statistics 2026 post shows most founders pivot at least once, so infrastructure should be cheap to throw away.
3. Hand-built auth and billing screens. Sign-in, password reset, MFA, subscription changes and invoices are solved problems with free or percentage-priced tools above. Writing them by hand spends build weeks on parts no user will praise. Keep the data in your database and use a provider for the flows.
What the stack costs in month one and month twelve
| Layer | Month one: 1 developer, first paying users | Month twelve: 2 developers, 20,000 users, MFA, 30,000 emails |
|---|---|---|
| Vercel | $20 (Pro, 1 seat) | $40 (Pro, 2 seats), plus any usage above the credit |
| Supabase | $25 (Pro) | $25 (Pro), plus compute above Micro if needed |
| Clerk | $0 (Free) | $25 (Pro, for MFA) |
| Resend | $0 (Free) | $20 (Pro) |
| PostHog | $0 (Free) | $0 if under 1M events |
| Sentry | $0 (Developer) | $29 (Team, monthly) |
| Total before Stripe fees | $45 | $139 |
Source: our arithmetic on the vendor pricing pages listed below, read 2026-10-18. Excludes usage overage, domains and developer time.
The larger cost is build time; our MVP Development Statistics 2026 post collects agency estimates, and we quote per project. Tooling is rarely why an MVP budget fails, so pay for the plans that keep the product online.
How to use this
Pick the default unless you can name the reason not to. Use Supabase Auth instead of Clerk if the user count will pass 50,000 early or the budget is tight. Use Xano instead of Supabase if non-developers will own the backend logic. Use Firebase if the product is mobile-first and the data fits documents. Use Auth0 when an enterprise buyer's security review names it. Use Cloudflare when bandwidth, not developer time, is the main cost.
Then do four things in week one: move hosting off any non-commercial plan, set Vercel's spend budget and Supabase's spend cap deliberately, upgrade Resend before launch day, and wire PostHog and Sentry before the first user arrives so the first bug report has a stack trace. Our web app development service page explains how we scope a build on this stack.
