Everyone has an opinion about what indie makers build with. Very few of those opinions come with numbers attached.
We had the numbers sitting in our database, so we counted.
This is the tech stack breakdown of 466 live products on BuiltByIndies, submitted by independent makers over the past year. Every product here is real, shipped, and publicly listed. Makers pick their own stack tags when they submit, so this reflects what people say they actually built with — not what a survey respondent aspires to use next quarter.
Some of it confirmed what we expected. Some of it genuinely surprised us.
The Short Version
Vercel is on 27% of all products — more common than any framework, language, or database.
Next.js is the default framework at 104 products, more than double the next non-React option.
Firebase beats Supabase 65 to 41, which was not what we predicted.
Gemini is the most-used AI model at 46 mentions, ahead of Claude (24), OpenAI (10) and Anthropic's API (10).
Only 7% of products are open source — 32 out of 466.
81% are free or freemium. Only 54 products are paid-only.
Hosting: Vercel Won, and It Isn't Close
Platform | Products | Share |
|---|---|---|
Vercel | 126 | 27.0% |
Cloudflare | 23 | 4.9% |
Google Cloud | 9 | 1.9% |
AWS | 9 | 1.9% |
Railway | 7 | 1.5% |
Netlify | 3 | 0.6% |
Vercel appears on more products than React does. It is the single most common tag in the entire dataset.
The interesting part is the gap. Vercel has 5.5x the adoption of Cloudflare and 14x that of AWS. Among indie makers, the "just push to Vercel" path has effectively become the default, and the traditional cloud providers barely register.
That gap says something specific: for a solo maker, deployment friction matters more than unit economics. AWS is cheaper at scale, but almost nobody in this dataset is at scale. They are trying to ship this weekend.
Frameworks: React, Then Everyone Else
Framework / Library | Products |
|---|---|
Next.js | 104 |
React | 74 |
Tailwind CSS | 32 |
Flutter | 11 |
Vite | 9 |
Astro | 8 |
Nuxt | 6 |
Expo | 6 |
Vue | 5 |
Svelte | 4 |
Angular | 4 |
Laravel | 4 |
Next.js appears more often than plain React, which tells you most React users here are not starting from Create React App or Vite — they are starting from a framework with routing and server rendering already decided.
The Vue and Svelte numbers are the ones worth sitting with. Both are widely liked, both have devoted communities, and both show up on fewer than 1% of products each. Combined, Vue, Svelte, and Angular total 13 products against Next.js's 104.
This is not a statement about quality. It is a statement about defaults. When you are a solo dev with limited hours, you pick the ecosystem with the most tutorials, the most Stack Overflow answers, and the most AI training data — and right now that is unambiguously React.
The Backend Surprise: Firebase Still Beats Supabase
Backend / Database | Products |
|---|---|
Firebase | 65 |
Supabase | 41 |
PostgreSQL | 35 |
MongoDB | 10 |
Redis | 9 |
MySQL | 9 |
Convex | 6 |
Prisma | 6 |
SQLite | 5 |
Neon | 3 |
If you spend time on tech Twitter you would assume Supabase had comprehensively won this category. In our data it has not — Firebase leads by roughly 1.6x.
Two plausible explanations. First, Firebase has a decade-long head start, and a lot of these products started as something else before becoming a launch. Second, Firebase Auth remains extremely easy to drop in, and auth is usually the first hard thing a solo maker hits.
Worth noting: PostgreSQL appears 35 times independently of Supabase, so the actual Postgres footprint is larger than the Supabase number alone suggests. The relational-vs-document question is fairly settled — Postgres-based options outnumber MongoDB by roughly 8 to 1.
AI: Gemini Leads on Volume
AI Provider / Model | Products |
|---|---|
Gemini | 46 |
Claude | 24 |
OpenAI | 10 |
Anthropic (API) | 10 |
Perplexity | 4 |
Grok | 3 |
OpenRouter | 3 |
DeepSeek | 2 |
Groq | 1 |
Gemini's lead is the most counterintuitive result in the dataset, and we think the explanation is mostly economic. Google's free tier is generous, and for a maker validating an idea with no revenue, "free" beats almost every other consideration.
Combining Claude and Anthropic tags gives 34 — second place, and notably strong given no comparable free tier.
The bigger story is the denominator. 94 products are tagged AI, our single largest category, but only around 100 stack mentions reference a specific model provider. A meaningful number of "AI products" either wrap a provider they did not tag, or use AI mostly as positioning.
Languages
Language | Products |
|---|---|
TypeScript | 41 |
Python | 34 |
JavaScript | 29 |
.NET / C# | 30 |
Swift | 15 |
PHP | 12 |
Rust | 6 |
Go | 5 |
Java | 3 |
Kotlin | 3 |
Elixir | 3 |
TypeScript over JavaScript, but by less than the online discourse implies — plenty of makers are still shipping plain JS and doing fine.
The .NET number at 29 is the quiet outlier. It gets almost no attention in indie maker circles, yet it is roughly as common as TypeScript here. There is a whole population of makers building profitable things in C# who simply do not post about it.
Rust and Go together total 11 products. Both are beloved in surveys. Neither is where indie makers actually ship consumer products.
What Everyone Is Building
Category | Products |
|---|---|
AI | 94 |
Productivity | 88 |
SaaS | 56 |
Marketing | 34 |
DevTool | 27 |
Mobile App | 22 |
SEO Tools | 17 |
Education | 16 |
Analytics | 16 |
Gaming | 14 |
AI and Productivity together account for around 39% of everything listed. If you are building an AI productivity tool, you are not early — you are in the most crowded room on the internet.
The thin categories are more interesting than the fat ones. Healthcare (3), Legal (1), Accounting (1), Real Estate (4), and Invoicing (2) are barely represented. These are unglamorous, regulated, boring-sounding markets where customers have budgets and competition is thin. The distribution suggests makers chase interesting problems rather than profitable ones.
Pricing: Almost Nobody Charges Upfront
Model | Products | Share |
|---|---|---|
Freemium | 242 | 51.9% |
Free | 135 | 29.0% |
Paid | 54 | 11.6% |
Free Trial | 30 | 6.4% |
Lifetime Deal | 5 | 1.1% |
Four out of five products are free or freemium. Only 11.6% ask for money at the door.
Freemium at 52% is the default, and it is worth asking how many of those chose it deliberately versus defaulted into it because charging felt presumptuous. A free tier is a real ongoing cost — support, infrastructure, and the psychological weight of users who will never pay.
The 135 fully free products are a mix of genuine passion projects and products whose makers have not yet decided what they are. Both are fine. But "free" is a decision, not a default, and the data suggests it is often treated as the latter.
Open Source Is Rarer Than You Think
Just 32 of 466 products (6.9%) are open source.
Indie hacking and open source get discussed as if they were adjacent cultures. In practice they overlap far less than the conversation implies. Most makers here are building closed products they hope to monetise, which is a perfectly reasonable thing to do — it just does not match the community's self-image.
What We'd Take From This
The default stack is real. Next.js on Vercel with Tailwind, a hosted backend, and an AI API is what a large share of new indie products look like in 2026. Boring and well-trodden is a legitimate advantage when you are one person.
Your stack is not your differentiator. Nothing here correlates with a product doing well. Choosing Svelte over React will not save a product nobody wants, and choosing React will not sink a good one. Optimise for how fast you can ship and how easily you can debug at midnight.
The gaps are the opportunity. 94 AI products and 3 healthcare products is not a statement about market size. It is a statement about what feels fun to build.
Decide your pricing on purpose. If you are going to be free, be free for a reason you can articulate.
Method and Caveats
Data comes from 466 publicly visible products on BuiltByIndies as of this writing, covering products with LAUNCHED, SCHEDULED, or featured status. Stack tags are self-reported by makers at submission, from a picker plus free text.
Three limitations worth stating plainly. Tags are optional, so absence is not proof of non-use. Some tools appear under multiple spellings, which we merged by hand where obvious. And our audience skews toward web-first indie makers — a dataset from a mobile-first or enterprise community would look very different.
We will re-run this as the number grows. If you want your stack in the next edition, submit your product — it is free, and you can browse what everyone else is using by tech stack or by category.




