StackScope Crawled 40,000 Indie Launches. Here's What Founders Actually Ship With.

Quick answer: StackScope (stackscope.dev) is a free public catalogue that finds indie launches on Product Hunt, Hacker News and PeerPush, crawls each site to detect its tech stack, and scores it for launch readiness. Across its corpus of indie launches, roughly one in three runs on Vercel, about 54% use Tailwind CSS and about 36% use React. Strip out Vercel deployments and Tailwind falls to 46% while React falls to 20%, so the headline framework numbers are largely a Vercel story.

---

Twitter has been declaring framework deaths for four years. StackScope just crawled tens of thousands of real product launches and showed us what founders actually ship.

Not what they tweet about. Not what gets upvoted in framework comparison threads. What they actually deploy when they have something to sell and a launch date to hit.

The data is both obvious and genuinely surprising — and if you're about to make a stack decision for your next product, it's more useful than any benchmarking article you'll read this month.

---

Key Takeaways

• Vercel is the concentration point: Approximately 33% of all indie products in StackScope's corpus are deployed on Vercel.
• The Vercel effect inflates framework numbers: Excluding Vercel deployments, Tailwind drops from 54% to 46% and React drops from 36% to 20%.
• Tailwind leads the frontend: Around 54% of all tracked launches are detected with Tailwind CSS in the frontend.
• T3 and Supabase are the default shape: Next.js, TypeScript, tRPC, Prisma and Tailwind show up repeatedly, with Supabase as the database layer for auth, Postgres and real-time.
• Python came back for AI reasons: FastAPI appears on more than 30% of AI-focused launches in the 2026 dataset.
• Most AI is consumed, not built: The typical indie launch calls an API from OpenAI, Anthropic or Google rather than running its own models.
• Vercel share is slipping slowly: Fly.io, Railway and raw VPS deployments are gaining ground, mostly among founders who have shipped before.

---

What StackScope Actually Is

StackScope.dev launched quietly as a free public catalogue of indie product launches. The mechanism is simple and the output is dense: it finds new launches on Product Hunt, Hacker News, and PeerPush, crawls each product's site, detects the technology stack running underneath, and scores the launch for technical readiness. Every product gets its own page with a full stack breakdown.

What makes StackScope genuinely interesting for founders isn't the per-product breakdowns — it's the aggregated view across thousands of launches over months. You're not reading about which stack a specific successful founder chose. You're seeing what the entire indie founder population actually ships when the choice is theirs to make, the budget is real, and the timeline is tight.

By mid-2026, StackScope has tracked and analyzed thousands of launches across multiple months, with data drawn from a corpus that covers the full range of indie products — solo-founder micro-SaaS tools, small-team B2B products, AI-native utilities, and developer-focused infrastructure projects. The dataset is large enough that the patterns are meaningful, not just noise.

---

The Headline Numbers: What Founders Actually Build With

Let's start with the number that reframes everything else.

One in three indie launches runs on Vercel.

Not "a lot use Vercel." Not "Vercel is popular." Approximately 33% of all indie products in StackScope's corpus are deployed on Vercel. This is a level of market concentration in deployment infrastructure that has no direct parallel in previous tech cycles. Three years ago, saying a third of new web products ran on one hosting platform would have sounded like a vendor dream. In 2026, it's the reality for the indie market.

The other top-line numbers from StackScope's aggregated data:

• Tailwind CSS: ~54% of all launches detected with Tailwind in the frontend
• React: ~36% of all launches using React as the UI layer
• Google Analytics: consistently common across non-Vercel deployments

Those numbers look like Tailwind and React won. Then you control for Vercel.

---

The Vercel Effect Is Inflating Everything

This is the most interesting data point StackScope has surfaced, and it doesn't get enough attention.

When you strip Vercel deployments out of the dataset and look only at launches hosted elsewhere, the numbers shift dramatically:

• Tailwind drops from 54% to 46%
• React drops from 36% to 20%
• Google Analytics becomes more prominent relative to other analytics tools

What that tells you: Vercel isn't just a hosting choice. It's a complete development paradigm that pulls along a specific set of tools. When a founder chooses Vercel, they're very likely also choosing Next.js (Vercel's own framework), which practically mandates React, which pairs naturally with Tailwind and shadcn/ui, and integrates cleanly with Vercel's own analytics or Vercel-adjacent tools.

The Vercel-Next.js-Tailwind-shadcn/ui stack isn't four separate technology decisions. For a significant portion of the indie founder population, it's one decision: deploy to Vercel. Everything downstream follows from that.

This matters if you're reading stack adoption numbers anywhere. A lot of "Tailwind is winning" and "React is dominant" reporting is really "Vercel is dominant" dressed up in framework language. The actual framework diversity in the non-Vercel indie market is higher than the headline numbers suggest.

Notably, this also connects to a real tradeoff worth thinking through — Vercel's dominance comes with vendor concentration risk, which is why founders are increasingly exploring self-hosting alternatives to Vercel and Supabase as their products mature.

---

What the Full Stack Actually Looks Like

Beyond the Vercel effect, StackScope's data reveals a few dominant configurations.

The T3 Stack — Next.js, TypeScript, tRPC, Prisma, and Tailwind CSS — shows up repeatedly as the most complete indie SaaS configuration. It's opinionated, it moves fast for solo founders, and the tooling has reached a level of maturity where the rough edges from 2023 are mostly gone. If you're building a B2B SaaS that needs auth, database, API, and a responsive frontend, T3 on Vercel is the path of least resistance for the indie market.

Supabase as the database layer has cemented itself. The Postgres-as-a-service model with built-in auth and real-time subscriptions has become the default for founders who want a managed database without the configuration overhead of raw Postgres or the pricing unpredictability of DynamoDB. StackScope's detection data shows Supabase appearing alongside Vercel and Tailwind at high rates — these three tools have become a coherent primitive for the MVP phase. The open-source AI SaaS boilerplates that dominate GitHub right now are almost all built on this stack.

Python on the backend is back, for AI-specific reasons. Across AI-focused launches in 2026, FastAPI appears on more than 30% of products. WebReveal's 2026 dataset confirms the trend: Python reclaimed backend dominance for AI-native applications because the ML/AI library ecosystem — LangChain, LlamaIndex, Pydantic, Hugging Face — doesn't have TypeScript equivalents that are production-ready at the same level. If your product has a meaningful AI layer (RAG pipelines, LLM orchestration, model serving), Python backend with a Next.js frontend is now a standard configuration, not an unusual choice.

---

AI in Real Indie Launches: What the Data Shows

The AI question is the one every founder is asking, so let's be precise about what StackScope and the surrounding data actually reveal.

Most indie founders are consumers of AI, not builders of it. The AI capability in the average indie launch isn't a fine-tuned model or a custom RAG implementation — it's an API call to OpenAI, Anthropic, or Google. The AI layer is a product feature built on commodity access, not an infrastructure differentiation.

This isn't a criticism. It's a realistic read of where the leverage is. Calling an API for GPT-5.5, Claude, or Gemini Flash lets a solo founder ship AI-powered features in days that would have required a team of ML engineers two years ago. The entire category of vibe coding and AI-assisted development is built on this pattern.

The founders where AI shows up in the actual tech stack — where FastAPI and Python backends appear rather than just an OpenAI key in a .env file — are generally building AI-native vertical products: tools where the AI capability is the core product, not a feature bolted onto a traditional SaaS workflow. For these products, the stack choice matters in ways it doesn't for AI-enhanced horizontal tools.

The practical implication for stack decisions: if AI is a feature in your product, your stack choice doesn't change much from the pre-AI default. If AI is your product, Python backend is increasingly the right call. The zero-cost developer toolkit for shipping production apps covers what you need at either level.

---

What This Means for Your Next Stack Decision

Here's the honest answer the data supports: for most indie SaaS products, the stack decision is less important than the problem you're solving.

StackScope's data shows that the winning stack in the indie market is deeply concentrated. Next.js + Tailwind + Supabase + Vercel covers the majority of successful launches. Not because these are objectively the best tools for every use case — but because they're good enough for most use cases, have the deepest community support, and let a solo founder or small team move fast without hitting walls.

If you're trying to differentiate on tech stack — betting that building in Elixir or Rust or SvelteKit gives you a meaningful product edge — the data doesn't support that as a strategy. Your competition is using the boring stack and shipping faster than you can architect the interesting one.

The places where stack differentiation actually matters:

• High-scale read-heavy workloads where Next.js server-rendering costs add up and a more performant backend matters economically
• AI-native products where Python's ecosystem advantage is real and persistent
• Self-hosted products where your pricing model depends on eliminating your own infrastructure costs, making self-hosting alternatives worth the setup complexity
• Developer tools where the stack is part of the product signal — a dev tool built on Go ships a credibility message that a Node.js tool doesn't

For everything else: pick the boring stack, use a validated boilerplate, and spend the engineering time you save on distribution and user research. For the distribution side, our directory submission service can handle the directory listings for you.

---

The Patterns Worth Watching

StackScope's aggregate data is a lagging indicator — it shows what founders decided three to six months ago when they started building what just launched. Two patterns in that data are worth tracking forward.

Self-hosting is growing as a deliberate choice. The Vercel concentration in StackScope is real, but it's been slightly declining month-over-month as founders who experienced provider pricing changes or vendor lock-in concerns in 2025 make different choices. Fly.io, Railway, and raw VPS deployments are gaining share, particularly among founders who've shipped at least one product before and know what scale looks like.

AI adoption in stacks is broadening beyond the API call. The next wave of StackScope data will likely show local embedding models, vector database integrations, and self-hosted inference endpoints appearing more frequently — not because founders stopped using OpenAI, but because the economics of production AI workloads push toward hybrid architectures once you have real users and real token bills.

StackScope itself is a product worth bookmarking if you care about these trends. The per-launch scoring is useful for competitive intelligence, and the aggregate trend data will get more valuable as the corpus grows. You can browse the live data at stackscope.dev.

---

The Bottom Line

Forty thousand indie launches don't lie, even if the data is messier than a benchmark. The indie SaaS stack in 2026 is boring in exactly the right way: concentrated, well-supported, fast to ship, and increasingly AI-augmented at the feature level rather than the infrastructure level.

One in three launches runs on Vercel. More than half use Tailwind. Most AI products are making API calls, not training models. The founders winning aren't out-engineering the field — they're using the default stack to move faster and putting their real effort into finding problems worth solving.

That's what 40,000 data points look like when you zoom out.