You're So Mainstream: The State of Indie Launches, June 2026
Published 2 July 2026
How developer tools cross over into the indie web.
18,130 indie launches analysed in June 2026. We promised a hosting and infrastructure deep dive. It is here, and a sharper question came out of it.
You can tell how new a piece of infrastructure is by who uses it.
Last month we found that Vercel hosts about a third of all indie launches, and said the next issue would go deeper on infrastructure: which CDNs front which origins, whether edge compute is actually arriving, the long tail of self-hosted launches. We did all of that, and it is below. But the market-share tables were the less interesting half. The other half is who actually adopts each tool.
We can now classify what a launch is, not just what it runs, into product categories. Cross that against the tech stack and a pattern falls out that no market-share chart shows. Cloudflare Email Sending, the newest outbound-email service in our data and still in beta, already skews about as developer-heavy as anything we track: 48% of the launches using it are developer tools or AI products, against a corpus baseline of about 25%. Nearly twice the technical average. Meanwhile Resend, the email service everyone calls "the developer's choice," sits just below the baseline at 24%. It is not a developer tool anymore. It went mainstream.
That gap is the story. New developer infrastructure launches into a developer-only audience and then diffuses outward as it matures. The fraction of a tool's users who are technical is a clock on its adoption curve. This issue reads that clock across the whole catalogue.
Five things stood out
- Cloudflare Email Sending's audience is about as developer-heavy as anything we track. 48% of its users are dev or AI products, nearly twice the ~25% baseline. That is what a brand-new piece of infrastructure looks like before it crosses over.
- Resend already crossed over. The "developer's email" is now at the baseline (24%) and drifting below it. With 8,213 users it is a mainstream default, not a developer secret.
- AI builders are quietly mainstreaming "developer" tools. Vite, Firebase, and Lovable all skew non-technical, because AI builders scaffold them into the hands of founders who never chose them.
- At the edge, the indie web is one company; at the origin, it is no one. Cloudflare fronts 42.4% of launches. Below Vercel's third, no host clears 8.5%.
- 52 founders have now claimed their launch and recrawled. Median StackScope score lift: +1.8 out of 10.
A quick word on methodology
The scored launches come from Product Hunt (84.9%), PeerPush (6.9%), and Hacker News Show HN (8.1%). Every launch in this set was crawled during June 2026. Same baseline crawl as the last two issues: static HTML, rendered DOM via Playwright, DNS and RDAP, legal-page discovery. Tech detection runs against 5,676 fingerprints. Full methodology at /methodology.
New this issue: the product-category classifier that shipped last month. It sorts each launch into a product category from its tagline and description, and we only use classifications it is at least 90% confident about. That gives a clean sub-corpus of 41,981 launches across the full catalogue, not just the June cohort, where we know both what the product is and what it runs. The tech-share numbers below are read off this whole sub-corpus; a single month is too small to read a tool's audience from.
Two definitions used throughout. A launch is a technical build if its product category is Developer Tools or AI; everything else is a non-technical build. A tool's tech-share is the percentage of its users that are technical builds. Across the whole corpus, about 25% of launches are technical builds, so 25% is the line a tool sits on if its audience looks like the market average. Above it, the tool skews to developers; below it, to everyone else. That 25% baseline has barely moved month to month (25.2%, 25.1%, 25.1% across April, May, June), so it is a stable yardstick.
The Bolt AI-builder match was tightened this month to remove a few false positives; it moved named-builder counts by a rounding error, not a trend.
The cohort
| Source | Launches | Share |
|---|---|---|
| Product Hunt | 15,393 | 84.9% |
| Hacker News (Show HN) | 1,471 | 8.1% |
| PeerPush | 1,254 | 6.9% |
| Manual / other | 12 | <0.1% |
| Total scored | 18,130 | 100% |
The mix is holding at the steady state of the last three months. Nothing in the cohort composition is doing anything surprising, which is what lets the infrastructure story stand on its own.
The Ultimate Question of CDN, the Edge, and Everything
First, the deep dive we promised, because the answers set up everything after.
The edge is one company. Cloudflare fronts 42.4% of every launch we scored, rising to 52.8% of the launches not on Vercel. It is the single most common piece of infrastructure in the dataset, ahead of any host and any framework. The next CDN, Fastly, is on 5.8%; AWS CloudFront and the rest trail far behind. HTTP/3 (39.0% of launches, 54.5% non-Vercel) and Cloudflare DNS (34.4%, 39.1% non-Vercel) move with it, because founders get both for free once they are on Cloudflare.
Here is the coincidence we could not resist. Cloudflare's Q1 2026 investor deck states, as of March 31 2026, that 42% of the Fortune 500 are paying customers (the same number leads their homepage). We do not track the Fortune 500. We track indie launches, the products at the furthest possible end of the market from a Fortune 500 company: newer, smaller, usually one person. Cloudflare fronts 42.4% of those too. Douglas Adams readers will know where this is going: the answer, at both ends of the web, turns out to be about 42. The two numbers measure different things, theirs a customer count among the largest companies on earth, ours an edge detection on the smallest, so the match is a coincidence and not a finding. But it is a striking one, worth filing next to Cloudflare's other public figure: about 20% of the web at large runs through its network.
The origin is no one. Vercel hosts 32.2% of launches, a number that has barely moved across three editions. But it is the only host with real concentration. Strip it out and the leader of the remaining 12,292 launches is Hostinger at 8.5%, with no provider above 8.5% and a long flat tail after it. There is a Cloudflare of the edge. There is no Cloudflare of origins.
Edge compute has not arrived, as far as we can tell. This is the one promised thread we have to answer with a caveat instead of a number. We do not fingerprint the edge-compute runtimes that matter most: Cloudflare Workers, Vercel Edge Functions, Netlify Edge Functions, Fastly Compute. We can see OpenNext (1.7% of the non-Vercel cohort) and a scatter of Deno Deploy and Wasmer, and that is about it. So the finding is not "edge compute is rare," it is "we cannot measure it yet, and the visible signal is small." We would rather say that plainly than dress a detection gap up as a trend.
The self-hosted long tail is real. Hostinger, traditional shared hosting with a cPanel and a one-click WordPress installer, is the second-biggest host overall. WordPress holds 7% of the non-Vercel cohort and jQuery 6%, neither of which appears on the full-cohort top 15. Hold that thought, because this cohort turns out to be the same one the next section calls the non-technical pole.
Who runs what
Now the question the map cannot answer: not how big each tool is, but who uses it.
Sort every tool we track by tech-share, the percentage of its users that are technical builds, and the catalogue splits into two clean poles around the 25% baseline.
The developer pole (tech-share well above 25%):
| Tool | Tech-share |
|---|---|
| Cloudflare Email | 48% |
| GitHub Pages | 42% |
| Fly.io | 39% |
| Caddy | 37% |
| OpenNext | 36% |
| Astro | 35% |
| Fastly | 34% |
| SvelteKit | 34% |
The mainstream pole (tech-share well below 25%):
| Tool | Tech-share |
|---|---|
| WooCommerce | 8% |
| Elementor | 9% |
| Apple App Store | 9% |
| Rank Math | 9% |
| Site Kit by Google | 10% |
| WordPress | 11% |
| PHP | 12% |
| LiteSpeed Web Server | 12% |
Tech-share = % of a tool's users that are Developer Tools or AI products. Corpus baseline is ~25%. Tools with at least 120 high-confidence users shown; the developer table is the dev-infrastructure and framework subset of that pole (the raw ranking also includes B2B and ad tools, a different story).
The developer end of the axis is the modern dev-infra and dev-native-framework cluster: Cloudflare's newer products, Fly.io, the build-it-yourself frameworks (SvelteKit, Astro), the static-host-for-coders (GitHub Pages). The mainstream end is the WordPress and PHP world, plus app stores and ecommerce plugins. The self-hosted long tail from the infrastructure map and the non-technical pole here are the same launches, seen from two angles.
A cleaner way to draw the same axis is the diverging view: each tool's adoption rate among technical builds minus its rate among everyone else, centred on zero. Developer-coded tools lean right, mainstream tools lean left.
The caveat on magnitude: nothing is wildly partisan. The strongest developer lean is Cloudflare DNS at about +6 points, the strongest mainstream lean is Google Analytics at about -6. The indie stack is a near-monoculture and most tools are used by everyone. The tells are real but subtle, which is exactly why the tech-share composition (11% versus 48%) reads louder than the raw difference.
The adoption clock
Put the email category on the tech-share axis and you get an adoption curve laid out in a single column.
| Email tool | Users | Tech-share |
|---|---|---|
| Cloudflare Email Sending (new, beta) | 157 | 48% |
| SendGrid | 1,244 | 26% |
| Mailgun | 574 | 26% |
| Resend | 8,213 | 24% |
| Amazon SES | 1,625 | 24% |
User counts are within the classified sub-corpus (the ~80% of launches with a high-confidence product category), so they read lower than the all-time totals on the linked pages.
Cloudflare Email at the top is brand new and still developer-only. Resend, at 24%, has already made the trip: the tool the whole indie world calls "the developer's email" now has an audience that looks like the market average, because with 8,213 users it is the market. It started where Cloudflare Email is now and walked down the curve. The mature legacy senders cluster on the baseline. Below all of them sit the marketing-email tools non-technical founders reach for: Mailchimp and Klaviyo land between 13% and 17%.
It is worth saying why the newest tool starts so far to the developer side, because it is not only that early adopters are technical by nature. Cloudflare recruits them on purpose. Its own Q1 2026 investor deck leads the impact slide with the developer audience: more than 4,000 startups building on Cloudflare, over 500 of them AI startups, plus a Workers Launchpad accelerator backing 175 more. A new Cloudflare product launches straight into the audience the company spends to assemble. The adoption clock starts where the marketing does.
Is this a real diffusion over time, or just a snapshot of tools at different ages? Three months is too short to watch a single tool move, and the small new tools are too noisy: Cloudflare Email bounces around month to month on a few dozen launches. But two things hold. The cross-section reproduces every month against that flat 25% baseline, so it is not a one-month fluke. And the one tool big enough to show drift, Resend, has slid 24.9% to 23.9% to 23.1% across the April, May, and June cohorts (the table's 24% is the whole catalogue in one number), a gentle monotonic decline past the baseline. The dev-darling is not just mainstream, it is becoming slightly less technical than the average launch.
So we will treat the clock as a recurring measurement from here. Each month we will report the tech-share of the newest infrastructure and watch it descend. Cloudflare Email is the one to watch: if the theory is right, its number falls over the next year as it crosses over.
The tools AI builders are mainstreaming
The diverging view turned up three tools that sit on the wrong side of the line. Vite (22%), Firebase (17%), and Lovable (15%) all skew non-technical, Firebase and Lovable clearly, Vite by a smaller lean. A non-technical founder is more likely to ship these than a developer is, and for a build tool, a backend platform, and an AI app-builder, any lean that way is backwards.
The mechanism is the AI builders. Lovable, Bolt, and v0 scaffold their output as Vite-plus-React apps, often wired to Firebase, and the people using them to spin up a marketing site or a small productivity tool are mostly not developers. So a stack that reads as "developer infrastructure" is being pushed into non-technical hands by a generation of tools that write the boilerplate for you. The clearest single case is Lovable itself, which we detect on the finished sites: those sites skew non-technical, because not-coding is the entire pitch.
This is the same diffusion clock running in reverse. Normally a tool starts technical and broadens. These tools are being carried into the mainstream by an abstraction layer on top of them, faster than they would have travelled on their own.
Risers and fallers
| Field | May 2026 | June 2026 | Δ |
|---|---|---|---|
| Vercel hosting share | 32.1% | 32.2% | +0.1pp |
| Tailwind CSS adoption | 53.9% | 52.0% | -1.9pp |
| React adoption | 35.6% | 35.6% | +0.0pp |
| Next.js adoption | 33.7% | 34.1% | +0.4pp |
| Resend share of email | 18.5% | 19.3% | +0.8pp |
| Cloudflare Email Sending | 0.44% | 0.45% | +0.01pp |
| Named AI-builder cohort | 8.5% | 8.6% | +0.1pp |
| Lovable share of named cohort | 51.9% | 50.2% | -1.7pp |
| Avg Vibe Score (PH) | 32.1 | 31.7 | -0.4 |
| Avg Launch Readiness (PeerPush) | 90.7 | 91.3 | +0.6 |
| Avg Vibe Score (HN Show HN) | 20.0 | 19.7 | -0.3 |
Auto-derived from last month's data. Anything inside about two points is noise.
The stack stayed sticky again: Tailwind, React, Next.js and the Vibe Scores all held inside a point or two. The infrastructure line that matters this issue is Cloudflare Email Sending, which is the subject of the adoption-clock section above rather than a delta to extrapolate. Cloudflare Email Routing, its inbound sibling, sat on 1,829 launches, the largest Cloudflare-in-email footprint in the data and the natural conversion pool for the outbound product.
Tuesday is launch day, mostly
We checked launch timing across the whole corpus again (52,287 launches). The finding from May holds: Tuesday leads (21.3% of launches against Wednesday's 17.3%), the weekend is the only real cliff, and the Tuesday peak is a Product Hunt artifact. PeerPush, where most founders take a scheduled queue slot rather than picking a day, is flat across the week and peaks gently on Friday. The Tuesday peak only exists where founders choose the day, which makes it look self-fulfilling.
The claim funnel
52 founders have now claimed their listing and triggered a recrawl, up from 28 at the end of May. Of the 45 we could compare cleanly, 32 shipped at least one observable fix. Most claimers were already polished, with a median starting score of 7.4. The median per-launch lift was +1.8, mostly from adding the easily-forgotten files and headers (llms.txt, security.txt, a Referrer-Policy header, a privacy policy, robots.txt).
If you launched on Product Hunt, PeerPush, or Show HN in June and haven't checked your listing, find it on /browse. Claiming is free. The Launch Readiness Check runs the same analysis on any URL.
What we can now measure
Two new things, and both are why this issue exists and where the series is going next.
Product categories. Every launch now gets sorted into a product type (developer tools, AI, ecommerce, marketing, and a dozen more) from its tagline and description. That classifier is the engine of this whole issue: without knowing what a product is, there is no tech-share, no adoption clock, no developer-versus-mainstream split. It also now drives the category filter on /trends and the breakdown on every /tech page.
Country of origin. New: an estimate of where each launch is actually based. Most "where is this from" signals really measure hosting, which for the indie web means Cloudflare or Vercel, not the founder. So we lean on signals that point at the company rather than the server, and we say "unknown" rather than guess. Launch pages now show an estimated-origin flag where we are confident enough, and about 31.6% of June launches carry one. For any published country number we tighten to high confidence across the full catalogue, not just the June cohort. That is a smaller slice (about 4.2% of the corpus), and on it the early leaders are:
| Estimated origin (high-confidence) | Launches |
|---|---|
| India | 568 |
| United States | 431 |
| United Kingdom | 259 |
| Canada | 164 |
| Australia | 67 |
| Brazil | 55 |
| Germany | 52 |
| Japan | 47 |
The high-confidence slice only (~4.2% of the corpus so far); most launches still resolve to "unknown". Directional, not a census, and it grows as coverage improves.
Among the launches we can place with high confidence, India edges the United States, which is not the order most people would guess. That is a slice, not the whole market, but it is a good question to pull on.
What's next
This is issue #3. The recurring shape holds: a monthly feature, the deltas table, the day-of-week check, the claim funnel. New recurring item from this issue: the adoption clock, the tech-share of the newest infrastructure, tracked month over month. Cloudflare Email is the first tool on the watch list.
Next month we point the lens at geography. With country-of-origin estimates now live, issue #4 asks whether different countries favour different stacks: does an Indian launch reach for a different default than a German or Brazilian one, or is the indie monoculture global? The tech-share machinery from this issue, crossed against origin instead of product category, should answer it.
Bookmark /blog.
Credits and links
Thanks to the PeerPush team for the ongoing integration support, and to the readers whose questions on the April and May posts shaped both.
Explore the June data
- Find your launch: see how your June launch was scored.
- Claim a listing: verify ownership from your launch page and trigger a fresh crawl.
- Slice the trends: stackscope.dev/trends, now filterable by product category.
- Browse the tech catalogue: stackscope.dev/tech.