Product Hunt is swamped now

Everyone building software can feel that AI has changed the internet, but I wanted a number for one small corner of it.

Product Hunt is where I noticed it most. I have used it for years as a rough way to see what people are launching, and it used to be something I could scan in a few minutes. Now the feed feels full in a different way: AI wrappers, agents, “for X” tools, relaunches, prototypes, experiments, and products that look like they went from idea to launch page in a weekend.

So I counted one comparable day across time: the second Wednesday of May, every year from 2015 to 2026.

Product Hunt leaderboard entries on the second Wednesday of May.

That is the whole finding. From 2015 to 2023, this sample stays between 22 and 67 launches; then it goes to 139 in 2024, 258 in 2025, and 783 in 2026. That is about 17x from 2023 to 2026. If you use 2020 as the base, it is about 22x.

Method

I used the “all” daily leaderboard, not just the featured page, at https://www.producthunt.com/leaderboard/daily/YYYY/5/DD/all. I picked the second Wednesday of May to avoid comparing weekdays with weekends, then followed Product Hunt’s infinite-scroll GraphQL cursor until the day was exhausted and counted unique launch IDs whose launch date matched the requested day. That date filter matters because the endpoint can inject off-date posts into old leaderboards.

This is still a small sample: one weekday per year, one site, one leaderboard definition. A stronger version would scrape every Wednesday in May, or every day of the year, and plot medians. I also counted leaderboard entries, not unique companies, durable products, or products that still work six months later.

AI is changing the internet at the production layer. The visible part is chatbots and image generators, but the bigger effect for builders is that the cost of turning a half-formed idea into something launchable has fallen: name, logo, copy, landing page, screenshots, scaffolded app, auth, billing, LLM integration, demo video, and launch post.

That does not make the software better. It does make the launch queue longer, and Product Hunt is one of the places where that queue is easiest to see.

The same pattern shows up elsewhere. GitHub said in an April 2026 availability update that it started planning for 10x capacity in October 2025, then by February 2026 had to design for a future requiring 30x current scale, after agentic development workflows drove sharp growth in repository creation, pull requests, API usage, automation, and large-repository workloads. GitHub’s Octoverse 2025 report gives the broader background: more than 230 new repositories per minute, nearly 1 billion commits in 2025, 1.13 million public repositories depending on generative-AI SDKs, and 4.3 million AI-related repositories.

Product Hunt has also said the quiet part out loud. In its “All the AI that launched in 2025” newsletter, 13 of the top 15 Product Hunt launches that year were tagged “Artificial Intelligence.” A recent paper, Generative AI Fuels Solo Entrepreneurship, but Teams Still Lead at the Top, uses more than 160,000 Product Hunt launches and argues that entrepreneurial entry increased sharply after ChatGPT-3.5, especially for solo entrepreneurs, while much of the increase looked like low-commitment experimental entry.

That sounds right to me. Product Hunt used to feel like a shelf; now it feels like a queue. The count is not a precise measure of startup formation, but it is a decent measure of what a reader has to wade through, and by that measure Product Hunt is swamped now.

Reproduce it

Open any Product Hunt page in a browser where the site works, then paste this into DevTools. It writes the table to window.phLaunchCounts, the individual launches to window.phLaunches, and tries to copy a CSV to your clipboard.

Show the script

(async () => {
  const FROM = 2015;
  const TO = 2026;
  const HASH = "4b8a0e19d053a2e8d06a1df97d13499cafc864148d7316de06d6a2292f9cf2bb";
  const ORDER = "VOTES";
  const DELAY_MS = 250;
  const MONTH = 5;
  const WEEKDAY = 3; // Wednesday. 0=Sunday, 1=Monday...
  const OCCURRENCE = 2; // Second Wednesday of May.
  const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
  function nthWeekdayOfMonth(year, month, weekday, occurrence) {
    const first = new Date(year, month - 1, 1);
    const offset = (weekday - first.getDay() + 7) % 7;
    return new Date(year, month - 1, 1 + offset + 7 * (occurrence - 1));
  }
  function gqlUrl({ year, month, day, cursor }) {
    const variables = { featured: false, year, month, day, order: ORDER, ...(cursor ? { cursor } : {}) };
    const extensions = { persistedQuery: { version: 1, sha256Hash: HASH } };
    return `/frontend/graphql?${new URLSearchParams({
      operationName: "LeaderboardDailyPage",
      variables: JSON.stringify(variables),
      extensions: JSON.stringify(extensions),
    })}`;
  }
  async function fetchPage(params) {
    const res = await fetch(gqlUrl(params), {
      credentials: "include",
      headers: {
        "content-type": "application/json",
        "x-requested-with": "XMLHttpRequest",
        "x-ph-timezone": Intl.DateTimeFormat().resolvedOptions().timeZone,
      },
    });
    if (!res.ok) throw new Error(`${params.year}: HTTP ${res.status}`);
    const json = await res.json();
    if (json.errors?.length) throw new Error(`${params.year}: ${json.errors.map((e) => e.message).join("; ")}`);
    return json;
  }
  function findConnections(value, path = []) {
    if (!value || typeof value !== "object") return [];
    const out = [];
    if (!Array.isArray(value)) {
      const edges = Array.isArray(value.edges) ? value.edges : null;
      const nodes = Array.isArray(value.nodes) ? value.nodes : null;
      const pageInfo = value.pageInfo;
      if (pageInfo && (edges || nodes)) {
        const items = edges
          ? edges.map((edge) => ({ node: edge.node ?? edge, cursor: edge.cursor }))
          : nodes.map((node) => ({ node, cursor: null }));
        out.push({ path, items, pageInfo });
      }
    }
    for (const [key, child] of Object.entries(value)) {
      if (child && typeof child === "object") out.push(...findConnections(child, path.concat(key)));
    }
    return out;
  }
  function looksLikeLaunch(item) {
    const n = item?.node?.post || item?.node;
    if (!n || typeof n !== "object") return false;
    return Boolean(n.id || n.slug || n.name || n.tagline || n.votesCount || n.commentsCount || n.product);
  }
  function hasLaunchDate(item, yyyyMmDd) {
    const n = item?.node?.post || item?.node || {};
    const dates = [n.featuredAt, n.scheduledAt, n.createdAt].filter(Boolean);
    return dates.some((value) => String(value).startsWith(yyyyMmDd));
  }
  function pickLaunchConnection(json) {
    return findConnections(json.data)
      .map((connection) => {
        const path = connection.path.join(".");
        let score = 0;
        if (/post|launch|leaderboard/i.test(path)) score += 20;
        score += connection.items.filter(looksLikeLaunch).length;
        if ("hasNextPage" in connection.pageInfo) score += 5;
        if ("endCursor" in connection.pageInfo) score += 5;
        return { ...connection, score };
      })
      .sort((a, b) => b.score - a.score)[0];
  }
  function normalizeLaunch(item) {
    const n = item.node.post || item.node;
    const product = n.product || {};
    return {
      id: n.id || n.databaseId || n.slug || product.id || product.slug || item.cursor || JSON.stringify(n).slice(0, 200),
      name: n.name || product.name || "",
      tagline: n.tagline || n.description || product.tagline || "",
      slug: n.slug || product.slug || "",
      votes: n.votesCount ?? n.voteCount ?? n.votes ?? "",
      comments: n.commentsCount ?? n.commentCount ?? "",
    };
  }
  const rows = [];
  const allLaunches = [];
  for (let year = FROM; year <= TO; year++) {
    const date = nthWeekdayOfMonth(year, MONTH, WEEKDAY, OCCURRENCE);
    const month = date.getMonth() + 1;
    const day = date.getDate();
    const yyyyMmDd = `${year}-${String(month).padStart(2, "0")}-${String(day).padStart(2, "0")}`;
    const seen = new Map();
    let cursor = null;
    let page = 0;
    while (true) {
      page += 1;
      const json = await fetchPage({ year, month, day, cursor });
      const connection = pickLaunchConnection(json);
      if (!connection) {
        console.warn(`${year}: no launch connection found`, json);
        break;
      }
      for (const item of connection.items) {
        if (!looksLikeLaunch(item)) continue;
        if (!hasLaunchDate(item, yyyyMmDd)) continue;
        const launch = normalizeLaunch(item);
        seen.set(launch.id, { year, date, ...launch });
      }
      const nextCursor = connection.pageInfo.endCursor;
      const hasNext = Boolean(connection.pageInfo.hasNextPage);
      if (!hasNext || !nextCursor || nextCursor === cursor) break;
      cursor = nextCursor;
      await sleep(DELAY_MS);
      if (page > 150) throw new Error(`${year}: stopped after 150 pages`);
    }
    rows.push({
      year,
      date: yyyyMmDd,
      weekday: date.toLocaleDateString("en-US", { weekday: "long" }),
      launches: seen.size,
      pagesFetched: page,
      url: `${location.origin}/leaderboard/daily/${year}/${month}/${day}/all`,
    });
    allLaunches.push(...seen.values());
    console.log(`${year} ${yyyyMmDd}: ${seen.size} launches`);
    await sleep(DELAY_MS);
  }
  console.table(rows);
  window.phLaunchCounts = rows;
  window.phLaunches = allLaunches;
  const csv = [
    "year,date,weekday,launches,pagesFetched,url",
    ...rows.map((row) =>
      [row.year, row.date, row.weekday, row.launches, row.pagesFetched, row.url]
        .map((value) => `"${String(value).replaceAll('"', '""')}"`)
        .join(",")
    ),
  ].join("\n");
  await navigator.clipboard.writeText(csv).catch(() => console.log(csv));
  console.log("Done. Results are in window.phLaunchCounts and window.phLaunches.");
})();

Sources

Product Hunt daily leaderboard pages used for the scrape, example: https://www.producthunt.com/leaderboard/daily/2026/5/13/all
Product Hunt GraphQL endpoint used by infinite scroll: https://www.producthunt.com/frontend/graphql?operationName=LeaderboardDailyPage
GitHub availability update, April 2026: https://github.blog/news-insights/company-news/an-update-on-github-availability/
GitHub Octoverse 2025: https://github.blog/news-insights/octoverse/octoverse-a-new-developer-joins-github-every-second-as-ai-leads-typescript-to-1/
Product Hunt newsletter, “All the AI that launched in 2025”: https://www.producthunt.com/newsletters/archive/46593-all-the-ai-that-launched-in-2025
Paper using Product Hunt launches, “Generative AI Fuels Solo Entrepreneurship, but Teams Still Lead at the Top”: https://arxiv.org/abs/2605.10291

Product Hunt is swamped now

Method

This is AI-related

Reproduce it

Sources