TL;DR
~2× acceleration in shipping output-not per-dev productivity, but total stuff shipped.
Still accelerating, not slowing: Peaks continue into Jan 2026 (Show HN hit an all-time high). GitHub: record repos, commits, AI repos (+178% YoY), agents just went mainstream.
Interpretation: AI (Copilot → Cursor → agents) didn’t just make devs faster-it lowered the threshold to ship, so many more products get released.
Bottom line: AI caused a clear step-change in 2025 when AI crossed from “assistive” to “agentic + universal”. (~2× more shipping), and early-2026 data says the curve is still bending upward, not flattening.
What we’re trying to find out
We wanted to get a sense of whether AI is changing how much new software-and how many new products-actually get shipped. Not just hype or adoption surveys, but something that reflects volume of output: more launches, more repos, more public “ships.”
There’s no single metric for “impact of AI on software shipment,” so we didn’t try to measure it directly. Instead we looked at proxies-signals that tend to move when more people are building and shipping:
- Product Hunt - Counts of new product launches. A proxy for “new products brought to market” and maker/startup activity.
- Show HN (Hacker News) - Counts of “Show HN” posts. A proxy for developers and makers shipping in public.
- GitHub (Octoverse) - Total repositories and growth over time. A proxy for how much software is being created and maintained in the open.
- .ai domain registrations - Monthly registered .ai domains. A proxy for “AI” branding and new ventures betting on the TLD.
Each proxy is imperfect and can be driven by other factors (e.g. more PH launches could mean more marketing, not necessarily more products). Taken together, though, they give a picture of whether more new software and products are going out the door-and whether that acceleration lines up with the rise of AI tooling (Copilot, Cursor, agents, LLM SDKs). Below we lay out the data and what it suggests.
Key milestones: AI and development
| Month & year | What happened |
|---|---|
| June 2022 | GitHub Copilot generally available |
| November 2022 | ChatGPT (GPT-3.5) |
| July 2023 | Cursor started getting traction |
| June 2024 | Claude 3.5 |
| October 2024 | Agentic coding went mainstream |
| May 2025 | Claude Code went mainstream |
Product Hunt: more products launched
Product Hunt is a useful proxy for “new products launched” in the wild. Counts of launches by month show a clear ramp through 2025.
Numbers from the data:
- December 2024: 4,444 launches → December 2025: 9,903 launches (~123% year-over-year for that month).
- Full-year 2025 total: ~90,086 launches.
- Upward trend through 2025; peak months July (8,663), December (9,903), and January 2026 (10,629).
So by this measure, roughly twice as many products were launching at the end of 2025 as at the end of 2024, and January 2026 set a new high at 10,629 launches.
Show HN (YC Hacker News): more developers shipping in public
Show HN is a proxy for developers and makers shipping in public on Hacker News. Monthly counts of Show HN posts also show a strong increase through 2025.
Numbers from the data:
- December 2024: 1,622 posts → December 2025: 3,525 posts (~117% year-over-year for that month).
- Full-year 2025 total: ~27,844 Show HN posts.
- 2024 levels were ~1,400-1,600 per month; 2025 ramps to 2,200-2,600+ and December hits 3,525; January 2026 peaks at 4,782.
| Period | Posts (sample) |
|---|---|
| 2024 (e.g. May-Dec) | ~1,500-1,600/mo |
| 2025 (Jan-Dec) | 1,731 → 3,525 (ramp) |
| Dec 2025 | 3,525 |
| Jan 2026 | 4,782 (peak) |
So more people are “showing” new work on HN in 2025-and the trend continues into early 2026.
GitHub (Octoverse 2025 vs 2024): record repos, commits, and TypeScript #1
GitHub’s Octoverse 2025 and Octoverse 2024 reports give a third signal: record growth in repos, commits, and language use, with AI and TypeScript front and center. Total repositories on GitHub have grown from about 144M (2019) to 630M (2025), per Octoverse reports (2020, 2021, 2022, 2023, 2024, 2025).
2025 highlights (Octoverse 2025):
- 180M+ developers on GitHub; on average, more than one new developer joined per second.
- 230+ new repositories created per minute.
- 43.2M pull requests merged per month (+23% YoY); nearly 1 billion commits in 2025 (+25.1% YoY).
- TypeScript became the most used language on GitHub (overtook Python and JavaScript).
- 4.3M+ AI-related repositories; 1.1M+ public repos with an LLM SDK (+178% YoY).
- GitHub Copilot Free and coding agents widely adopted; ~80% of new developers use Copilot in their first week.
2024 context (Octoverse 2024):
- 518M projects; nearly 1 billion contributions to public and open source.
- Python overtook JavaScript as most used language; 137k public generative AI projects (+98% YoY); Jupyter Notebooks surged.
So GitHub is seeing more repos, more commits, more PRs, and a structural shift to TypeScript and AI-related projects-all consistent with more software being written and shipped, and with AI tooling as part of the stack.
.ai TLD: monthly registered domains
The .ai top-level domain (Anguilla’s country-code TLD) has grown sharply as a proxy for “AI” branding. Monthly totals of registered .ai domains show strong growth through 2024.
Numbers from the data (Domain Name Stat):
- December 2023: ~354k registered .ai domains → November 2024:
573k (62% year-over-year). - December 2024: ~573k (Identity Digital took over registry management that month).
- About 40k registrations in the two months Oct-Nov 2024; growth of ~219k domains over the 12 months to Nov 2024.
So by this measure, .ai registrations accelerated through 2024-consistent with more products and brands adopting an “AI” identity.
Conclusion
Three different views-Product Hunt launches, Show HN posts, and GitHub’s Octoverse-all point the same way: 2025 saw a big step-up in how much new software and how many new products were built and shipped. Product Hunt launches roughly doubled (Dec-to-Dec ~123%) and hit a new peak in Jan 2026 (10,629); Show HN posts roughly doubled (Dec-to-Dec ~117%); GitHub hit record repos, commits, and PRs and put TypeScript and AI repos at the center of the story. AI (Copilot, agents, LLM SDKs) doesn’t explain everything, but it correlates strongly with this acceleration. If that correlation holds, we should expect even more new software and products in the years ahead.