The State of Self-Hosted Software Demand: September 2026
Nine self-hosted and privacy asks, 90 distinct people counted from public posts. The loudest: family backup (27 people). The widest gap: homelab management (12 asking, 4 rivals detected).
Self-hosting stopped being a hobby. We count software demand by tracking distinct people who publicly ask for the same thing, and the self-hosted and privacy cluster keeps growing louder: the nine biggest asks below add up to 90 different people, counted one by one from public posts.
Here's the September 2026 state of that demand, sorted by how many people are asking.
| What people want | People asking | Rivals detected | Read |
|---|---|---|---|
| Family data backup that just works | 27 | 13 | evidence |
| Self-hosted ticketing system | 14 | 14 | evidence |
| Homelab management in one place | 12 | 4 | evidence |
| A home server you can't hear | 9 | 7 | evidence |
| Private cloud storage | 6 | 26 | evidence |
| Android keyboard that doesn't phone home | 6 | 16 | evidence |
| Maps without data harvesting | 6 | 9 | evidence |
| Gmail exit, office suite included | 6 | 5 | evidence |
| Self-hosted music streaming | 4 | 13 | evidence |
The loudest ask is the most boring one
Family backup, 27 distinct people. That's one of the loudest signals in our entire dataset, any category. The verbatim tells you why the 13 detected rivals haven't closed it: "I don't need anything overly complicated — just a reliable way to back up both the physical servers and VMs." The market keeps shipping enterprise-grade complexity to people who want an appliance.
The widest gap: homelab management
Twelve people asking, only 4 rivals detected — the best demand-to-supply ratio on this list. The pain has a specific shape: every self-hosted tool manages itself, nothing manages the pile. One asker put it perfectly: "I just wanted to self-host Home Assistant, so how hard could it be?"
Crowded doesn't mean done
Private cloud storage has 26 detected rivals — the most crowded row here — and people are still publicly asking. Same pattern we've documented across the whole dataset: when the asking survives 26 shipping products, the products are failing at something the spec sheet doesn't show. Here it's usually trust and setup pain.
The de-Google exits are still understaffed
Mail, maps, keyboard — the same migration wave we flagged in August's snapshot is still running. The Gmail row is the one to watch: 6 people, only 5 rivals detected, and the ask is bigger than email ("I don't need Google Docs, but need something to substitute it"). Whoever ships the full exit path, not just the inbox, gets the wave.
How to read this
Counts are distinct humans, not posts or upvotes. "Rivals detected" is what our scan found shipping — a tool with no web presence won't show up, so low numbers mean "we found few," never "none exist." All numbers match what each linked page renders today, September 4, 2026, and they move as the dataset grows.
If you want the method for turning asks like these into a validated build decision: how we validate with public data.
From GapMine's live tracking of builder demand. Next self-hosted census when the data warrants it.