For 15 years the answer was automatic: rent it. In 2026 that's breaking down. Cloud costs got real, data became the moat vendors learn from, and AI made building cheap — three shifts that moved the optimal point back toward ownership for more workloads than conventional wisdom admits. An honest scorecard of both sides.
Self-hosting didn't get free. The things it wins on got more valuable, and the thing it lost on got cheaper.
For fifteen years, the answer to "where should we run our software?" was automatic: the cloud, someone else's SaaS, someone else's servers. Owning infrastructure was for dinosaurs. In 2026, that automatic answer is quietly breaking down — and a growing number of companies are asking a question that would have sounded backward two years ago: what if we ran it ourselves? Here's the honest case on both sides of self-hosted versus SaaS, and why the pendulum is swinging.
Why SaaS won in the first place
Give the incumbent its due, because the reasons were good. SaaS and cloud won because they removed real pain: no servers to rack, no updates to apply, no ops team to hire, someone else on the hook at 3am. You traded ownership for convenience, and for most of the last decade that was a brilliant trade. Speed mattered more than control, and renting was faster than building.
None of that stopped being true. The trade just stopped being obviously one-sided.
What changed the maths
Three forces are pushing companies to reconsider, and they're all intensifying at once.
Cost stopped being trivial. The era of cheap cloud is over. Compute is scarce, GPU pricing is volatile and now financialised, and the SaaS bill that was a rounding error at small scale becomes a serious line item at medium scale. "FinOps" — the discipline of controlling cloud spend — exists because the spend got big enough to need a discipline. When renting is expensive enough, owning starts to pencil out.
Data became the asset, and vendors learned from it. In the AI era, the data your business generates is the moat — and a growing worry is that when you run everything through a vendor's platform, that vendor can learn from your data, potentially folding your proprietary knowledge into a product it sells to others. Ownership stopped being an ideological preference and became a competitive one.
Building got cheap. The historical killer of self-hosting was effort: standing up your own stack meant months of undifferentiated plumbing. AI coding tools and modern source-available frameworks collapsed that cost. The thing that made renting obviously easier — that building was hard — is much less true than it was.
The honest scorecard
This isn't a case for self-hosting everything. It's a case for choosing deliberately, because each side genuinely wins on different axes.
SaaS still wins on: zero operational burden, someone else's uptime guarantee, instant setup, and not needing the skills to run infrastructure. For a two-person team without ops capability, or a workload that isn't core to your business, managed SaaS is often correct — paying someone to make a problem disappear is a legitimate trade.
Self-hosting wins on: cost at scale, data ownership and residency, no per-transaction platform cut, freedom from a vendor changing terms or pricing under you, and the ability to keep your customer relationship and your data on your own side of the line. The price is real: you run the server, you apply the updates, you own the 3am page.
The pendulum is swinging not because self-hosting became free — it didn't — but because the things it wins on (cost, data, control) got more valuable, and the thing it lost on (effort) got cheaper.
Where the modern option lives
The reason "self-hosted" no longer means "rebuild everything from scratch" is a new class of source-available, own-your-stack platforms that ship the plumbing pre-built. VBWD is a clean example of the category: a self-hosted, source-available framework with a backend, web and mobile clients, subscription billing, and an AI layer already assembled — so you get the ownership of self-hosting without the year of foundation-building that used to be its price. Your data lives in your own database, there's no platform transaction cut, and it's free for commercial use below a defined revenue threshold. The pitch isn't "self-host out of principle." It's "self-host because the maths finally works, and the tools finally exist." You can see how the pieces compose in the plugin catalogue and the architecture.
The honest caveat stands: it's still self-hosted, so someone runs it, and for some teams that cost outweighs the benefits. This is a "choose deliberately" argument, not a "rip out all your SaaS" one.
The read
The self-hosted-versus-SaaS question isn't ideological anymore, and it isn't settled the way it was in 2015. Cloud costs got real, data became the moat, and building got cheap — three shifts that quietly moved the optimal point back toward ownership for more workloads than the conventional wisdom admits. Most companies should still rent most things. But the reflexive "obviously SaaS" is over. In 2026, the smart move is to actually run the maths for each part of your stack — and to notice that, for the pieces where cost, data, and control matter, owning is a live option again in a way it hasn't been for a decade.
Analytical commentary on infrastructure trends; the right choice depends on your team, scale, and workload. The VBWD reference illustrates the self-hosted approach and is not an endorsement. Not investment advice.
Learn more about VBWD
VBWD is a self-hosted, source-available platform for building subscription products, marketplaces, and AI-powered apps. Explore it further:
- 🌐 Website and documentation: vbwd.cc — see the plugins, architecture, and developer docs.
- 💻 Source code and plugins on GitHub: github.com/VBWD-platform
- 🎥 Watch VBWD in action: demo video 1 and demo video 2
- 💼 Follow the project on LinkedIn: linkedin.com/company/vbwd https://vbwd.cc/blog/2026/vbwd/cms-ai-an-llm-port-built-into-your-content-system
No comments:
Post a Comment