Open-source billing gives you the code, self-hosting, and independence from any vendor's roadmap; hosted platforms give you managed infrastructure and faster launches. In 2026 — after the two leading metering specialists were acquired by processors — the independence argument strengthened, but self-hosting still costs real engineering time.
What each option actually is
Open-source billing means the billing software's source code is publicly available under an open-source license. The leading example is Lago: usage metering, rating, invoicing, prepaid credits and wallets, entitlements, and subscription management, with code published on GitHub under the AGPL license. You can self-host it on your own infrastructure or use the vendor's managed cloud. The defining properties: you can read the code, audit the rating logic, modify it, and — crucially — keep running it even if the vendor disappears.
Hosted platforms (Stripe Billing, Chargebee, Recurly, Maxio, and the now processor-owned Metronome and Orb) are closed-source SaaS: the vendor operates the infrastructure, you configure through dashboards and APIs. You get managed reliability, continuous updates, and support — and you accept that the code, the roadmap, and your data's home are the vendor's.
The comparison is not "free vs. paid." Open-source software is free to download; running it in production is not free. Hosted platforms charge subscription or volume-based fees; they also absorb operational costs you would otherwise pay in engineering time. The real comparison is where the costs sit and who controls the roadmap.
Why 2026 changed the calculus
Two acquisitions reframed this debate. Stripe completed its purchase of Metronome in January 2026 and Adyen's acquisition of Orb was announced on June 11 and closed on July 1, 2026 — the two most prominent independent usage-billing specialists, both now on payment processors' roadmaps, per industry reporting. Lago's team argued in their response to the Orb announcement that independence from any processor's roadmap is precisely their advantage — read it as an interested competitor's case, but the structural point stands on its own: buyers who chose a specialist partly for processor neutrality — keeping billing logic independent of any single payment processor — now need to ask where that logic lives long-term. For buyers with a five-plus-year horizon, "which payment processor will my billing logic be tied to?" is now an explicit evaluation question.
This does not make open source automatically right — it makes independence a priced feature rather than an assumed one. Which leads to the honest accounting both sides need.
The case for open-source billing
- Code visibility and auditability. Billing code has direct financial consequences — a rating bug is lost or overcharged revenue. Readable, auditable code matters for financial software in a way it does not for a CMS. AGPL licensing means the code stays open.
- Self-hosting for regulated environments. Healthcare (HIPAA), financial services (data residency), government contractors (security review) — for some buyers, billing data cannot leave their own infrastructure, full stop. Cloud-only platforms are disqualified regardless of features.
- Roadmap independence. No vendor can deprecate the metering engine, change pricing, or converge it with a processor's priorities. If the vendor company stopped operating, the codebase remains forkable — a different risk profile, not a risk-free one.
- Customization depth. Need custom deduplication logic, specific event-transformation rules, or billing logic in version control alongside product code? With open source, you implement it directly in the codebase instead of filing a feature request.
- Cost predictability at scale. Open-source self-hosted has no per-event or volume-based vendor fee; managed open-source clouds typically charge flat fees. At high event volumes, the economics vs. consumption-priced vendors can be significant.
The case for hosted platforms
- Time to value. A hosted platform goes from signup to first invoice in days; a self-hosted deployment is an infrastructure project. For teams whose billing needs are standard, the speed difference is decisive.
- No operational burden. Upgrades, scaling, failover, security patching, and 3 a.m. pages belong to the vendor. Self-hosting billing means owning a revenue-critical system's operations permanently.
- Depth of managed features. Sophisticated dunning, many gateway integrations, tax automation, revenue recognition tooling — hosted platforms have had years and large teams to build these. Match the feature list to your actual requirements before assuming parity.
- Support and accountability. When invoicing breaks on a hosted platform, there is a vendor with an SLA. When it breaks on your self-hosted deployment, there is your team.
- Ecosystem integration. Processor-owned stacks now offer genuinely valuable billing-payments feedback loops (authorization optimization, fraud signals from billing data). Independence has a price; so does integration.
The honest costs of self-hosting
"Open source eliminates vendor risk" is too clean — it trades vendor risk for operational risk. Budget honestly:
- Engineering time, permanently. Initial deployment plus ongoing upgrades, monitoring, incident response, and keeping pace with upstream releases. For a revenue-critical system, this is not intern work — it needs engineers who understand both the infrastructure and the billing domain.
- Upgrade complexity. Billing systems accumulate state (subscriptions, meters, wallets); upgrades must migrate that state without disrupting active billing. Test every upgrade against a production-like dataset.
- Feature lag. The managed specialists ship features continuously; your self-hosted deployment gets them when you upgrade, and custom features you build yourself are yours to maintain.
- Compliance is still yours. Self-hosting gives you data control, which regulators like — but PCI scope, audit evidence, and security posture are your responsibility, not the vendor's.
- The license matters. AGPL (Lago's license) has copyleft obligations that some companies' legal teams scrutinize — particularly around network use and modifications. Clear it with legal before committing, not after deploying.
A reasonable rule of thumb: self-hosting makes sense when you have the DevOps capability and a strategic reason (regulation, data control, processor independence, scale economics). If you have neither the team nor the reason, the hosted platform's fee is buying you something real.
Vendor risk: both directions
Every option carries vendor risk; they just carry different kinds:
| Risk | Open-source self-hosted | Hosted independent | Hosted processor-owned |
|---|---|---|---|
| Vendor disappears | Code survives; you operate it | Data export; migrate | Data export; migrate |
| Pricing changes against you | N/A (self-hosted) | Possible; contracts help | Possible; bundle leverage |
| Roadmap converges away from your needs | N/A — your fork, your roadmap | Possible | Most likely (processor priorities) |
| Operational failure | Your team, your pages | Vendor SLA | Vendor SLA |
| Forced migration | Never | Rare; usually acquisition-driven | Possible on processor timeline |
The 2026 lesson: acquisition-driven roadmap change is not hypothetical — it happened twice to the metering category in one year. That does not mean hosted is wrong; it means portability deserves explicit evaluation.
Side-by-side comparison
Drawing on the published Lago–Orb comparison (written by Lago, so read it as an interested party's honest accounting rather than neutral truth):
| Dimension | Open-source (Lago as example) | Hosted specialist (as example) |
|---|---|---|
| Source code | AGPL, fully auditable | Proprietary |
| Self-hosted deployment | Yes | No (cloud-only typical) |
| Event ingestion | REST, batch, SDK, streaming | Strong — API, streaming |
| Prepaid credits/wallets | Native, real-time enforcement | Strong, managed |
| Pricing rule definition | JSON API + UI | Varies (e.g., SQL-like syntax) |
| Revenue recognition | Integrations; lighter native depth | Enterprise-grade native (varies) |
| PSP neutrality | Strong — processor-independent | Watchlist post-2026 acquisitions |
| Pricing model | Free (self-hosted) / flat cloud fee | Custom/quote or volume-based |
| Who operates it | Your team (self-hosted) | The vendor |
Decision framework
Choose open-source self-hosted when: regulation or security policy requires data on your infrastructure; processor independence is a strategic requirement; you have DevOps capacity and want billing logic in version control; or your event volumes make flat economics beat consumption pricing.
Choose hosted independent when: you want managed operations without processor entanglement; your billing needs are standard-to-complex but not exotic; and the vendor's export story is solid.
Choose hosted processor-owned when: you are committed to that processor's ecosystem; the billing-payments integration (authorization lift, unified data) is worth the roadmap coupling; and you have evaluated the neutrality trade deliberately.
Revisit annually. The market is consolidating; today's independent vendor can be next year's acquisition. Your portability answers should be re-verified at every renewal, not just at selection.
For the broader 2026 landscape, see our market overview; for metering specifics, our AI metering guide; and for selection criteria, our buyer's checklist.
- 2026 made independence a priced feature: the two leading metering specialists are now inside Stripe and Adyen — processor neutrality must be evaluated explicitly.
- Open source buys code visibility, self-hosting, and roadmap independence — at the cost of permanent operational ownership.
- Hosted platforms buy speed, managed operations, and feature depth — at the cost of roadmap dependence and migration friction.
- "Open source eliminates vendor risk" is false — it converts vendor risk into operational risk. Budget the engineering honestly.
- Portability is the tiebreaker: demand concrete answers on data and logic export from every vendor, and re-verify at renewal.
Frequently asked questions
Is open-source billing really free?
The software is free to download and use; running it in production is not free. Budget engineering time for deployment, upgrades, monitoring, and incident response — permanently. Managed open-source clouds charge flat fees that split the difference.
What does AGPL mean for billing software?
AGPL is a copyleft open-source license: the code is public and auditable, and if you modify it and offer it over a network, you generally must share your modifications. Most companies using (not modifying-and-distributing) billing software are unaffected, but clear it with legal before committing.
Can self-hosted billing handle high usage volumes?
Yes — self-hosted metering pipelines are designed for high event volumes; the constraint is your infrastructure and operations, not the license model. The economics often favor self-hosting at scale versus per-event vendor pricing.
What is PSP neutrality and why does it matter in 2026?
Payment-service-provider neutrality means your billing logic is not tied to a specific payment processor. It matters more since 2026 because the two leading metering specialists were acquired by processors (Metronome→Stripe, Orb→Adyen), putting their roadmaps under processor priorities. If you might change processors, neutrality protects your options.
Which is better for a startup: open-source or hosted billing?
Most startups should start hosted — speed to first invoice beats architectural purity, and few early teams have spare DevOps capacity for billing infrastructure. Revisit when regulation, scale economics, or processor-independence requirements emerge.
How do I avoid lock-in with a hosted billing platform?
Verify data export (customers, subscriptions, meters, wallets, full history) and billing-logic export (rating rules, pricing config) before signing; prefer platforms with documented migration tooling; and re-verify portability at each renewal — especially in a consolidating market.
Conclusion
The 2026 reality check is not that open source won or hosted lost — it is that the choice got sharper. Processor-owned metering is powerful and deeply integrated; independent and open-source options are the counterweight for buyers who need their billing logic to outlive any vendor's strategy. Make the choice with both cost columns filled in — the vendor's fee and your engineering time — with the license cleared by legal, and with portability verified rather than assumed. Do that, and either path is defensible. For context, see our 2026 market overview, AI metering guide, and buyer's checklist.
—
Market developments verified on cited pages in September 2026; confirm current product status with vendors. Research-based guide; no vendor testing involved.