At platform scale the in-house layer moves up: buy the ledger, build the monetization platform
Multi-product incumbents do not follow the startup rule of building the meter and buying the biller. They take one of two routes: build the billing engine outright, or buy a Zuora or Stripe Billing ledger and build the catalog, entitlement and consumption layer above it. Across 432 companies the build share of decided billing entries rose to 26% (44 of 169), and the builders cluster at platform scale.
What's happening — and why
What's happening: growth-stage AI companies in this corpus mostly build the usage meter in-house and rent invoicing and payments from Stripe Billing, Orb or Metronome. This trend is about what happens once a company has several product lines and hundreds of millions in recurring fees. There, the in-house layer moves up the stack.
Two signal waves took the decided billing base from 74 to 169 stack entries, and the build share went from 18% to 26% (44 of 169). Of the 40 companies with an in-house billing entry, 21 company pages (17 distinct parent organizations) are platform-scale multi-product incumbents or frontier labs.
Route one, build the engine, is taken by web-native self-serve platforms: Microsoft's Commerce Transaction Platform, depended on for more than 20 years; Dropbox; Rippling, whose Monetization Platform manages over $700M in SaaS fees a year; monday.com; Sentry; Google Workspace; Shopify, which now meters and invoices third-party apps on its own engine; and OpenAI, which evaluated third-party usage-billing platforms and turned them down.
Route two, buy the ledger and build the layer, is taken by enterprise-contract incumbents. Zuora or Stripe Billing stays the system of record, and an in-house platform sits on top: Gusto's Commerce Platform (catalog, promotions, subscription, pricing experimentation, entitlements), Workday's Platform Consumption Console, Zoom's Enterprise Billing services and Intercom's Fin billing platform. Twilio runs both at once.
The dated arc points toward consolidation in-house. Superhuman (2026-09-12) is folding the fragmented billers of Docs, Go, Grammarly and Mail into one Monetization Platform it builds, the mirror image of Replicate's 2025 move from a homegrown meter to Metronome.
How it works
Evidence over time
14 supporting · 5 counter — hover or tap a point for detail, click to jump to the row.
Evidence
| Company | Date | What happened |
|---|---|---|
| Microsoft 365 Copilot | Oct 2021 | BUILDS THE ENGINE. The in-house Commerce Transaction Platform, "depended on heavily for more than 20 years", runs purchases across Office 365, Bing Ads, Windows Store and Dynamics. Microsoft's billing and rev-rec rail is built, not bought. |
| Dropbox | Jun 2023 | BUILDS THE ENGINE. "At Dropbox we were very in favor of building, not buying." The billing spine has been homegrown since the Series B era, and card processing is the only part bought. |
| Rippling | Dec 2025 | BUILDS THE ENGINE. The Monetization Platform team owns onboarding, contracts, invoicing and payments, "managing over $700M in SaaS fees annually". CRM is the only bought layer. |
| OpenAI | Feb 2026 | BUILDS THE ENGINE, after an explicit buy evaluation: "We evaluated third-party usage billing and metering platforms ... [they] didn't meet two critical requirements." It built real-time access, credit balances and usage tracking in-house. Stripe and RevenueCat are bought as payment rails only. |
| Shopify | May 2026 | BUILDS THE ENGINE AND RENTS IT OUT. The App Events API lets third-party apps "define meters in the Partner Dashboard, and Shopify handles aggregation, calculation, and invoicing". App charges land on the merchant's Shopify invoice. Shopify's own billing platform (2022-06-29 engineering post) is in-house. |
| monday.com | Jun 2026 | BUILDS THE ENGINE. "a cutting-edge, homegrown platform" runs the whole subscription lifecycle, and the team is now building "a self-serve monetization engine" to "track, price, and enforce usage for millions of users and autonomous agents in real-time." |
| Intercom | Jun 2026 | BUILDS ABOVE A BOUGHT BILLER. The team is "building and enhancing Fin's billing, subscription, invoicing, and metering capabilities" with Stripe Billing as an integration beneath it. The outcome-priced product gets its own in-house layer. |
| Gusto | Sep 2026 | BUYS THE LEDGER, BUILDS THE LAYER. NetSuite and Zuora (Zuora inferred from a 2026-09-03 req, jobs/7369003) sit underneath, while an in-house Commerce Platform owns "the product catalog, promotion engine, subscription, pricing experimentation platform, and entitlement services". |
| Workday | Aug 2026 | BUYS THE LEDGER, BUILDS THE LAYER. Zuora runs billing operations and revenue management. The in-house Platform Consumption Console, "the central nervous system for usage data", is being built for the Flex Credits consumption model. |
| Sentry | Aug 2026 | BUILDS THE ENGINE. An in-house Billing Platform team owns checkout and invoicing on top of an in-house usage pipeline "processing hundreds of billions of events daily", with three simultaneous Billing Platform reqs open. |
| Zoom | Sep 2026 | BUYS THE LEDGER, BUILDS THE LAYER. Zoom's first stack pass shows every system of record bought: Salesforce, Zuora (Zuora's own case study), Stripe, Adyen/PayPal and Gainsight. Zoom then builds the in-house Enterprise Billing microservices that stitch them together. |
| Google Workspace | Sep 2026 | BUILDS THE ENGINE. A Workspace Monetization org is "building core infrastructure for metering, billing, and invoicing based on SKUs and resource consumption," plus in-house quota and API-tiering controls. |
| Twilio | Sep 2026 | BOTH AT ONCE. An in-house Commerce Platform is "responsible for managing the billing lifecycle". A same-day req (jobs/8048661) staffs a Zuora billing engineer for enterprise quote-to-cash. The engine is built and the enterprise ledger is bought. |
| Superhuman | Sep 2026 | THE CONSOLIDATION ARC. "Consolidating billing, checkout, and payments ownership across all Superhuman business units: Docs, Go, Grammarly, and Mail, into a single Monetization Platform". It is migrating from multiple fragmented billing systems to one unified in-house stack, and Stripe is kept as the rail. This is the corpus's clearest dated case of a company that acquired its way to several billers and is replacing them with one it builds. |
Counterexamples
- Anthropic · Jun 2026 — Frontier scale and still buying the engine: Metronome for metering, plus Zuora, Stripe, NetSuite and Salesforce. It builds only a homegrown ledger on top. This is the strongest counterexample to "scale builds the engine", though it fits the buy-the-ledger, build-above-it half.
- Replicate · Sep 2025 — Ran the opposite arc: it tore out a homegrown metering-and-fraud stack for Metronome and shipped prepaid credits in 14 days. Build-to-buy at growth stage, the reverse of Superhuman's buy-to-build at platform stage.
- 15Five · Feb 2026 — A scaled seat-SaaS incumbent that buys the whole revenue engine (Zuora billing, Salesforce, Vitally) and meters nothing in-house. Some incumbents build no layer at all.
- Yellow.ai · Feb 2024 — Outcome- and usage-metered, and it bought BOTH layers. Zuora runs subscriptions, and when Zuora's metered layer "hit performance and feature limits at high usage" Yellow.ai added Togai's metered quote-to-cash on top rather than building one. This is the cleanest counterexample to the claim that consumption products force an in-house layer.
- Zendesk AI · Jan 2014 — Buys its quote-to-cash stack whole (Zuora billing, Salesforce CPQ). Its outcome-based AI pricing is being built INSIDE the bought CPQ rather than in a separate in-house platform.
Trivia
-
Three corpus pages (Superhuman, Coda and Grammarly) cite one and the same 2026-07-16 job posting as their billing evidence. It describes consolidating "multiple fragmented billing systems" across Docs, Go, Grammarly and Mail into a single in-house Monetization Platform.
-
Rippling's in-house Monetization Platform handles "over $700M in SaaS fees annually", per a 2025-12-07 req. Rippling's own pricing page publishes no price.
-
OpenAI (2026-02-13) is the only corpus company that says in writing it evaluated third-party usage-billing platforms and turned them down. They were "well-suited for invoicing and reporting, but didn't meet two critical requirements".
For buyers
If you are pricing a usage or AI-credit product inside a multi-product company, expect to own the catalog and entitlement layer even if you buy the invoice. The vendor decision is about the ledger, not the monetization logic. Scope a billing-platform evaluation accordingly: judge Zuora, Stripe Billing, Orb or Metronome on ledger, invoicing and revenue-recognition fit, and plan staffing for the layer above. The counterexamples say buying more is viable too. Anthropic buys Metronome and Zuora at frontier scale and builds only a ledger on top, and Yellow.ai added Togai on top of Zuora when Zuora's metered layer hit performance limits at high usage, rather than building one. OpenAI is the one corpus company on record explaining why it did not buy: third-party platforms were well-suited for invoicing and reporting but did not meet two critical requirements.
For vendors
For billing vendors, the corpus says the platform-scale sale is increasingly the ledger, not the whole stack. Route-two incumbents such as Gusto, Workday and Zoom keep Zuora as the system of record and build catalog, consumption and entitlement services above it, so API depth and clean integration with an in-house catalog matter more there than an all-in-one UI. Shopify shows the other direction: a platform that built its engine can rent it out, letting third-party apps define meters while Shopify handles aggregation, calculation and invoicing. And watch the consolidation risk. Superhuman is replacing several fragmented billers across four business units with one platform it builds, which is exactly how a bought biller loses an account after an acquisition.
Outlook — what to watch
The durable claim is narrower than 'scale builds the engine'. Where a platform-scale company's monetization layer is disclosed at all, its catalog, entitlement and consumption logic is built in-house far more often than it is bought, and the one dated consolidation in the window, Superhuman, moved toward building. The trend is falsified if two or more multi-product incumbents that meter usage are shown, not merely undisclosed, to run fully bought monetization stacks with no in-house catalog, entitlement or consumption layer, or if a second platform-scale company runs Replicate's build-to-buy arc on its billing engine. Four usage- or outcome-metered Zuora buyers (Klaviyo, Contentful, Zendesk, Otter) disclose no in-house layer yet, so they are the cases to watch.
Bottom line
Once a company runs several product lines, the part that decides what a customer is entitled to and what a unit costs is usually built in-house, whether the ledger underneath is bought or not. The build share of decided billing entries reached 26% at 432 companies, and the builders cluster at platform scale.
FAQ
Do large SaaS companies build or buy their billing systems?
At platform scale, more build than the growth-stage pattern suggests. Across 432 corpus companies, 26% of decided billing stack entries are in-house builds (44 of 169), up from 18%. Web-native platforms such as Microsoft, Dropbox, Rippling, monday.com, Sentry, Google Workspace, Shopify and OpenAI build the engine. Enterprise-contract incumbents such as Gusto, Workday and Zoom buy a Zuora ledger and build the layer above it.
What does 'buy the ledger, build the layer' mean?
A bought billing system (usually Zuora or Stripe Billing) stays the financial system of record for invoicing and revenue, while an in-house platform owns the product catalog, pricing, entitlements and consumption tracking. Gusto's Commerce Platform, for example, owns the product catalog, promotion engine, subscription, pricing experimentation platform and entitlement services, with NetSuite and Zuora underneath.
Why did OpenAI build its own usage billing?
In its 2026-02-13 post, OpenAI said it evaluated third-party usage billing and metering platforms and found them well-suited for invoicing and reporting, but they did not meet two critical requirements. It built real-time access, credit balances and usage tracking in-house, and buys Stripe and RevenueCat as payment rails only.
Are there large companies that buy the whole billing stack?
Yes, and they keep this from being a law. Anthropic buys Metronome, Zuora, Stripe, NetSuite and Salesforce and builds only a ledger on top. 15Five buys Zuora billing, Salesforce and Vitally and meters nothing in-house. Zendesk buys Zuora and Salesforce CPQ and builds its outcome-based AI pricing inside the bought CPQ.