Your GRR Benchmark Is Gone. The 2026 Data Shows SaaS Lost Four Retention Points — and Pricing Model Is Why.
Orb built the billing infrastructure that AI companies like Vercel, Glean, and Replit use to handle usage-based pricing at scale. Adyen's July 2026 acquisition reveals that billing has become the underrated AI monetization moat — and the race to control it has started.
On July 1, 2026, Adyen completed its $335 million all-cash acquisition of Orb, a San Francisco-based startup that builds pricing, billing, and revenue intelligence software for AI and SaaS companies. The press coverage framed it as a fintech story: a payments giant buying a billing startup, vertical integration, one more move in the consolidation of financial infrastructure. That framing missed the point entirely.
Adyen did not acquire Orb to expand its payment processing capabilities. Adyen acquired the infrastructure layer that makes AI pricing models commercially viable at scale. The distinction matters for every AI company deciding how to monetize in 2026 — and for every founder who still thinks of billing as a back-office function rather than a strategic moat.
What Orb Actually Built
Orb is not a traditional billing platform. The established players in SaaS billing — Chargebee, Recurly, Zuora, and Maxio — were built for a world where pricing meant monthly or annual subscriptions, seat counts, and usage tiers defined in advance. They handle flat recurring charges well. They handle real-time metering at AI scale very poorly.
Orb built a fundamentally different architecture, designed from the start for consumption-based pricing at high event volumes. Its three core capabilities are:
Real-time event ingestion at scale. Orb can ingest millions of usage events per hour — API calls, tokens consumed, tasks completed, queries answered — and immediately translate those events into accurate billing data. This is technically tractable for a startup billing a few hundred customers. It is technically very hard for a company billing thousands of enterprise customers with different pricing agreements simultaneously, without delays, errors, or reconciliation backlogs.
A flexible price configuration engine. AI companies price in ways that legacy billing systems cannot model: per-thousand-token prices that vary by model and context window, per-task prices that differ by task type and complexity, hybrid structures combining a base subscription with usage overages, enterprise-specific volume discounts that reset monthly, outcome-based pricing that triggers on verified results. Orb's price configuration layer can represent all of these simultaneously and apply them accurately to each customer's usage data in real time. Adding a new pricing tier or changing a rate card does not require an engineering sprint.
Revenue intelligence. Orb translates metered usage data into forward revenue projections, customer health signals, expansion opportunity identification, and churn risk indicators. The billing system becomes a revenue intelligence layer — a source of commercial insight — not just a transaction processor. Finance teams can forecast accurately because the billing data reflects actual usage in real time. Sales teams can identify which customers are approaching usage limits and should be upgraded.
Orb's customer list at the time of acquisition included Vercel, Glean, Replit, and Supabase — all usage-intensive AI and developer platforms with complex pricing models that legacy billing systems cannot handle without significant custom engineering. As Adyen's press release stated, the acquisition price reflects the technical moat Orb built in the years when traditional billing vendors assumed usage-based pricing was a niche.
The Usage-Based Surge That Made This Acquisition Inevitable
The strategic context for the Adyen-Orb deal is the acceleration of usage-based pricing across the software industry. In 2021, 27% of public SaaS companies had a usage-based component in their pricing. By 2026, that number has crossed 51%. The shift to AI-powered products has made usage-based pricing not just common but structurally necessary.
Per-seat pricing rests on the assumption that pricing should scale with the number of users doing a roughly equivalent amount of work. AI shatters both assumptions simultaneously. A single prompt engineer can produce the output of a five-person team. A single sales representative using an AI agent can run ten times the sequences they ran manually. Charging per seat when per-seat value has increased by a factor of ten is a way to systematically undercharge your best customers and structurally misalign your revenue with your value delivery.
As Signal has documented extensively, per-token, per-call, and per-outcome pricing models are replacing seat counts as the primary unit of commercial exchange in AI-native SaaS. The 2026 AI pricing crisis — where companies face simultaneous pressure from falling model costs and customer demand for value-based pricing — is collapsing the economics of fixed subscription contracts for AI products. The billing systems designed for fixed subscriptions cannot handle what comes next.
Orb's acquisition price reflects the cost of owning the infrastructure that can.
Why Billing Is the Underrated AI Infrastructure Layer
The billing layer is the system that converts a company's pricing decisions into actual revenue. It sounds administrative. It is, in practice, the constraint that limits how creatively you can price your product.
Consider the situation facing a typical AI company in 2026. It has three or four pricing tiers, a per-token component for API usage, an enterprise tier with custom discounts and usage caps, a startup tier with discounted rates, a handful of beta customers on outcome-based contracts, and a free tier with conversion targets. Its legacy billing system — built for subscription management — can handle one of these models at a time. To support all of them simultaneously, the company has either built a custom billing layer in-house at significant engineering cost, or grafted workarounds onto a subscription billing tool not designed for the purpose.
The workarounds have downstream costs that compound over time. Revenue recognition becomes inconsistent. Customer-facing invoices become inaccurate or delayed, generating support tickets and trust erosion. Sales cannot offer creative pricing without engineering involvement to model the billing impact. Finance cannot forecast accurately because the billing data does not reflect actual usage in real time. The CEO hears "we can't price it that way" from engineering when the real answer is "our billing system can't support that pricing."
The companies that get billing infrastructure right — that invest in metering and pricing engines designed for consumption-based products — gain the ability to price in ways that create real competitive differentiation. Outcome-based pricing — charging customers per successful resolution, per completed task, per verified result — is only commercially viable if you have billing infrastructure that can measure and verify the outcomes at scale. Most AI companies in 2026 do not have it.
Orb provided that infrastructure. Adyen's willingness to pay $335 million for it is a data point about what the infrastructure is worth to the companies that have it versus the companies that do not.
The Competitive Landscape After the Acquisition
The Lago blog post responding to the Adyen-Orb deal is worth reading for what it reveals about the competitive dynamics. Lago — an open-source alternative to Orb — positioned the acquisition as proof that AI companies should choose open-source billing infrastructure rather than proprietary platforms owned by payment processors. The reasoning: a billing system acquired by Adyen may eventually be optimized for Adyen's payment rails and commercial interests rather than the billing flexibility its customers need.
The argument is not implausible. Adyen's business model is transaction-based. Once Orb is fully integrated into the Adyen stack, the commercial pressure to route transactions through Adyen's payment rails — and to deprioritize features that benefit customers using other processors — may reshape the product roadmap in ways that serve Adyen's margins rather than Orb's historical customers' flexibility.
This tension creates an opening for open-source and independent billing infrastructure. The market is bifurcating in a pattern familiar from other infrastructure acquisitions: enterprise consolidation on one side (Adyen/Orb), open-source independence on the other (Lago, Metronome, and new entrants).
| Approach | Best For | Trade-offs |
|---|---|---|
| Adyen + Orb | Companies already using Adyen for payments | Deep integration; potential lock-in to Adyen's commercial roadmap |
| Stripe Billing | Companies on Stripe; improving metering | Not AI-native at scale; better for simpler usage models |
| Lago (open source) | Companies needing maximum flexibility and control | Engineering maintenance overhead; no managed service |
| Metronome | AI-native companies wanting usage-based depth | Growing fast; used by AI infrastructure leaders |
| Build in-house | Companies with highly specific metering needs | High cost, high long-term maintenance |
The acquisition does not eliminate the competition. It clarifies the market structure and creates urgency for AI companies that have not yet made an infrastructure decision to make one.
The Pattern Behind the Deal
The Adyen-Orb acquisition is not an isolated event. It is part of a broader consolidation of AI infrastructure layers that are becoming strategic assets:
The inference infrastructure layer is consolidating around scale and specialization — Fireworks AI raised $1.5 billion, Baseten raised $1.5 billion, Together AI continues growing — as companies recognize that the managed infrastructure for running AI models at production scale is a distinct and valuable business.
The billing infrastructure layer is now consolidating around AI pricing complexity — Orb's acquisition is the signal event, but Metronome's continued growth and Lago's open-source traction indicate the market is real.
The data infrastructure layer has been consolidating for years — Databricks at over $60 billion, Snowflake maintaining enterprise dominance — driven by the recognition that the systems storing and querying the data that trains and serves AI models are indispensable.
Each of these infrastructure layers was a commodity or irrelevant when AI was a feature in someone else's product. Each is becoming a strategic asset as AI becomes the product itself. Stripe's evolution toward becoming a financial infrastructure layer is the canonical earlier example: the infrastructure that processes transactions becomes indispensable, acquires adjacent capabilities, and becomes the moat against which competitors must price.
Billing infrastructure is following the same trajectory. Adyen paid $335 million to own this layer in 2026. The market price for owning the equivalent layer in 2028 will be higher, because the companies that do not own it will have spent two more years accumulating billing debt that limits their pricing flexibility and constrains their revenue intelligence.
What AI Companies Should Do Right Now
The Adyen-Orb deal is a market signal, not a deadline. But it indicates the billing infrastructure market is entering a consolidation phase — which means the window for making a deliberate infrastructure choice, rather than defaulting to whatever is easiest to integrate with your current payment processor, is narrowing.
1. Audit your billing system for usage-based compatibility. Can your current system handle real-time event ingestion at your actual API call volume? Can it represent your complete pricing model — all tiers, all overages, all custom enterprise contracts — without workarounds? If the answer to either question is no, you are carrying billing debt that is limiting your pricing options.
2. Quantify the revenue leakage from billing limitations. Companies with billing systems that cannot accurately meter usage typically undercharge customers by 8–15% of actual usage. A billing audit often reveals both revenue leakage — usage that occurred but was not billed — and invoice inaccuracies that generate support costs and customer trust erosion.
3. Evaluate the payment processor concentration risk. If you use Adyen for payments and now also use Orb/Adyen for billing, you have significant commercial concentration in one vendor. Model what happens to your billing flexibility if Adyen's product roadmap diverges from your requirements.
4. Consider open-source alternatives before the window closes. Lago's positioning as the independence-preserving alternative may appeal to companies that want usage-based billing depth without lock-in to a payment processor's commercial agenda. The trade-off is engineering maintenance. That trade-off is reasonable for companies with strong platform engineering teams. It is less reasonable for companies that cannot staff a billing infrastructure team.
5. Build toward outcome-based pricing as your next capability. The billing infrastructure that handles per-token and per-call metering today is the same infrastructure that will handle per-outcome pricing tomorrow. Companies that invest in metering capability now are building the foundation for the most defensible pricing architecture in AI: charging customers on verified results rather than on access or consumption.
Takeaway: The Adyen-Orb acquisition is a strategic signal about where AI monetization infrastructure is heading, not just a fintech transaction. Billing systems designed for flat subscriptions cannot handle AI's pricing complexity — per-token metering at millions of events per hour, outcome-based contracts that require outcome verification, real-time usage-to-invoice translation for thousands of enterprise customers simultaneously. The companies that own or can access this infrastructure layer have an asymmetric advantage in how they can price and monetize their AI products. Adyen paid $335 million to understand this. The AI companies that have not yet audited their billing infrastructure should take note.
Frequently Asked Questions
What did Adyen acquire and why?
Adyen acquired Orb, a San Francisco-based startup that builds pricing, billing, and revenue intelligence software for AI and SaaS companies, for $335 million in an all-cash transaction that closed on July 1, 2026. Orb's core capability is handling the complexity of usage-based pricing at scale — real-time event ingestion, flexible price configuration, and accurate invoice generation for companies with millions of API calls, token events, or task completions per day. Adyen's co-CEO stated the rationale plainly: 'The structural complexity of modern billing has become an infrastructure problem Adyen is built to take on.' The acquisition was not primarily about expanding payment processing capabilities. It was about owning the layer between AI product usage and AI product revenue — the infrastructure that determines whether a company can price its AI products creatively or is constrained to simple subscription tiers its legacy billing system can handle.
What is usage-based billing and why do AI companies need specialized infrastructure for it?
Usage-based billing (also called consumption pricing or metered billing) charges customers based on how much of a product they actually use — API calls, tokens consumed, tasks completed, data processed — rather than a flat monthly fee or per-seat license. Traditional billing platforms like Chargebee, Recurly, and Zuora were designed for predictable subscription and seat-based pricing. They handle monthly invoices and annual renewals well. They struggle with high-frequency metering, real-time usage translation, mid-period pricing changes, and the simultaneous management of complex pricing agreements across thousands of enterprise customers with different rate structures. AI companies generating millions of usage events per hour need billing infrastructure built for that volume and complexity. Orb built exactly that architecture — a real-time event ingestion system, a flexible price configuration engine, and revenue intelligence that converts metered data into actionable business signals. This infrastructure gap between traditional billing and AI pricing requirements is what the Adyen acquisition is ultimately about.
How does usage-based pricing affect SaaS retention and revenue growth?
Usage-based pricing companies post a 108% median net revenue retention (NRR) in 2026 benchmarks compared to 95% for seat-based models — a 13-point structural gap that compounds annually. Monthly attrition for usage-based models is 2.1% versus 3.9% for per-seat models, a 46% reduction. The mechanism is structural: usage-based pricing aligns the vendor's revenue with the customer's success. When AI makes each user more productive, usage increases and so does the customer's payment, creating a positive feedback loop between customer value and vendor revenue. Seat-based pricing creates the opposite dynamic: when AI reduces headcount, seat-based revenue falls even as workflow dependency deepens. As of 2026, 51% of public SaaS companies have a usage-based component in their pricing, up from 27% in 2021. The shift is driven by AI's effect on human productivity and by the accumulated evidence that usage-based pricing produces significantly better retention economics.
What are the alternatives to Orb for usage-based billing after the Adyen acquisition?
The main alternatives to Orb for usage-based billing infrastructure include Lago (open-source, self-hostable, positioned as the freedom-preserving alternative to Orb after the acquisition), Metronome (venture-backed, specifically designed for AI and infrastructure companies, growing rapidly with customers including Anthropic and Vercel), Stripe Billing (improving metering capabilities but not natively designed for AI-scale event volumes), Maxio (formerly Chargify and SaaSOptics, serving mid-market), and Zuora (enterprise billing, strong on revenue recognition but complex to implement for pure usage-based models). The Lago blog's response to the Adyen acquisition explicitly positioned it as proof that open-source billing infrastructure is needed to avoid vendor lock-in in a market where proprietary billing providers are being acquired by payment processors with their own commercial interests. Companies evaluating billing infrastructure in 2026 face a real trade-off: the depth and managed-service quality of proprietary platforms like Orb/Adyen versus the independence and flexibility of open-source alternatives like Lago.
What does the Adyen-Orb acquisition mean for AI companies choosing their billing infrastructure?
The acquisition sharpens a decision that was already urgent. Adyen acquiring Orb creates a consolidated billing and payments stack that offers deep integration advantages — if you already use Adyen for payment processing, the combined stack simplifies reconciliation and reduces vendors. But it also creates concentration risk: a billing system owned by a payment processor may eventually be optimized for Adyen's commercial interests rather than the billing flexibility its customers need. AI companies that do not use Adyen for payments gain less from the integration and face the standard lock-in risk. The more important implication is structural: the acquisition signals that usage-based billing infrastructure is now a strategic asset valuable enough to attract nine-figure acquisition prices. Companies still running complex usage-based pricing on top of subscription billing tools built for simple recurring charges are carrying increasing operational debt — billing workarounds that limit pricing creativity, cause revenue leakage, and produce inaccurate invoices. The Adyen-Orb deal is a market signal that the cost of that debt is rising.