In May 2021, when the trade press was declaring subscription pricing dead and consumption pricing its inevitable replacement, we argued against the consensus: the consumption-versus-subscription debate was the wrong fight, because the real lever is the value metric beneath either model. The billing mechanism was never the decision. The unit you charge on, and who controls it, was.
Five years on, the infrastructure market made that point concrete. The independent meter, the layer most vendors treated as neutral plumbing, has been bought up, while the one layer no acquirer can purchase, the pricing decision itself, sits exactly where it always did. That is the layer 2021 was about, and it is the layer that matters most now. Here is what this usage-based billing consolidation changes, and what to do about it.
TL;DR: In six months, the three leading independent usage-billing vendors were absorbed by payments and platform companies: Stripe closed its purchase of Metronome in January, Salesforce announced its acquisition of m3ter on June 8, and Adyen announced its $335 million acquisition of Orb on June 11. The infrastructure that operationalizes consumption pricing is now owned by companies that earn on the metered flow. That changes the economics of your billing layer, hardens the lock between your value metric and your systems, and makes the one layer that remains independent, the pricing decision itself, more valuable than it has ever been.
Three deals in six months
The independent usage-billing layer was built over the past five years on a clear promise: instrument any metric, bill any model, and keep the flexibility to change both as your pricing evolves. That layer was just consolidated.
Stripe completed its acquisition of Metronome, reported at roughly $1 billion, in January 2026. Salesforce announced a definitive agreement to acquire m3ter on June 8, folding its metering and rating into Salesforce’s revenue management suite. Three days later, Adyen announced a definitive agreement to acquire Orb, the usage-billing platform behind Vercel, Replit, and Supabase, for $335 million, with both June deals expected to close July 1.
Three different acquirers, one pattern: a payments or platform company bought the meter. Nobody bought the decision layer, because the decision layer is not a product you can acquire. It is a discipline.
The deal coverage has treated these as fintech consolidation stories. For software vendors, they are pricing stories, and the mechanism deserves a closer look than the M&A press has given it.
What a payments company buys when it buys a meter
Every computing era settles around a toll layer, the place where one party collects on everyone else’s activity. Operating systems taxed software distribution. Search taxed discovery. App stores took their thirty percent. Cloud providers metered egress. The AI era’s toll layer is the metering and billing stack, the machinery that counts consumption and turns it into an invoice, and these acquisitions are payments and platform companies buying position at that toll.
An independent billing vendor earns a software fee. Its revenue scales with the number of vendors it signs and, loosely, with the events it processes. Its commercial interest is in usage-based billing existing as a category. Crucially, it competes on flexibility: the ability to define custom metrics, restructure plans, and backtest pricing changes against a stored event stream was the differentiator these vendors sold, because losing a customer to a rival billing platform was a live threat. The independent layer had to make metric change easy. That was the product.
A payments platform earns a percentage of every dollar that moves across its rails. Its revenue is indexed to processed volume, not to software seats. When a payments platform owns the meter, the meter’s roadmap answers to an owner whose economics favor billing structures that generate frequent, automatic, on-rail collection. Consumption pricing is the best possible customer for a payment rail: it produces continuous transaction flow that grows with usage. A large annual invoice settled by wire transfer is the worst: one collection event, often off-rail entirely.
Here is a thought experiment that makes the shift concrete. Imagine a vendor that bills its retail customers on total inventory SKUs under management: a clean value metric that tracks the customer’s business scale. That metric produces one billing event per period. The meter computes SKUs times rate, an invoice goes out, and the customer pays by wire on net-60 terms. Ask what the new owner of that vendor’s billing platform gains from this deal, and the honest answer is almost nothing: no transaction flow, no retry optimization, no fraud scoring, no take. The acquisition economics only pay off on transaction-dense metrics. Which tells you exactly where the investment, the tooling, and the roadmap attention will go.
We call this infrastructure gravity: the metric menu a vendor offers its customers is shaped by what its infrastructure makes easy and what that infrastructure’s owner rewards. Yesterday the gravity came from a metering company that profited from metering anything. Today it comes from owners paid on the flow.
The metric lock just got a landlord
Most vendors cannot change their value metric without re-engineering the systems underneath it. The metric is wired into billing and entitlements, into quoting, into revenue recognition, into the contract paper itself. The meter you instrument becomes the metric you are stuck with. Vendors charge on the units they can bill, not the units they would choose with a free hand.
While the billing layer was independent, that lock had an escape valve. A vendor unhappy with what its billing platform could express had options: pressure the provider, or switch. Billing migrations are painful, but they are tooling decisions, and the competitive dynamics of the independent layer meant flexibility kept improving.
Now the meter is a strategic asset inside a payments platform. The stated direction of the Adyen acquisition is convergence toward a single infrastructure experience across billing and payments, and the announcement language describes the destination as “self-driving infrastructure for revenue optimization.” Read that phrase carefully: it describes pricing operations run autonomously from inside the payment stack, by a party with a financial position in what the pricing produces. Unwinding a metric no longer means swapping a billing tool. It means unwinding from a platform that may also be your payment processor, your dunning system, and, as these platforms extend credit products, your lender.
Pricing agility was already rare. It does not get more common when your billing system is a line item in a payment platform’s growth plan.
Who Controls Your Value Metric When the Billing Landlord Does?
If your value metric is wired into a consolidated billing platform, changing it requires the landlord’s permission. We can assess whether your licensing, packaging, and pricing can survive that constraint.
How pricing models drift when the infrastructure has an owner
None of this arrives as a memo announcing that your pricing model now serves someone else’s economics. It arrives as defaults. Three mechanisms to watch:
Erosion by hybridization. The billing platforms advertise the ability to combine usage-based pricing, fixed fees, and per-seat charges in a single plan. That flexibility is real, and it is also the on-ramp. When a vendor adds AI features, the path of least resistance in its billing stack is not to rethink the value metric. It is to keep the existing base and bolt metered AI credits on top. The clean metric does not get killed; it gets barnacled. Two renewals later, half the invoice rides a unit nobody chose deliberately. We documented the herding dynamics in our field guide to usage-based pricing backfires, and the GitHub Copilot move to credits shows how fast a category’s unit conventions migrate once a large vendor moves. A major conversation-intelligence platform’s June 2026 credit launch shows the pattern in real time: the per-seat base stays, and a shared credit pool layers on top to meter AI features, integrations, and API-driven workflows. The seats keep the humans; the meter takes the machines.
Roadmap neglect of value metrics. The features that made business-scale metrics workable, custom metric definitions, mid-cycle quantity adjustments, backtesting against a stored event stream, were the differentiators of independent metering companies fighting for vendor customers. Under platform ownership, those become maintenance surface. The features that deepen transaction-dense billing and payment convergence become the investment thesis. A vendor’s ability to honor a buyer’s request to re-cut a deal on a different metric quietly degrades, not because anyone vetoed it, but because the tooling stopped improving.
Menu convergence. Flat-fee conversions reduce a platform’s transaction take. Usage floors, overage clauses, and auto-escalators preserve it. As platform-side pricing tools, contract templates, and optimization recommendations tilt vendor deal desks, expect the contract structures offered across a category to converge on consumption with floors. Procurement teams reviewing dozens of vendor contracts a year are better positioned to spot this convergence early than any single vendor is: they see cross-vendor patterns from the buying side that a vendor, seeing only its own paper, cannot.
The metrics only the meter can see
There is a second asymmetry buried in this shift, and it is the one we expect procurement teams to find first.
A value metric tied to the customer’s business is usually verifiable by the customer. A retailer can count its own SKUs. A logistics company knows its own shipment volume. If the invoice is wrong, the buyer can prove it from systems the buyer controls.
Machine-side metrics are different. AI credits, tokens, metered events, and compute units are observable only by the meter. When a buyer disputes a consumption spike, and agentic workloads make spikes routine, because agents consume in patterns no human usage curve predicts, the usage record that adjudicates the dispute lives inside the vendor’s metering infrastructure. After this consolidation wave, that infrastructure increasingly belongs to a party that earns a percentage of the invoice when it is paid.
Enterprise buyers already know this problem from software license audits, and they have a sharp instinct for it. The drift from buyer-verifiable metrics toward meter-only metrics is not just a pricing change. It is a transfer of audit power from the buyer to the infrastructure, at the exact moment the infrastructure’s new owners took a financial position in the total. Vendors should expect sophisticated buyers to start asking who owns the meter, and to negotiate exportable usage records and audit rights in response. The vendors that can answer cleanly will close faster.
There is a deeper reframe here that sophisticated buyers feel before they name it. The old instinct in a billing dispute was buyer against vendor. After these deals, you do not fully control your own meter either: the definitions that compute the invoice answer to the platform beneath you. The adversary changed, and most contracts were written for the old one. AI is becoming the enterprise’s fastest-growing metered spend, and no weights-and-measures office stands behind the meter; the contract is the only certificate of accuracy anyone gets. That is the ground your renewal conversations will increasingly be fought on: who owns the meter that computes the bill, and whether its definitions still hold.
How a vendor protects its independence
The response is not to flee platform-owned infrastructure. The platforms are good at what they do, and on the day a deal closes nothing about your stack has changed. The leverage point is timing: every mechanism that makes unwinding expensive accrues gradually, through integration cycles and renewal terms. The cheapest moment to protect your independence is now, while the promises are fresh and the entanglement is shallow.
Four moves, in priority order:
Contract for exit before the entanglement. Export rights on raw usage events are table stakes; the trap is derived state. Commitment drawdown balances, credit wallets, proration history, what each customer currently owes: that is the data that makes billing migrations die, and export rights to it almost never appear in a standard order form. Add payment-credential portability while you are at it. Tokenized cards and debit mandates are tied to the processor, and re-collecting them mid-migration is the hardest part of leaving a payment platform. Almost nobody negotiates token portability at signing. It is cheap to ask for now and nearly impossible to retrofit.
Refuse cross-product coupling in commercial terms. Bundled pricing that conditions your payments rate on using the billing product, volume commitments that span both layers, termination clauses that tie one to the other: each reads as a discount at signing and converts into renegotiation leverage against you at renewal. Price each layer as if you could replace it alone, because one day you may need to.
Keep the pricing architecture documented outside the stack. Your value metric, entitlement rules, packaging boundaries, and discount logic should exist as an artifact you own, not as the sum of whatever is configured in the billing platform. If the only complete description of your pricing model is your vendor’s configuration screens, your vendor holds your pricing model.
Re-validate the metric on your own evidence, on a cadence. The drift mechanisms above work through defaults, and defaults win when nobody is checking. A metric reviewed against transaction evidence on a regular rhythm stays deliberately chosen. One that is not reviewed becomes a metric the infrastructure chose for you, retroactively.
The decision layer is the only independent layer left
Here is the part we care about most, because it is the layer where we work.
A pricing architecture is the output of three structural decisions: the licensing model, the packaging model, and the pricing model. The licensing model is the upstream-most of the three: the pairing of a value metric with entitlement rules that define what each unit grants. Everything downstream, including the meter, operates on what that decision has defined. We have made this argument since long before the current consolidation: the metric beneath the model is the lever, and the model’s shape, subscription, consumption, or hybrid, is a consequence, not a starting point.
The billing layer has spent two years telling vendors the opposite: that the model shape is settled, hybrid is inevitable, and the urgent decision is which infrastructure to onboard. That message always served the sellers of infrastructure more than the buyers of it. It serves them even better now that the infrastructure earns on the flow. When every layer that operationalizes consumption pricing is owned by someone who profits from consumption, “usage-based is inevitable” stops being a forecast and starts being a business model.
The countermeasure is sequencing, and it has not changed: validate the pricing architecture first, then instrument it. A meter purchased before the metric decision does not skip the decision. It makes the decision for you, on terms you did not examine, and then enforces it. Variable AI pricing that suppresses customer exploration is one well-documented version of that failure; an AI cost-control architecture that absorbs variance at the licensing layer is what the deliberate version looks like.
This is also why we keep the decision layer deliberately independent. We do not sell metering, we do not process payments, and we hold no position in which model shape wins. Our practice is Continuous Monetization: pricing iterated on the same cadence your team ships product, not a one-time project repeated every few years, with LevelSetter as the instrumentation that observes how your metric performs against actual transactions. The substrate tells you when the metric is drifting away from how customers produce value. What you bind customers to remains your decision, made with evidence, not inherited from whatever your billing stack made easy.
For the broader practice context, the pricing strategy consulting hub maps how the decision layer fits together.
If your AI features are forcing a pricing change this year, or your billing vendor just changed owners and you are wondering what that means for the model you run on it, the sequencing question is worth an outside read before the defaults answer it for you. Architecture first, instrumentation second is the sequence worth pressure-testing against your current stack: see our approach, then book a working session to stress-test your metric before the next renewal cycle locks it in deeper.
