Investment Thesis · Subscription Infrastructure

The infrastructure repricing

AI is forcing SaaS to evolve. The companies that modernize their payment infrastructure will re-rate. The rest won't survive the transition.

Santiago Roel Santos  ·  February 2026  ·  12 min read

$0B
Lost to payment churn annually
0×
Median public SaaS EV/Rev
0%
Of failures never recovered
0-3×
NRR premium above 120%
CORE THESIS

The SaaS market now rewards retention, margins, and revenue quality with 2-3x multiple premiums.[1] Payment infrastructure migration is the only single operational change that improves all four valuation-driving metrics simultaneously.

This article is a companion to our Payment Infrastructure Calculator, which models the specific dollar impact for individual businesses. What follows is the case for why we believe subscription SaaS is mispriced and why payment infrastructure is the value creation lever the market has not yet recognized.

01 · The Friction Tax

Payment infrastructure is eroding subscription unit economics.

Every subscription business running on traditional card rails pays an infrastructure tax that never appears on an invoice. It materializes as customers who silently disappear when a card expires, as 2.9% + $0.30 per transaction in processing fees, and as the entire dunning management industry that exists to compensate for infrastructure deficiencies with retry logic and email campaigns.

Stripe recovered $6.5 billion in failed payments through its Smart Retries system in 2024.[4] That is a single processor. The best dunning tools salvage 60-70% of failed payments; the industry average is closer to 47%.[5] The remainder is permanent revenue leakage: customers who were using the product, had no intention of canceling, and were lost because a piece of plastic expired every 3-5 years.[6]

$0B
Recovered by Stripe Smart Retries (2024)
One processor. Revenue that would have been permanently lost to payment failures.
0+
Decline codes a processor can return
Most are opaque to the merchant. 'Do not honor' provides no actionable information.
0%
Of failed payments never recovered
Industry median recovery rate is 47.6%. More than four in ten failed charges are permanent losses.
ARR Divergence Over 36 Months
Illustrative: same business, different infrastructure · 5,000 starting customers · $200/mo ARPU · +300 new/mo
MONTH 36
$5.6M
gap · 2,350 customers
$12M$14M$16M$18MStartM6M12M18M24M30M36$18.2M$12.6M
Stablecoin rails
Card rails
Drag to explore · Assumes 3.5% monthly voluntary churn, 5% payment failure rate, 56% dunning recovery · Full methodology in sources [17]

Same product. Same price. Same voluntary churn rate. Same customer acquisition rate. The only variable is the payment infrastructure. Over 36 months, the cumulative gap compounds to millions in lost ARR. This is not a product problem or a market problem. It is an infrastructure cost that accrues in every cohort, every month, permanently.

02 · The Rerating Moment

The market now rewards exactly the metrics that payment infrastructure suppresses.

The post-2022 SaaS rerating changed what drives multiples. Growth alone no longer commands premiums. The 2026 market rewards retention, margin quality, churn discipline, and revenue predictability. Companies clearing the Rule of 40 command multiples roughly 2x those that fall below.[7] NRR above 120% is one of the strongest predictors of premium valuations.[1]

Net Revenue Retention
>120%2-3× multiple premium
Gross Margin
>75%+150-200 bps
Logo Churn
<5% annualUnlocks premium tier
Revenue Auditability
Real-timeDe-risks underwriting

Payment infrastructure migration touches all four of these levers simultaneously. No product improvement, no go-to-market optimization, no cost reduction initiative moves NRR, gross margin, logo churn, and revenue auditability in a single deployment. This is why we underwrite payment infrastructure in every subscription deal we evaluate.

03 · The Pricing Shift

AI broke the pricing model. What comes next is harder to bill.

Per-seat pricing, the model that powered two decades of SaaS growth, is in measurable decline. Seat-based pricing dropped from 21% to 15% of B2B SaaS companies in 12 months, while hybrid pricing surged from 27% to 41%.[8] When an AI agent can do the work of five people, charging per seat penalizes the customer for efficiency and the vendor for delivering it.

Intercom moved from per-seat pricing to $0.99 per resolution for AI agents, down from $39 per agent seat.[9] HubSpot and Salesforce adopted credit-based pricing models in 2025.[10] Gartner projects that, in its best-case scenario, agentic AI could drive up to 30% of enterprise software revenue by 2035.[11] The pricing model is migrating, and every new model is harder to bill on card rails.

Payment Failure Surface by Pricing Model
Per-seat
1 fixed charge/mo
Native
Usage-based
Variable charges
Strained
Credit / token
Top-ups + metered draws
Fragile
Outcome-based
Event-triggered charges
Breaking
AI agent billing
Continuous compute
Incompatible

Smart contract rails: zero decline opportunities across all models. The payment either executes or it does not.

The industry is simultaneously making billing more complex and more frequent while running it on infrastructure designed for one fixed monthly charge. Each additional billing event on card rails is another opportunity for a decline code, another retry window, another point of involuntary churn. Smart contract rails are not just a solution to the churn problem. They are the native infrastructure for the pricing models the industry is migrating toward.

04 · The Architecture Gap

Stablecoins do not patch the failure mode. They eliminate it.

Credit cards were designed in the 1950s for discrete, in-person transactions. Recurring billing is a layer added after the fact: the merchant stores a credential and re-presents it on a schedule, hoping the credential remains valid and the authorization chain returns a success code. When any link breaks, the payment fails and the merchant has no recourse except retry.

Card Rails
Credentials expire every 3-5 years
2,000+ possible decline codes
3-7 intermediaries per transaction
Authorization chain fails at any link
Multiple retries over days per failure
Settlement: 2-3 business days
Processing: 2.9% + $0.30
Stablecoin Rails
Wallets do not expire
Binary: funds present or not
Direct wallet-to-wallet settlement
No intermediary authorization chain
Auto-pause if insufficient balance
Settlement: seconds
Processing: ~1-1.5%

Credit-based AI billing is converging on the same architecture as stablecoin subscription wallets: a pre-funded balance, programmatic draws against that balance, and automated replenishment. Card rails introduce a failure mode at the top-up charge. Stablecoin rails eliminate it. The infrastructure that AI-driven pricing models require is the infrastructure that stablecoins already provide.

05 · Why Now

The prerequisites converged in 2024-2025.

Production-grade billing[12]Oct 2025

Stripe launched stablecoin subscription payments via smart contract, with over 400 wallets supported. Payments settle in fiat through existing Stripe billing infrastructure. No parallel stack required. Merchants manage stablecoin and traditional subscriptions from the same dashboard.

Wallet UX parity[13]2025

Coinbase Smart Wallet creates a non-custodial wallet using a passkey (FaceID, fingerprint) in seconds. No seed phrase. No browser extension. No app download. The friction gap between entering a card number and connecting a wallet has compressed significantly.

Regulatory clarity[14]Jul 2025

The GENIUS Act established a federal framework for payment stablecoins, including a pathway for OCC-chartered payment stablecoin issuers. The regulatory ambiguity that kept institutional capital on the sideline has been substantially resolved.

06 · What Migration Looks Like

You keep Stripe. You add a second payment option.

Migration does not mean ripping out your billing stack. It does not mean asking your customers to buy cryptocurrency. It means adding stablecoin as a payment method alongside cards, the same way businesses added Apple Pay or ACH. Stripe already supports this natively. The merchant dashboard, invoicing, revenue recognition, and reporting stay the same. Payments settle in fiat. Your finance team never touches a token.

01
Add stablecoin checkout

Enable USDC as a payment method through Stripe. Your existing checkout flow gets a second button. Customers who prefer cards keep using cards. No forced migration, no disruption to current subscribers.

Engineering lift: a Stripe API flag.
02
Incentivize adoption

The savings are larger than they appear. Processing fees drop from 2.9% + $0.30 to ~1-1.5%, but that is the smaller number. The bigger one is the involuntary churn you no longer lose. A stablecoin customer who would have churned involuntarily in month 8 now pays for months 9 through 36 and beyond. The expected LTV recovery per customer dwarfs the processing savings. That means you can offer 5-10% off for stablecoin payment and still come out well ahead on a per-customer basis. The discount is funded by revenue you were previously losing to expired cards.

Discount is self-funding. LTV math does the selling.
03
Cohort diverges

The stablecoin cohort has zero involuntary churn from day one. No expired credentials, no decline codes, no retry windows. Over 12-24 months, the retention gap between your stablecoin and card cohorts becomes visible in the data. The blended metrics improve with every customer that switches.

No product change required.

The critical point: this is additive, not replacement. Card payments continue to work exactly as they do today. Stablecoin is a second rail that runs in parallel. Adoption is gradual, voluntary, and incentive-driven. There is no migration event. There is no switchover date. The economics do the work.

07 · The Thesis

Invest. Migrate. Re-rate.

AI is repricing SaaS — not killing it. New delivery models and new pricing require new payment infrastructure. That's the core thesis at Inversion. The SaaS companies that make this transition will be the winners of the next decade.

Value Creation Sequence
Day 1
Fee compression
Add stablecoin payment option. Processing drops from 2.9%+ to ~1-1.5%. Savings flow directly to EBITDA.
Month 3-6
Churn reduction
Stablecoin cohort: involuntary churn drops to zero. Retained customers compound revenue every subsequent month.
Month 6-12
Pricing flexibility
Usage-based and conditional pricing on stablecoin cohort. Better price-to-value alignment lifts ARPU and reduces voluntary churn.
Year 2-5
Re-rating
Improved NRR, lower churn, higher margins, auditable on-chain revenue. The multiple expands because the metrics justify it.

At current multiples, the re-rating from 5x to 7.5-8.5x contributes more enterprise value than the EBITDA improvement itself. That asymmetry is the entire thesis.

Run Your Numbers

Quantify the infrastructure cost for your business.

Five inputs. The calculator shows what payment-caused churn is costing you in lost revenue, compressed LTV, and suppressed margins.

Sources
[1]Bessemer Venture Partners, "Cloud Index" (2025). NRR >120% correlates with 2-3x EV/Revenue multiple premium. Software Equity Group corroborates ~63% premium at NRR >110%.
[2]Stripe standard pricing: 2.9% + $0.30 per successful charge. Stablecoin processing: 1.5% via Stripe (Dec 2025); <0.1% for direct on-chain settlement per FinchTrade.
[3]Butter Payments (2024): involuntary churn adds approximately 4% to annual revenue loss. Consistent with Recurly and ProfitWell benchmarks for B2B SaaS.
[4]Stripe.com (2024): Smart Retries recovered $6.5 billion in revenue that would otherwise have been lost to failed payments.
[5]Slicker, "2025 Dunning Benchmarks": industry median recovery rate 47.6%. Top-performing dunning tools recover 60-70%.
[6]Visa, Mastercard issuer guidelines: standard card expiry cycle is 3-5 years (36-60 months).
[7]The SaaS CFO (2025): Rule of 40 companies command roughly 2x the EV/Revenue multiple of those below the threshold. Flippa 2026 data corroborates.
[8]Growth Unhinged, "2025 State of B2B Monetization": seat-based pricing 21% to 15%, hybrid 27% to 41% year-over-year.
[9]Intercom Fin AI launch (2023): $0.99 per AI resolution vs. $39/seat. Pricing confirmed via Pilot.com analysis.
[10]SaaStr (2025): HubSpot credit limits and auto-upgrade pricing. Salesforce pricing restructure confirmed across multiple sources.
[11]Gartner press release (Aug 26, 2025): best-case scenario projection that agentic AI could drive up to 30% of enterprise software revenue by 2035.
[12]Stripe blog (Oct 14, 2025): stablecoin subscription payments launch with 400+ wallet support, fiat settlement through existing billing infrastructure.
[13]Coinbase Smart Wallet (2025): passkey-based non-custodial wallet creation. No seed phrase, no browser extension, no app download.
[14]GENIUS Act signed into law Jul 18, 2025. Establishes federal framework for payment stablecoins including OCC charter pathway.
[15]Recurly (Jan 2024): $129B annual payment churn estimate. Derived from 8.6% revenue recovery rate applied to ~$1.5T global subscription market.
[16]PublicSaaSCompanies.com (Jan 9, 2026): median public SaaS EV/Revenue multiple of 4.01x across 157 companies.
[17]ARR divergence chart methodology: 5,000 starting customers, $200/mo ARPU, +300 new customers/month (identical for both cohorts). Voluntary churn: 3.5%/mo (~35% annualized), typical for SMB SaaS at this price point per Recurly/ProfitWell benchmarks. Card payment failure rate: 5%/mo (industry average). Dunning recovery: 56% per Stripe Smart Retries [4]; 44% of failures are permanent, yielding 2.2% involuntary monthly churn (5% × 44%). Total gross monthly churn: 5.7% on card rails (3.5% voluntary + 2.2% involuntary), 3.5% on stablecoin rails (voluntary only, zero involuntary). Chart compares 100% card vs. 100% stablecoin as a structural ceiling, not a migration projection.

AI is repricing SaaS — not killing it.

New delivery models and new pricing require new payment infrastructure. That's the core thesis at Inversion. The SaaS companies that make this transition will be the winners of the next decade.

Open Calculator →Get in Touch
← Back to Insights