AI Underwriting Is Changing Who Gets Funded

For most of its history, merchant cash advance underwriting has run on a fairly narrow set of inputs: a handful of months of bank statements, a personal credit score, and a judgment call about whether a business's card sales looked stable enough to support daily repayment. In 2026, that model is being rapidly displaced — and the shift has real implications for who gets funded, at what price, and how competitive advantage is built in alternative lending.

From Bank Statements to Thousands of Data Points

Leading MCA and alternative-finance providers now deploy machine-learning models that evaluate thousands of data points per applicant — transaction-level bank account patterns, industry-specific benchmarks, supply chain signals, online review trends, and even macroeconomic indicators — to generate underwriting decisions within seconds rather than days. That is a meaningfully different underwriting process than the one that defined the industry for most of the last decade, and it's already changing approval economics.

Current approval rates for MCA products sit somewhere in the 65% to 85% range depending on the source and provider, dramatically higher than the roughly 25% to 30% approval rate typical of conventional bank small-business loans. That gap has always existed because MCA underwriting tolerates more risk in exchange for higher pricing — but AI-driven models are increasingly allowing providers to price that risk more precisely rather than applying a blunt, one-size-fits-all factor rate. In principle, that means lower-risk borrowers can be identified and offered comparatively better terms, while higher-risk segments are priced — and monitored — more accurately, rather than every applicant absorbing the same risk premium.

Data Access Is Becoming the New Competitive Edge

This underwriting shift is being reinforced by a parallel regulatory development: the expansion of open banking in the United States, driven by the CFPB's rulemaking under Section 1033 of the Dodd-Frank Act, which is pushing financial institutions toward greater data-sharing with consumers and, by extension, with the fintech platforms consumers authorize. For MCA underwriting specifically, broader and faster access to verified transaction data is likely to accelerate the trend toward automated, real-time decisioning — and to widen the gap between funders with strong data infrastructure and those still underwriting largely on static bank statement PDFs.

This is showing up in market structure, too. Consolidation among payment and fintech platforms — most notably Nuvei's agreement to acquire Payoneer, announced in mid-2026 — reflects a broader push toward embedding financing decisions directly into merchants' existing payment infrastructure. Embedded finance players such as Shopify Capital, along with financing tools built into PayPal, Stripe, and Block's Square platform, already underwrite advances using real-time transaction data pulled directly from the platforms merchants already use to process sales — collapsing the gap between "when a business needs capital" and "when a decision gets made."

Where the Growth Is Actually Coming From

The scale of this shift is easier to appreciate against the size of the underlying market. Global MCA market size is projected at roughly $32 billion in 2026, with forecasts suggesting it could approach $59 billion by 2033 — a compound annual growth rate near 9.5%. North America is expected to retain around 40% of global market share in 2026, reflecting the concentration of fintech lenders and payment-processing infrastructure in the region, while the Asia-Pacific region is forecast to grow fastest, at a compound annual rate near 11.6%, as embedded finance infrastructure expands into new markets.

Within that growth, the "MCA split" repayment structure — where providers collect a fixed proportion of a merchant's daily card sales rather than a flat daily debit — is expected to represent roughly 36% of the global market in 2026. That structure is inherently better suited to AI-driven underwriting, since it ties repayment directly to observed, ongoing revenue performance rather than a fixed schedule, giving automated models a continuous stream of performance data to monitor and act on.

The Risk Side of Faster Decisioning

Faster, more data-rich underwriting isn't a one-directional improvement, though. The same speed that lets a well-underwritten applicant get funded in seconds also lets a poorly-monitored one take on a second, third, or further advance just as quickly — a pattern known in the industry as "stacking." Industry practitioners have described cases of businesses carrying eight, ten, or even sixteen concurrent advances, with combined MCA exposure reaching into the millions of dollars for what were, on paper, mid-sized companies. AI underwriting models are only as good as the ongoing monitoring built around them; a strong initial decision doesn't guarantee visibility into a merchant's total exposure across other funders down the line.

That's pushed sophisticated funders toward two practices that pure automation doesn't solve on its own: recurring UCC-1 lien searches to detect other funders' claims on the same merchant, and closer attention to sudden deposit pattern shifts that can signal an undisclosed advance is already being serviced elsewhere.

The Takeaway for Investors

AI underwriting is quickly becoming table stakes rather than a differentiator on its own — the funders who will actually separate themselves are the ones pairing strong underwriting models with real portfolio-level monitoring, not just faster approvals at the front end. For capital allocators evaluating exposure to this space, the underwriting technology story is only half the picture; the other half is whether a funder has built the infrastructure to see a merchant's total risk, not just the risk of the deal directly in front of them.

At TMG Investment, we continue to monitor how technology and data infrastructure are reshaping underwriting across the alternative lending landscape.