Plaid's Zachary Perret on fintech's spring revival, stablecoins, and the company's fraud and credit scoring push

Summary

Plaid CEO Zachary Perret describes a company that has quietly outgrown its original identity. Founded to build the API layer connecting bank accounts to fintech apps — the authentication pop-up most users encounter when linking a Wells Fargo account to Venmo — Plaid now runs financial plumbing for major banks like Citi and large enterprises like Invitation Homes, and is pushing hard into two new product lines it expects to become significant businesses.

Fintech spring

Perret's read on the fintech cycle is worth taking seriously: Plaid sits on aggregate transaction data across essentially the entire US digital finance ecosystem, giving it a macro view few others have. His diagnosis of the 2022–2023 downturn is straightforward — demand pulled forward into 2020 as consumers went digital out of necessity, zero-rate conditions turbocharged growth through 2021 and into early 2022, and then rate rises hit a sector heavily exposed to lending and crypto. Companies retrenched, focused on operating profitability rather than growth, and retooled their product mixes. By 2024 the rebound was underway. Perret calls 2025 a "fintech spring" — early-stage formation is up, mid-stage companies are back to growth, and he expects Klarna and Chime to IPO soon.

Fraud detection

Online financial fraud is growing at 20–25% annually off an already large base, accelerated by deepfakes and AI tooling. Perret describes the dynamic bluntly: AI is driving fraud up, and AI is what Plaid is using to fight it back down. The product works like a network intelligence layer — if Plaid sees a fraud pattern on Robinhood, it can flag the same actor to Venmo or a bank before they strike again. He frames the model as analogous to Palantir's approach for financial services: anomaly detection that propagates alerts across the network in near real-time.

Alternative credit scoring

The second new line targets a structural flaw in traditional credit models. If someone's income triples tomorrow after a new job, that improvement takes years to surface in a FICO score because the system relies on loan origination and repayment history rather than real-time cash flow data. Plaid has built its own credit score using live transaction data, so lenders can underwrite on current financial reality rather than a lagged proxy. The pitch to lenders is more accurate risk assessment; the pitch to consumers is access to credit they're already eligible for but can't get.

Stablecoins and payments infrastructure

Perret is unambiguous that stablecoins will grow significantly, particularly in B2B and cross-border payments where ACH's three-day settlement and fragmented real-time alternatives leave obvious gaps. He's skeptical of near-term consumer adoption at the point of sale — nobody is buying coffee with a stablecoin — but sees the B2B case as near-certain. On domestic instant payments, he argues the Federal Reserve's FedNow system already works technically; the problem is adoption. A few hundred mostly small banks use it. His view is that a mandate requiring all banks to support it, combined with better incentive structures, could make instant bank-to-bank transfers universal quickly — but the political will to impose that on small banks is the open question.

AI in banking

Perret doesn't expect AI to visibly transform consumer-facing banking products in the near term. Regulatory overhead and trust requirements will keep banks cautious about deploying AI directly to customers. The more likely near-term effect is AI-powered dashboards behind the scenes — wealth advisors and tellers getting real-time portfolio analysis and recommendations — rather than autonomous AI agents managing consumer accounts. Perret personally wants AI bots optimizing his cash across treasuries, loans, and bill payments, but acknowledges he's well outside the mainstream on that.