S03.1 · Financial Services & Capital Markets

Banking & Depository Institutions

Deposit-taking banks and their $7T+ global revenue pool, as agentic AI automates KYC, underwriting and back-office operations.

S03.1

What is on this page. Market structure, and how AI is reshaping this segment. Ownership, buyer universes, transaction comparables and deal-timing analysis are maintained privately by El Dorado Capital and are not published.

Deposit-taking institutions are the core of global banking: roughly $7.0-7.3T in 2024 revenue (McKinsey Global Banking Annual Review 2025) on total system assets exceeding $180T. The development that matters most right now is not the rate cycle. It is agentic AI moving from pilot to production inside KYC/AML, underwriting and back-office reconciliation — and the operations headcount cost it collapses is falling faster than the branch-distribution moat that has historically protected deposit franchises.

Market structure

Revenue grew ~9% in 2023-2024, almost entirely a rate story, and McKinsey projects deceleration to low-to-mid single digits as net interest margins normalize and industry return on equity settles near 11-12%. The US and China are the largest national revenue pools, but they run on incompatible regulatory architectures — Basel III/IV capital rules and dual state/federal chartering in the US, single-market passporting across the EU, state-linked banking systems across much of Asia. Cross-border scale is therefore harder than the revenue map suggests; a charter or passport that works in one jurisdiction rarely transfers cleanly to another.

Concentration splits in two. The top four to five US banks hold outsized deposit share; beneath them, roughly 4,000 US community and regional banks — plus thousands more internationally — compete on relationship density, not scale. Upstream sit wholesale funding markets and core banking technology vendors; downstream, payments networks, wealth and asset management, and insurance distribution. The economics are net-interest-margin economics: high capital intensity through regulatory capital ratios, recurring interest income, and a profile that has little in common with fee- or subscription-based financial businesses. Charter and licensing gating is the primary barrier to entry, and it is the real one.

How AI is reshaping this segment

Where does the adjacent money sit? In payments, wealth management, specialty consumer lending and capital markets — businesses banks either compete with directly or increasingly want to own outright, because NIM-only exposure has become a less attractive place to hold capital. Diversifying into fee income defends against margin compression. Cross-selling wealth, lending and payments products into an existing deposit base is where the new dollars come from, and it is where AI is doing the most immediate commercial work — agentic systems can now originate and service adjacent products against a bank's own customer data without proportional headcount growth. That is a reallocation of the customer's wallet toward the bank rather than net-new spend, and it accrues to whoever holds the deposit relationship.

Operationally, the shift is furthest along in KYC/AML screening, credit underwriting workflows and back-office reconciliation, where agentic AI performs tasks that previously required entire compliance and operations departments. That compresses the branch and operations cost line — historically one of the largest fixed-cost blocks on a bank's income statement — and it chips at the branch-distribution moat that protected incumbent deposit franchises for a century. A loan underwritten in minutes rather than days needs no branch visit.

Two competitive categories are forming in response. AI-native neobanks carry no legacy branch infrastructure and no core-system debt, so their unit economics on account acquisition and servicing look structurally different from an incumbent's. And core-banking infrastructure providers — the technology layer underneath deposit-taking, lending and payments — are becoming strategically important on their own terms, because whoever controls the AI-native core system sets the pace at which incumbents can automate.

The timeline is uneven. Operations automation is already underway and compounding quarter over quarter. Core-system disruption is the harder, more structural shift — likely a 5-10 year process, gated less by the technology than by the regulatory drag of migrating a chartered deposit institution's core infrastructure without interrupting service or breaching capital and reporting requirements.