Five scenarios for banks, credit unions, insurers, and wealth managers; the probabilities behind each, and the indicators that will tell you which future is arriving
Bottom-line first: financial services is not going to be disintermediated out of existence. We are asked that question in various forms: will AI-native fintechs kill the banks, will crypto rails make charters obsolete, will robo-advice finish off the advisor. Those are the wrong fears, and institutions that spend the next decade debating them will miss the discontinuity that actually matters. The biggest change coming to financial services is not AI inside your institution. It is AI representing your customer.
WE PREDICT that financial services in 2040 will remain a large, profitable, and heavily regulated industry, and that its profit pools will have moved. Every business model built on customer inattention will have eroded: deposit stickiness, renewal persistency, assets that stay put out of inertia, relationship pricing that survives because nobody re-shops it. As customers deploy AI agents that sweep, shop, refinance, and negotiate continuously, margin migrates to the things agents cannot replicate (the regulated balance sheet, custody and trust, and judgment under genuine ambiguity), and it concentrates among far fewer institutions than are profitable today. The claims on those positions are being staked in the next thirty-six months, not in 2039.
That conclusion comes from looking at what financial services actually sells, function by function, and asking which functions AI absorbs and which it structurally cannot. Financial services does not get the luxury we described in our CRE brokerage predictions of watching industries eighteen to thirty-six months ahead on the adoption curve; in that article we named wealth management as one of the industries already ahead. Financial services is writing the playbook others will study. That is an advantage for institutions that move, and a shortened fuse for those that wait.
What Financial Services Actually Sells
Strip the industry down, and it’s clear financial services sells six things:
- information asymmetry (knowing the products, rates, structures, and risks the customer doesn’t)
- transaction processing and money movement,
- risk pricing and underwriting,
- capital intermediation on a regulated balance sheet,
- custody and safety, and
- fiduciary blame absorption for boards, CFOs, and families who need a named institution standing behind a consequential decision.
AI absorbs the first three almost completely by 2040. Advice-as-information, (e.g. which products exist, what rates are available, how a structure works), is already free to anyone with a frontier model, and it quietly ends the era in which knowing more than the customer was a defensible service. Transaction processing is a pure automation target. Underwriting is the interesting case: AI models will price risk better than any human committee, but when every institution’s models converge on the same signal, pricing stops being a differentiator and becomes table stakes, with fair-lending and explainability regulation determining how fast anyone is allowed to get there.
The last three resist automation for structural rather than technical reasons. A charter does not automate; it is a legal position, not an informational one, and its value may actually rise as everything around it commoditizes. Custody is a trust business, and trust attaches to accountable institutions, not to software. And the CFO signing off on a nine-figure credit facility, the family office restructuring across three generations, the board approving a benefits captive; they need an institution whose reputation is collateral. That need does not disappear when the institution stops holding an informational edge; it becomes the entire job. The 2040 financial institution is a trust-and-judgment layer, not an information layer, and the economics of a judgment layer favor fewer, more senior institutions.
The Inattention Subsidy
Here is the uncomfortable arithmetic underneath much of the industry’s profitability: a meaningful share of it is monetized customer inattention. Deposits earn near zero while short-term Treasuries yield multiples of that, and the spread persists because moving money is tedious. Insurance carriers reprice renewals upward on the actuarial bet that most customers won’t shop. Assets sit in higher-fee products for years after better alternatives exist. Loan relationships go unrefinanced long past the point the math justifies it. None of this is an accident; it is a profit pool, and in some segments it is the profit pool.
AI agents are inattention-killers. An agent that sweeps cash to best yield nightly, re-shops every insurance renewal in seconds, monitors refinance spreads continuously, and negotiates fee schedules without embarrassment or fatigue does not have a twenty-year relationship with your institution. It has an objective function. And when the institution responds by deploying its own agents (pricing, retention, negotiation), the two sides converge on competitive-frontier pricing, and the subsidy is gone. This is the agentic version of what transparent marketplaces did to trading commissions, compressed into every product line at once.
The strategic question is not whether this happens; the technology substantially exists today, and the regulatory frameworks for agent-initiated transactions are the only real gate. The question is how far up the complexity stack agent-driven commoditization climbs before it hits the trust ceiling, the point where customers still insist on an accountable human institution exercising judgment. Everything below that line reprices. Everything above it becomes more valuable, because it is scarcer.
Five Futures, with Probabilities
Scenario 1: The Consolidated Trust Layer (~32%)
The base case, and the winnable one. Financial services survives as an industry, but the number of institutions falls dramatically in an acceleration of the consolidation already underway among community banks and credit unions. Surviving institutions operate as trust-and-judgment layers on top of agentic rails: products are commoditized and largely agent-shopped, while advice, complexity, and accountable judgment retain pricing power. Margins per institution improve even as industry-wide inattention revenue disappears, because the economics concentrate.
The dividing line in this scenario is not asset size, capital, or brand. It is whether an institution’s customer intelligence lives in institutional systems or in individual bankers’, agents’, and advisors’ heads; and whether its infrastructure is ready to transact with customer-side agents rather than pretending they won’t arrive. This scenario requires no coordination problem to be solved and no platform monopoly to emerge; it is simply every institution optimizing independently, which is why it carries the highest probability.
Scenario 2: Platform & Embedded Capture (~22%)
The customer relationship migrates to consumer platforms, and chartered institutions become regulated utilities behind someone else’s brand. Embedded finance matures from a distribution channel into the primary channel: the platform’s agent is the customer’s interface for banking, insurance, and investing, and the chartered institution earns wholesale margins for balance-sheet capacity and regulatory absorption. Big Tech never wanted your charter; it wants your charter working for it. Institutions persist and can be profitable in this future, but they hold the position power companies hold relative to the devices in your home – essential, invisible, and priced accordingly.
Scenario 3: Agent-Commoditized Spread Collapse (~18%)
Scenario 1’s mechanism at a speed that outruns repositioning. Regulators permit agent-initiated money movement broadly and early, adoption is faster than deposit modeling assumed, and the inattention subsidy unwinds within a few years rather than a decade. Deposit betas spike as historically insensitive balances begin behaving rate-sensitive; insurance persistency collapses toward re-shopped equilibrium; spread businesses compress brutally and fee businesses reprice. Institutions survive, but the industry profit pool shrinks before the trust-layer economics arrive to replace it, and the shakeout is disorderly. The tell for this scenario is velocity: the same indicators as Scenario 1, arriving years early.
Scenario 4: Tokenized Disintermediation (~8%)
Stablecoin and tokenized-asset rails route around bank intermediation: B2B payments settle on-chain by default, tokenized cash equivalents siphon deposits, and some credit forms directly between capital and borrowers on programmable rails. Technically plausible well before 2040 for payments where it is already materially underway, but full disintermediation requires customers to abandon deposit insurance, regulators to permit charter bypass at scale, and institutional treasurers to accept novel counterparty risk. Those constraints bind. Meaningful probability in payments and wholesale settlement, narrow elsewhere; the realistic version of this scenario is banks losing the payments layer while keeping the balance sheet.
Scenario 5: Regulatory Muddle-Through (~20%)
Efficiency gains everywhere, structural change nowhere, and this scenario deserves more respect in financial services than in any other industry, because the incumbents’ moat is written into law and regulators move deliberately. Agent-initiated transactions raise genuine questions of liability, KYC/AML accountability, and fair treatment that could take a decade to resolve, and every year of delay is a year the inattention subsidy survives. Financial services in 2040 looks like today with better tooling and a thinner middle tier. But muddle-through is a probability, not a plan; and even in this future, wealth advice and insurance distribution restructure anyway, because those shifts do not require regulator permission at the same depth.
The Margin Is Moving, Not Vanishing
Read the scenarios together and one pattern dominates. In four of the five futures, financial services remains a profitable industry; in every one of those four, the profits move away from monetized inattention and toward trust, judgment, complexity, and the regulated balance sheet; they also accrue to a much smaller set of institutions than are profitable today. There is a path to still being standing in 2040, and for the institutions on it, a path to better economics than they have ever had. But the amount of claimable ground above the trust ceiling is far smaller than the current population of successful institutions, and the claims are being staked now.
The institutions that survive above that line share a common architecture: customer intelligence that belongs to the institution rather than to individual producers, infrastructure that is ready to serve, and transact with, customer-side agents rather than resist them, a deliberately senior advice-and-judgment layer, and the data discipline across core systems that most mid-market institutions cannot honestly claim today. None of that is built in a budget cycle, which is exactly why the window matters.
What to Watch
Scenario probabilities are only useful if you can tell which future is arriving. Five leading indicators are worth tracking:
- Regulatory frameworks for agent-initiated transactions. The first clear rules permitting autonomous agents to move money, bind coverage, or execute trades at scale is the starting gun for every agentic scenario; watch the liability allocation, because whoever absorbs agent error keeps the customer.
- Deposit beta behavior in historically insensitive balances. When non-rate-sensitive deposits begin behaving rate-sensitive without a rate shock to explain it, agentic sweeping has arrived, whether or not anyone announces it.
- Insurance persistency and re-shop rates. The share of renewals shopped by software rather than by humans is the cleanest single measure of the inattention subsidy unwinding.
- The first agent-to-agent negotiated financial product. One publicized commercial loan or insurance program negotiated substantially AI-to-AI moves the agentic scenarios from theoretical to priced-in.
- A major consumer platform bundling full-stack finance behind an agent. When a platform’s assistant becomes the customer’s default interface for banking, insurance, and investing in one motion, the platform-capture scenario is underway and the fight is over who owns the customer’s agent.
The Strategic Question
The question for a bank, credit union, insurer, or wealth firm leadership team in 2026 is not whether AI will change the industry; that debate is finished everywhere except the industry’s own conferences. The question is which scenario your institution is positioned for, and more pointedly, which side of the customer’s agent you intend to be on. Serving agents, resisting agents, and becoming the trusted institution behind the agent are three different strategies with three different architectures, and efficiency investments that merely make the present cheaper build toward none of them.
Innovation Vista works with financial institutions to answer exactly that question: separating the investments that position you for the consolidated future from the ones that only optimize a disappearing profit pool, and turning that distinction into a positioning strategy for the repricing ahead. If you want to pressure-test which side of 2040 your institution is building toward, that conversation is where we start.
