Bank of America Corporation (BAC)

Deep Dive Research Report

Report date: 10 September 2026 Listing: NYSE: BAC | Sector: Financial Services | Fiscal year-end: 31 December

Most recent reported period: Q2 2026, released 14 July 2026. BAC reports quarterly with a December year-end and releases in the second week of the month following quarter-end. Q3 2026 results are therefore due around 13-15 October 2026, after this report date. The six concalls used throughout are Q1 2025 (15 Apr 2025), Q2 2025 (16 Jul 2025), Q3 2025 (15 Oct 2025), Q4 2025 (14 Jan 2026), Q1 2026 (15 Apr 2026) and Q2 2026 (14 Jul 2026).

1. What the company does

Bank of America takes in money from people and businesses that want somewhere safe and convenient to keep it, and lends or invests that money at a higher rate than it pays out. That spread is roughly half the company's revenue. The other half comes from charging fees for things those same clients need anyway: managing their investments, executing their trades, underwriting their bonds, advising on their acquisitions, moving their payroll around the world.

The unglamorous truth of the business is that Bank of America's single most valuable asset is about $2 trillion of deposits, a large share of which sits in ordinary checking accounts paying close to nothing. In Q2 2026 the average rate paid on all deposits was low enough that the company's net interest margin sat at 2.08% while it earned a 17.0% return on tangible common equity (Q2 2026 press release). You cannot buy a deposit base like that. You accumulate it over decades, through branch networks, payroll relationships, mortgage escrow, small-business operating accounts and the plain inertia of people not wanting to move where their salary lands.

Everything else the company does is built to be attached to that anchor. As of Q2 2026 it served nearly 70 million consumer and small-business clients through roughly 3,500 retail financial centres, about 15,000 ATMs and roughly 60 million verified digital users, and it operates in the United States, its territories and more than 35 other countries and jurisdictions (company overview).

The founding story, and why the corporate history still matters

Bank of America is two companies stitched into one, and the seam is still visible in how it operates.

The name comes from San Francisco. Amadeo Pietro Giannini founded the Bank of Italy on 17 October 1904 to serve Italian immigrants who were being turned away elsewhere; it was renamed Bank of America National Trust and Savings Association on 3 November 1930 following its 1928 merger with Bank of America, Los Angeles (Bank of America corporate history).

The legal entity, however, is from Charlotte. Commercial National Bank was founded there in 1874, merged with American Trust Company in 1958 to form American Commercial Bank, became North Carolina National Bank in 1960, and merged with C&S/Sovran Corporation in 1991 to form NationsBank. In 1998 NationsBank of Charlotte acquired BankAmerica in what was then the largest bank acquisition in history, took the Bank of America name, and installed NationsBank's president, chairman and CEO Hugh McColl in the same roles at the combined company. That is why the SEC filer identity, the corporate headquarters and the institutional culture are Charlotte's while the brand is California's.

Two crisis-era acquisitions then defined the company that exists today. Countrywide Financial was announced on 11 January 2008 and completed in July 2008 for $4.1 billion, which brought a mortgage business and, with it, more than a decade of litigation and remediation cost. Merrill Lynch was announced on 14 September 2008, in the same weekend Lehman Brothers failed, and closed on 1 January 2009 for approximately $50 billion. Merrill is the reason Bank of America has a genuine investment bank and a genuine wealth management franchise at all. Without it, this would be a very large commercial bank with a small markets business.

Brian Moynihan became president and CEO effective 1 January 2010, following Ken Lewis's retirement. He inherited a company that was still absorbing two distressed acquisitions and facing a decade of legal settlements. The entire strategy of the sixteen years since is a reaction to that inheritance, and it is summarised in a phrase the company repeats relentlessly: Responsible Growth.

"By focusing on the principles that drive Responsible Growth, we've balanced risk and reward and managed costs appropriately to deliver strong earnings, capital returns and support for all who need us."

• Brian Moynihan, 2025 letter to shareholders

The most revealing statistic in that letter is not a financial one. Moynihan notes that Bank of America had approximately 210,000 employees at the end of 2007; in 2008 it acquired two businesses that brought in approximately 76,000 more; and today it has just over 213,000. In other words, the company absorbed a workforce equivalent to a third of its size and, eighteen years later, is running with essentially the same headcount it started with. That is the whole operating story: a fixed cost base held roughly flat while revenue compounds.

The core value proposition

For a retail client, the proposition is convenience plus rewards plus integration. You keep your operating account here, and in exchange your credit card earns more, your mortgage is cheaper, your brokerage account is in the same app, and the AI assistant in that app can answer questions about all of it.

For a corporate client, the proposition is that one institution can hold your operating cash across 35+ countries, extend you a revolving credit facility, underwrite your bond issue, advise you on an acquisition, hedge your currency exposure and execute your equity trades. The lending relationship is the door; the fee businesses are the rooms behind it.

For a wealthy client, the proposition is a Merrill advisor with the balance sheet of a systemically important bank behind them, so a securities-backed loan against your portfolio is a phone call rather than a new relationship.

The technical nature of the product, and what is hard to replicate

Banking looks like a commodity from the outside. Three things make Bank of America's version hard to copy.

The deposit franchise is a physical and behavioural asset. In Q2 2026 the company recorded its 30th consecutive quarter of net new checking account growth, adding over 160,000 accounts in the quarter, on a base of about 38.5 million consumer checking accounts. Roughly 92% of those are the client's primary account. Non-interest-bearing deposits grew for a seventh consecutive quarter, up 4% year over year (Q2 2026 concall, 14 July 2026). A competitor cannot buy this. A neobank can buy the balance with a promotional rate, but it buys hot money that leaves when the rate does; it does not buy the direct deposit of a salary.

The regulatory perimeter is a moat and a tax simultaneously. Bank of America is a Category I global systemically important bank. It carries a GSIB capital surcharge, submits to annual stress testing, and must hold a CET1 ratio that stood at 11.2% at the end of Q2 2026 against a 10% requirement. That is expensive. It is also the reason there are four institutions in the United States that can do everything Bank of America does, not forty.

Technology has become the operating leverage engine. The company has invested more than $100 billion in technology over the past decade and spends roughly $13 billion a year, of which about $4 billion is new-initiative spend rather than run-the-bank (2025 shareholder letter). Erica, the AI assistant launched in 2018, has surpassed 3 billion client interactions and averages more than 58 million interactions per month. By Q2 2026 the company had approved 300 AI use cases, had 114 live generative AI implementations with 34 fully deployed, and reported over 400,000 employee prompts a day (Q2 2026 concall, 14 July 2026).

What this actually looks like for one client

Take a mid-sized industrial distributor in Ohio with $400 million of revenue.

It opens an operating account with the commercial bank. Payroll, receivables and payables run through CashPro, the treasury platform used by more than 40,000 corporate and commercial clients globally, which handles payments, receivables, liquidity, FX and trade in one place and now includes machine-learning cash forecasting (CashPro). Those balances sit as deposits and fund the bank at close to zero cost.

The company draws a revolving credit facility for working capital. That is a loan, earning a spread. When it acquires a competitor, BofA Securities advises, earning an M&A fee, then underwrites the bond that finances it, earning an underwriting fee. Global Markets hedges the interest-rate exposure on that bond and the euro exposure on the acquired European subsidiary, earning a trading spread.

The 400 employees are enrolled in Employee Banking and Investing, which added 1.8 million eligible employees across 272 companies in 2025. Some open checking accounts, which become consumer deposits. The founder, now liquid after a partial sale, becomes a Merrill client, generating an asset management fee, and borrows against the portfolio rather than selling, generating another loan.

One relationship touches all four reporting segments. This is what management means when it says "integrated". It is also why the deposit franchise is the thing to watch: it is the root of the tree, not one of the branches.

2. Business segments

Bank of America reports four segments plus an "All Other" category. On FY2025 revenue net of interest expense, the four segments were: Consumer Banking $43,673 million, GWIM $24,883 million, Global Banking $24,108 million and Global Markets $24,096 million (FY2025 10-K). Total company revenue for 2025 was $113.1 billion with net income of $30.5 billion and diluted EPS of $3.81 (Q4 2025 earnings release).

The revenue split is deceptively even. The net income split is not: Consumer Banking earned $12.2 billion, Global Banking $7.8 billion, Global Markets $6.11 billion and GWIM $4.67 billion in FY2025.

2.1 Consumer Banking - the funding engine

What it does. Consumer Banking serves individuals and small businesses across the United States with checking and savings accounts, credit cards, residential mortgages, home equity, auto loans, small-business loans and a self-directed investing platform. It runs the physical network of roughly 3,500-3,600 financial centres and about 15,000 ATMs, and it owns the digital front door: approximately 49 million active digital banking users including roughly 41 million active mobile users as of 31 December 2025, with Zelle at 25 million active users in Q1 2026.

At scale: 38.5 million consumer checking accounts, average balances above $9,000, and 30 consecutive quarters of net new checking account growth as of Q2 2026. Average deposits in the segment were $951 billion in Q1 2026 against average loans of $322 billion. The segment is a massive net funder of the rest of the company.

The core capability. Two things, and neither is glamorous. First, primary-account acquisition at scale with low attrition: getting a customer to route their salary here and keep it here. Second, cost per account. The segment carries an efficiency ratio below 50%, which is the lowest of the four and the reason Consumer Banking is the largest single profit contributor despite lending only about a quarter of its deposits. That number is a direct product of the flat-headcount, high-automation approach: 79% digital adoption in Q1 2026, with Erica handling the queries that used to require a call centre.

Why it exists separately. Different regulator, different economics, different unit of production. Consumer banking is regulated for consumer protection (CFPB, fair lending, deposit insurance) rather than for market conduct, and its P&L is driven by account counts and cost per transaction rather than by deal flow. Bundling it with the investment bank would obscure both.

Competitive position. The direct rivals are JPMorgan Chase, Wells Fargo and Citigroup on the national branch side, Capital One on cards, and increasingly Chime, SoFi and the cash-management products of Charles Schwab and Fidelity on the digital side. Bank of America wins on integration and rewards: its bet is that a client with a checking account, a card, a mortgage and a brokerage account is far stickier than one with any single product. It loses to Capital One on pure card rewards richness for customers who do not want a banking relationship, and to digital-only players on price for customers who only want a place to hold cash.

The May 2026 relaunch of the loyalty programme is the sharpest expression of the strategy. BofA Rewards replaced Preferred Rewards on 27 May 2026, collapsing five tiers into four (Member, Preferred Plus, Preferred Honors, Premier) and, critically, removing the minimum balance requirement so that anyone with an eligible checking account can join. Within seven weeks over 3 million clients had enrolled, taking total enrolment to nearly 14 million, running at roughly 17,000 new members a week of whom about half were new to Bank of America (BofA Rewards enrolment release, July 2026). Management's stated ambition is to grow membership from about 11 million to 30 million.

How it fits. This is the cash cow and the funding source. Management talks about it in terms of account growth and operating leverage rather than revenue growth, which tells you what it is for.

Revenue mix: roughly 37% of the four segments' combined revenue in FY2025.

2.2 Global Wealth and Investment Management - the fee annuity

What it does. GWIM is Merrill (the mass-affluent and high-net-worth advisory business inherited from the 2009 acquisition) plus Bank of America Private Bank (ultra-high-net-worth, trust and estate services). It manages money, provides financial planning, and lends against portfolios. In Q2 2026 client balances reached $4.9 trillion, up 12% year over year, with assets under management of $2.3 trillion, up 17%.

The core capability. The advisor force and the brand. Merrill had 2,314 advisors recognised on the 2026 Forbes Best-in-State Wealth Advisor lists, more than any other firm, and 341 on the 2026 Barron's Top 1,500 list. More important than the accolades is what management reported in Q2 2026: advisor attrition at "near historic lows" (Q2 2026 concall, 14 July 2026). In wealth management, the advisor owns the client relationship. Attrition is the single number that determines whether AUM compounds or leaks.

The second capability is the banking attach. A Merrill client can get a mortgage, a securities-backed line of credit and a deposit account from the same institution, and those products both deepen the relationship and generate net interest income that a pure-play RIA cannot earn.

Why it exists separately. Regulatory: the broker-dealer (Merrill Lynch, Pierce, Fenner & Smith) and the registered investment advisory business operate under securities regulation with fiduciary and suitability obligations distinct from banking. Economically: this is a fee-on-assets business with an efficiency ratio around 74%, structurally the highest of the four segments because advisor compensation is a large share of revenue. Blending it into Consumer Banking would obscure both.

Competitive position. Morgan Stanley is the direct peer and has been the more aggressive acquirer (E*TRADE, Eaton Vance). Charles Schwab competes from below on price and self-directed platform. UBS competes at the top end. Edward Jones and the large independent RIA aggregators compete for advisors. Merrill's edge is the bank behind it; its historical weakness has been advisor recruitment and retention, which is exactly why the near-historic-low attrition figure matters.

How it fits. This is the growth bet with the most explicit target attached. On the Q2 2026 call management said Merrill is targeting 5% AUM growth over three to five years and was seven months into that plan, "off to a good start". Net income grew 42% year over year in Q2 2026 to $1.4 billion, the fastest of any segment bar Global Markets.

Revenue mix: roughly 21%.

2.3 Global Banking - the corporate relationship

What it does. Global Banking is commercial and corporate lending, treasury and cash management services, and investment banking (advisory, debt underwriting, equity underwriting) under the BofA Securities brand. It serves everything from middle-market companies through to multinationals and governments.

The numbers show what kind of business it really is. In Q1 2026 average deposits were $648 billion, up 13% year over year. In FY2025 investment banking fees were approximately $5.5 billion across the firm, with a #3 global ranking. In Q2 2026, investment banking fees jumped 50% year over year to $2.1 billion.

The core capability. Two things that look separate and are not. The first is the treasury platform, CashPro, with more than 40,000 corporate and commercial clients. Treasury management is genuinely sticky infrastructure: once a company's payment files, bank connectivity, entitlements and forecasting sit on one platform, moving is a multi-quarter IT project with real operational risk. The second is the balance sheet. Corporates award advisory and underwriting mandates disproportionately to banks that lend to them. The revolver is the loss leader; the fee is the product.

Why it exists separately. Investment banking is a licensed broker-dealer activity with conflict-of-interest walls that must be structurally separated from commercial lending. It is also economically different: lending is a spread annuity, advisory is lumpy fee income tied to the M&A cycle.

Competitive position. JPMorgan is the benchmark and holds the #1 global investment banking fee ranking with about 8.4% wallet share. Goldman Sachs dominates the top end of M&A advisory, having retained the #1 M&A adviser position for a 23rd consecutive year in FY2025. Bank of America sits reliably in the top three by fee volume, which is a real position but a follower's one. In middle-market commercial banking it competes against JPMorgan, Wells Fargo, PNC and U.S. Bancorp, where the balance sheet and CashPro give it a stronger hand than it holds in mega-cap advisory.

How it fits. This is the margin-stable core of the wholesale bank: FY2025 net income of $7.8 billion, roughly 25% of company net income, with a 50% efficiency ratio and a 16% return on allocated capital in Q4 2025.

Revenue mix: roughly 21%.

2.4 Global Markets - the cyclical upside

What it does. Global Markets makes markets in fixed income, currencies, commodities (FICC) and equities for institutional clients, provides financing (prime brokerage, securities lending, repo), and produces research. It is the descendant of Merrill Lynch's institutional business.

FY2025 revenue was $24.096 billion, up 10%, with net income of $6.111 billion, up 9%, on a 13% return on average allocated capital and a 63.99% efficiency ratio. The revenue composition tells you what the business is: $12.064 billion from market making and similar activities, $5.690 billion of net interest income (up 69% year over year, reflecting financing balances), $2.837 billion of investment banking fees and $2.511 billion of investment and brokerage services.

Q2 2026 was exceptional. Sales and trading revenue reached $7.2 billion, up 33%, with equities up 70% to a record $3.6 billion and FICC up 9% to $3.5 billion, marking the 17th consecutive quarter of year-on-year growth. Segment net income rose 72% to $2.6 billion at a 20% return on allocated capital.

The core capability. Balance sheet plus technology plus client franchise, in that order. Market making requires capital to warehouse risk, low-latency infrastructure to price it, and a distribution network wide enough that inventory turns. The equities financing business in particular is a scale game: prime brokerage balances are won by banks that can fund them cheaply, which loops back to the deposit franchise.

Why it exists separately. Risk profile. This is the only segment with genuine market risk, subject to Volcker Rule constraints, trading book capital rules and daily VaR limits. Its earnings are also the most volatile, and management is explicit that the return target is different from the rest of the firm.

Competitive position. Goldman Sachs, JPMorgan, Morgan Stanley and Citigroup domestically; Barclays, UBS and Deutsche Bank internationally. Equities is where Bank of America has gained most visibly, growing 23% in FY2025 and 70% in Q2 2026, well ahead of the market. FICC is a share-holding business rather than a share-gaining one.

How it fits. This is the operating leverage amplifier on the way up and the earnings-quality question on the way down. Management flagged this itself on the Q2 2026 call, noting that second-half comparisons are difficult because prior-year markets performance was elevated. A 72% year-over-year net income increase is not a run rate.

Revenue mix: roughly 21%.

Segment comparison

3. Products and business detail

Consumer product catalogue

Deposits. Advantage Banking checking (three tiers: SafeBalance with no overdraft, Advantage Plus, Advantage Relationship), Advantage Savings, CDs. The tiering exists to serve customers at very different balance levels without cannibalising fee income, and to create an upgrade path into the rewards programme.

Credit cards. The core lineup is Customized Cash Rewards (choose one 3% category from a menu including gas and EV charging, online shopping, cable/internet/phone, streaming, dining, travel, drug stores, home improvement, plus 2% on groceries and wholesale clubs), Unlimited Cash Rewards (flat cash back), Travel Rewards (3 points on travel booked through the Bank of America Travel Center, 1.5 points on everything else, no annual fee), Premium Rewards ($95 annual fee) and Premium Rewards Elite ($550 annual fee, with $300 airline incidental, $100 TSA PreCheck/Global Entry and $150 lifestyle credits). There is a student version of Customized Cash Rewards.

The rewards multipliers scale with BofA Rewards tier, which is the mechanism that ties the card economics to the deposit relationship. This is why the tier structure change in May 2026 matters commercially and not just cosmetically: opening the programme to anyone with a checking account puts a larger population on the path to the higher multipliers.

In June 2026 the company added three features (June 2026 product release):

• Refer-a-Friend, paying up to $100 cash or 10,000 points per approved referral, capped at $500 or 50,000 points a year, initially on Customized Cash, Unlimited Cash and Travel Rewards and expanding across the card range through 2026.
• Custom Pay Plan, converting purchases of $100 or more into fixed-fee instalment plans over 3 to 18 months while preserving rewards and card protections. This is the incumbent response to buy-now-pay-later.
• My Credit, free FICO score access, credit reports and alerts integrated with Erica, replacing a retiring FICO Score programme.

Lending. Residential mortgage and refinancing, home equity lines, auto and specialty vehicle loans, and personal lines. The mortgage business is materially smaller relative to the balance sheet than it was in the Countrywide era, which is a deliberate post-crisis posture rather than an accident.

Small business. Bank of America is the #1 FDIC-ranked small business lender, serving approximately 4 million small-business households. Products span the same deposit, card and lending set plus merchant services and payroll integration.

Investing. Merrill Edge Self-Directed for the do-it-yourself client, Merrill Guided Investing for the robo tier, and full Merrill advisory above that. The ladder is deliberate: a Consumer Banking customer can move up it without leaving the franchise.

The delivery process, and why the physical network still exists

Roughly 95% of client interactions are digital. The company nonetheless keeps building branches, and the reason is the mix of what happens in them. Digital handles transactions; branches handle complex, high-value, trust-dependent conversations - mortgage applications, small-business lending, investment advice, estate questions.

The network strategy is therefore a churn, not an expansion. Since 2014 the company has invested over $5 billion in the financial centre programme, has renovated more than 3,100 centres, and is opening more than 165 new centres across 63 markets by the end of 2026, including entry into nine new markets: Birmingham and Huntsville (Alabama), Boise (Idaho), Louisville (Kentucky), New Orleans (Louisiana), Omaha (Nebraska), Dayton (Ohio), and Madison and Milwaukee (Wisconsin). Alabama, Louisiana, Nebraska and Wisconsin are new states. On completion the network reaches more than 200 markets across 39 states. Alongside this the company runs a consolidation programme aimed at closing roughly two branches for every one it opens, which is why the total centre count has drifted down (approximately 3,700 at Q4 2024, 3,628 at Q4 2025, 3,540 US centres at Q1 2026) even as new markets open.

On the Q2 2026 call Moynihan characterised the approach as "rational, full market build-outs in new and growth markets rather than one-at-a-time locations". That distinction matters: a single branch in a new city has no marketing leverage, whereas a simultaneous multi-branch entry supports advertising, hiring and a small-business calling effort.

The technology stack as a manufacturing process

For a bank, technology is the factory. Some specifics worth holding:

Erica launched in 2018, has surpassed 3 billion client interactions, averages more than 58 million interactions a month, and in Q4 2025 alone was used by 20 million people nearly 200 million times. It is not only a consumer app feature: it is embedded across Merrill, Benefits OnLine and CashPro.

Generative AI deployment has moved from pilot to production across 2025-2026. On the Q4 2025 call management said AI-assisted coding had reduced coding work by about 30%, equivalent to roughly 2,000 positions, with those resources redeployed. By Q1 2026 there were 90 installations with all 200,000-plus employees having access. By Q2 2026 there were 300 approved use cases, 114 live generative AI implementations, 34 fully deployed, and more than 400,000 employee prompts a day.

CashPro carries the corporate side: real-time visibility across payments, receivables, liquidity, investments, FX and trade, with machine-learning forecasting, QR sign-in and CashPro Chat. Adoption of the newer features rose 21% in Q1 2025 over Q4 2024.

The data foundation is the part nobody sees. On the Q3 2025 call Moynihan noted the company spent approximately $3 billion on data infrastructure between 2014 and 2019 specifically to make later AI applications possible. This is the honest answer to why some banks can deploy AI at scale in 2026 and others cannot: the enabling work happened seven years earlier.

Geographies

The United States is overwhelmingly the revenue base. International operations span more than 35 countries and jurisdictions and are concentrated in Global Markets and Global Banking rather than consumer. Q2 2026 international revenue grew 38% year over year with Asia-Pacific called out as a standout on that call. In 2025 the company announced a Global Operations expansion to Belfast, Northern Ireland.

Milestones that changed the business

• 1904: Bank of Italy founded in San Francisco by Amadeo Pietro Giannini.
• 1958: BankAmericard distributed to 60,000 customers in Fresno, California - the product that became Visa.
• 1998: NationsBank acquires BankAmerica; the modern company is formed.
• 2008-2009: Countrywide (July 2008, $4.1bn) and Merrill Lynch (1 January 2009, ~$50bn). Merrill created the wealth and markets businesses that generate roughly 42% of segment revenue today.
• 2018: Erica launches, beginning the AI-led cost curve.
• 5 November 2025: First Investor Day in roughly 14-15 years, resetting medium-term targets upward.
• 27 May 2026: BofA Rewards replaces Preferred Rewards, removing the balance minimum.

4. Customers

Who buys

Consumers. Nearly 70 million client relationships, about 38.5 million consumer checking accounts. The base skews toward primary banking relationships (roughly 92% of checking accounts are the client's primary account) with average balances above $9,000.

Small businesses. Approximately 4 million small-business households, with Bank of America the #1 FDIC-ranked small business lender.

Wealth clients. Merrill and Private Bank clients holding $4.9 trillion of balances as at Q2 2026, of which $2.3 trillion is discretionary or advisory AUM.

Corporates and institutions. More than 40,000 CashPro clients globally, plus the institutional investor base of Global Markets.

Who makes the buying decision, and on what criteria

Retail. The individual decides, and the decision is made once and then rarely revisited. Acquisition criteria are convenience (branch and ATM proximity, app quality), price (fees, rates, rewards) and trust. The single highest-value acquisition event is a direct-deposit switch, which is why the Employee Banking and Investing programme exists: it goes to the employer, not the employee, and added 1.8 million eligible employees across 272 companies in 2025. Sales cycle: minutes to days.

Small business. The owner or bookkeeper decides. Criteria are credit availability first, then cash management and payments. Sales cycle: days to weeks for deposits, weeks to months for credit.

Corporate treasury. The Treasurer and CFO decide, usually through a formal RFP. Criteria are geographic coverage, platform capability, credit commitment, and price. Sales cycle: six to eighteen months for a treasury mandate.

Investment banking mandates. The CEO, CFO and board decide, on a per-transaction basis. Criteria are sector expertise, the individual banker's relationship, league table credibility, and - decisively for debt and leveraged transactions - willingness to commit the balance sheet. Sales cycle: continuous relationship coverage punctuated by a decision at deal time.

Institutional trading. The buy-side trading desk decides, continuously. Criteria are pricing, liquidity provision in size, research quality, and financing terms. Sales cycle: none. This is repeat business awarded trade by trade and measured in wallet share.

Why they choose Bank of America

The reasons differ by segment, and it is worth being specific rather than generic.

Retail clients choose it for the combination of branch density and app quality, which very few competitors match on both dimensions simultaneously - digital-only banks lose on the first, most regional banks lose on the second. The rewards programme is the explicit inducement to consolidate: multipliers on card spend scale with total balances held, so the marginal dollar moved to Bank of America is worth more than the marginal dollar left elsewhere.

Corporates choose it because the credit facility and the treasury platform come together, and because BofA Securities is credible enough to run the financing. Institutions choose Global Markets for equities financing scale and research.

Switching costs

These are the real thing, and they vary enormously.

Retail: moderate and behavioural. There is no contract, no exit fee, no qualification process. What stops people leaving is friction - re-pointing direct deposit, re-establishing autopay on a dozen billers, moving a mortgage escrow, losing rewards tier status. It is weak per customer and formidable in aggregate; it is why Bank of America has posted 30 straight quarters of net checking account growth and why a competitor offering a better rate does not simply take the base.

Corporate treasury: high and technical. Moving off CashPro means rebuilding ERP-to-bank connectivity, re-testing payment file formats, re-establishing entitlements and approval workflows, and re-papering account documentation across every jurisdiction the client operates in. Treasurers do this reluctantly and slowly.

Wealth: high but portable to the advisor, not the firm. The client's loyalty attaches to their Merrill advisor. If the advisor leaves for a competitor, a large share of the book typically follows. This inverts the usual switching-cost logic and explains why advisor attrition is the most important operational metric in GWIM, and why management chose to disclose that it is at "near historic lows" (Q2 2026 concall, 14 July 2026).

Institutional trading: low. Buy-side firms maintain relationships with a dozen dealers and route by price and liquidity. Share is won and lost quarter to quarter.

Concentration

There is essentially none on the customer side, and that is a genuine structural strength. No single client represents a material share of revenue. The concentration risk in this business sits elsewhere: in the US consumer economy as a whole, in the interest-rate path, and in specific credit sectors. On the Q2 2026 call management addressed the sector question directly, noting that commercial lending growth of 11% year over year was broad-based across business banking, the commercial bank and the corporate bank, with middle market growing around 8%, explicitly rebutting the suggestion that loan growth was a hyperscaler and data-centre phenomenon.

Contract structures and revenue predictability

The revenue base ranks roughly as follows, from most to least predictable:

1. Net interest income (about half of revenue). Not contractual in the usual sense, but highly forecastable because it derives from balances that move slowly and a repricing schedule that is already locked. On the Q3 2025 and Q4 2025 calls management quantified this: roughly $10-15 billion of mortgage-backed securities and mortgage loans roll off each quarter and are replaced at yields 150-200 basis points higher. That is a contracted tailwind independent of what the Fed does.
2. Asset management fees (GWIM). Recurring, charged on balances, but exposed to market levels.
3. Card and service charges. Recurring and stable.
4. Treasury and payments fees. Recurring, contract-based, very sticky.
5. Investment banking fees. Transactional and cyclical. Q2 2026's 50% year-over-year jump illustrates both the upside and the volatility.
6. Trading revenue. The least predictable, though the financing component within it is more annuity-like than the market-making component.

5. Competitive landscape

The structure of the industry

US banking is a barbell. At one end sit four universal banks that can do everything: JPMorgan Chase, Bank of America, Citigroup and Wells Fargo. Alongside them sit two firms that do wholesale and wealth but not mass-market retail: Goldman Sachs and Morgan Stanley. Below that are the large regionals (U.S. Bancorp, PNC, Truist, Capital One), then thousands of community banks, then a growing set of non-banks that attack individual product lines without carrying a bank charter.

Bank of America is second by assets and market capitalisation, first or near-first in US consumer deposits, and third in global investment banking fees. It is nowhere first by a wide margin, and that is the accurate summary of its competitive position: a strong number two across an unusually broad set of businesses.

Named competitors

Market capitalisations sourced from companiesmarketcap.com, September 2026. Figures move daily and are included only as a peer-size reference.

Where Bank of America wins and where it loses

Against JPMorgan: it mostly loses, and honestly so. JPMorgan is larger, earns a higher return on tangible equity, and holds the #1 investment banking position. Bank of America's counter is a deposit franchise of comparable quality and, since 2025, a demonstrably faster earnings growth rate: Q2 2026 net income rose 27% and EPS 34%. Whether that is convergence or a good quarter is the central question about this company, and Section 9 addresses it directly.

Against Wells Fargo: it wins on breadth. Wells has a comparable branch footprint and consumer franchise but a much smaller investment bank and markets business, so it has no equivalent of the Q2 2026 lift that took Global Markets net income up 72% and IB fees up 50%.

Against Citigroup: it wins in US retail decisively and loses in cross-border corporate treasury, where Citi's network is deeper.

Against Goldman Sachs and Morgan Stanley: it wins on funding cost. Bank of America finances its trading and financing books with $2 trillion of cheap deposits; Goldman and Morgan Stanley fund a larger share in wholesale markets. It loses on top-tier advisory prestige, where Goldman's position is entrenched.

Against Morgan Stanley specifically in wealth: it is roughly even on client balances but Morgan Stanley has been the more aggressive acquirer. Merrill's advantage is the banking attach through a much larger consumer bank.

Against digital-only banks: it wins on breadth and loses on cost to serve. A neobank with no branches has a structurally lower cost per account. Bank of America's answer is that it wants the primary relationship, not the balance, and a primary relationship needs a place to go when something goes wrong.

Against Capital One in cards: it usually loses on headline rewards richness for the rewards-maximising customer, and wins for the customer who wants their card, checking account and brokerage in one place. The BofA Rewards restructure is a direct attempt to move the boundary.

Barriers to entry

Genuinely high, and worth breaking down by what they actually prevent.

A bank charter takes years and substantial capital. This blocks new full-service entrants almost completely, which is why the disruption of the past decade has come from non-banks partnering with small chartered institutions rather than from new banks.

The GSIB capital and regulatory regime blocks scaling: a firm that grew to Bank of America's size would inherit a GSIB surcharge, resolution planning, annual stress testing and a living will. The cost is a moat for incumbents.

The deposit franchise cannot be bought at any reasonable price. This is the highest barrier of all and is not written down anywhere. Primary banking relationships accumulate over decades.

Technology at scale. The $100 billion of decade technology spend and the $3 billion of 2014-2019 data infrastructure work is not something a challenger replicates in a funding round.

Trust and brand in wealth. People hand over life savings to institutions they have heard of.

What the barriers do not prevent is unbundling. A fintech does not need a charter to offer a better rewards card, a better instalment-payment product or a better brokerage app. It just needs a partner bank. This is why the June 2026 product releases (Custom Pay Plan, My Credit, Refer-a-Friend) matter: they are defensive product responses to specific unbundling attacks.

Structural shifts

AI is changing the cost curve, and the incumbents are winning that race. Bank of America has held headcount roughly flat since 2007 while absorbing 76,000 acquired employees, and is now guiding headcount down. Moynihan's phrasing on the Q4 2025 call was blunt: "We can just make decisions not to hire and let the headcount drift down." Peer banks are pursuing the same path. This favours the largest players because AI deployment costs are largely fixed and the benefit scales with employee count.

Regulatory direction has reversed. On 19 March 2026 US regulators released revised proposals which, in aggregate, "would modestly reduce capital requirements for large banks and moderately reduce requirements for smaller banks", alongside a revision to how the GSIB surcharge is calculated (Basel III Endgame re-proposal coverage). The original July 2023 proposal was projected to increase aggregate large-bank capital requirements by 16-19%. This is a material swing in the operating environment for every bank in the table above.

Consolidation continues at the regional level but not at the top: the four universal banks are effectively barred from acquiring each other by deposit-cap rules.

Where Bank of America is strong and where it is exposed

Strong: the deposit franchise, the breadth of the model, the cost trajectory, and the equities franchise which has taken visible share (up 23% in FY2025, up 70% in Q2 2026).

Exposed: it is not first at anything except consumer deposits. Its Q2 2026 result leaned heavily on two of the most cyclical lines in banking (IB fees +50%, Global Markets net income +72%). Its efficiency ratio of 59% sits at the very top of the 55-59% medium-term band it set at Investor Day, which means the operating leverage story has to keep working for the returns story to hold.

6. Industry

What drives demand

Bank of America's revenue is driven by four things, in descending order of importance.

The level and shape of interest rates. About half of revenue is net interest income, which is the spread between what the bank earns on assets and pays on deposits. Higher short rates generally help because deposit costs reprice more slowly than asset yields. The current environment is unusual: the Fed has held the federal funds rate at 3.50%-3.75% since December 2025, held again on 29 July 2026 in a 9-3 vote with three policymakers dissenting in favour of a 25 basis point increase, and a September 2026 hike has become a live expectation, driven by supply-chain shocks tied to the Iran conflict keeping energy costs elevated (Fed outlook coverage). Bank of America's Q2 2026 guidance explicitly assumes one 25 basis point hike in September, and management quantified the sensitivity: a 100 basis point rate increase would add approximately $1 billion to NII over 12 months (Q2 2026 concall, 14 July 2026).

Loan demand, which tracks nominal GDP, business confidence and capital expenditure. Bank of America's average loans grew 8% year over year in Q2 2026 to $1.22 trillion, ahead of nominal GDP, with commercial up 11%.

Capital markets activity, which drives investment banking fees and trading. This is the most cyclical driver and depends on M&A confidence, IPO windows and market volatility.

Asset values, which drive wealth management fees. GWIM AUM rose 17% in Q2 2026 to $2.3 trillion, a figure that is part net flows and part market appreciation.

Industry size and structure

The 50 largest US banks held a combined $25.58 trillion in assets as at 31 December 2025, and reported a $186.20 billion increase in the most recent quarter with 38 of the 50 posting growth (S&P Global Market Intelligence, March 2026). JPMorgan Chase is the largest by assets at $4.143 trillion as at 30 June 2026. Bank of America is second.

Deloitte's 2026 banking and capital markets outlook frames the sector's NII growth for 2026 as modest at the industry level, with deposit costs continuing to fall - the average cost of interest-bearing deposits had already declined to 2.5% over the first six months of 2025. Bank of America's guided 6-8% NII growth therefore sits meaningfully above the industry baseline, which is a claim about share and repricing rather than about the rate environment.

On the consumer side, credit card balances are expected to reach $1.18 trillion by end-2026, up from $1.16 trillion in 2025, with delinquency rates forecast to be broadly flat; the 30-day delinquency rate dipped to 2.92% in Q1 2026 (TransUnion 2026 outlook). The caveat worth holding is that the deep-delinquency picture is less benign: the share of credit card balances 90 or more days delinquent rose from 7.6% in Q3 2022 to 12.8% in Q1 2026.

Where Bank of America sits in the value chain

It is the intermediary itself, not a supplier to one. It sits between savers and borrowers, between issuers and investors, and between corporates and the payment system. The inputs it buys are money (deposits and wholesale funding), technology, and people. The output is credit, liquidity, risk transfer and advice.

The relevant "import dynamic" for a US bank is not goods but capital and competition: the degree to which foreign banks (HSBC, Barclays, UBS, Deutsche Bank, MUFG) and non-banks (private credit funds, fintechs, asset managers) take share in lending and intermediation. Private credit in particular has taken meaningful share of leveraged and middle-market lending from bank balance sheets over the past decade. This is the most important structural competitive shift in the industry and it is one that regulation drove: capital rules made certain loans expensive for banks to hold, and the loans moved to entities without those rules.

Regulation

The regulatory environment is the single largest exogenous variable for this company, and it is currently moving in the industry's favour.

• Capital. Bank of America must maintain CET1 above a requirement built from a 4.5% minimum, a stress capital buffer set annually by the Federal Reserve's stress test, and a GSIB surcharge. Its requirement is 10%; it held 11.2% at Q2 2026. The 19 March 2026 re-proposals would modestly reduce large-bank requirements and revise the GSIB surcharge calculation, with comments due 18 June 2026.
• Liquidity. Liquidity coverage ratio and net stable funding ratio requirements, plus internal liquidity stress testing.
• Resolution. Annual living wills and total loss-absorbing capacity requirements.
• Consumer. CFPB supervision, fair lending, and product-level rules. Proposed credit card rate caps have been a live policy topic; on the Q4 2025 call Moynihan pushed back directly: "If you bring the caps down, you're going to get strict credit, meaning less people will get credit cards and the balance available to them on those credit cards will also be restricted."
• Financial crime. Bank Secrecy Act and anti-money laundering programmes, plus sanctions compliance. Bank of America has disclosed that it is in contact with several federal regulators on aspects of its BSA/AML and sanctions compliance programmes.

Cyclicality

Banking is procyclical in a specific and uneven way. Net interest income is relatively resilient because balances are sticky. Fee income from wealth management falls with asset prices. Investment banking and trading are violently cyclical, though volatility itself often helps trading while it hurts underwriting. Credit costs are the killer: provisions are low and stable for years and then rise very fast in a downturn.

The way to read Bank of America's current results is that all four cyclical drivers are pointing the same way at once - loan growth strong, capital markets open, asset prices high, credit benign. That is a favourable configuration, not a permanent one. Management's own framing on the Q4 2025 call was that through-cycle normalised losses should be 50-55 basis points; the FY2025 net charge-off ratio was 0.50% and the Q1 2026 rate was 0.48%. The company is running at the good end of its own normalised range.

Tailwinds and headwinds

Tailwinds: deposit costs falling as rates on the short end stay contained; the mechanical repricing of low-yielding legacy assets into higher-yielding new assets; a regulatory capital regime turning more accommodative; AI-driven cost reduction that benefits the largest players most; loan growth running above nominal GDP.

Headwinds: the possibility that rate cuts resume, which compresses asset yields faster than deposit costs fall; deep-delinquency deterioration in consumer credit; private credit continuing to take share in lending; the very large stock of unrealised losses on legacy securities across the banking system, discussed in Section 8; and the risk that data-centre and AI-related credit, which has driven a portion of commercial loan growth industry-wide, proves less durable than underwritten.

7. Growth triggers

All items below come from the six concalls in the review period. Nothing here is analysis or forecast; it is what management said is coming.

• 2026 net interest income growth guided to the upper end of 6%-8%, raised twice in six months. Guided at 5%-7% on both the Q3 2025 call (15 Oct 2025) and Q4 2025 call (14 Jan 2026); raised to 6%-8% on the Q1 2026 call (15 Apr 2026); raised again to the upper end of that range on the Q2 2026 call (14 Jul 2026). Repeated and upgraded twice.

"We now expect full year 2026 NII growth to be at the upper end of that 6%-8% range."

• Alastair Borthwick, CFO (Q2 2026 concall, 14 July 2026)

• Fixed-rate asset repricing continues mechanically through 2026. Roughly $10-15 billion of mortgage-backed securities and mortgage loans roll off each quarter and are replaced at yields 150-200 basis points higher (Q3 2025 concall, 15 Oct 2025). Restated as $12-15 billion per quarter on the Q4 2025 call (14 Jan 2026). Repeated. This is the most important trigger in the list because it does not depend on Fed policy, loan growth or market conditions - the roll-off schedule is already contracted.

• Net interest margin target of 2.30%, timeline pulled forward. Set at the 5 November 2025 Investor Day on an approximately three-year horizon; on the Q2 2026 call management moved the timeline in by roughly a year. NIM was 2.08% in Q2 2026.

"We still feel good about that 230 number that we're aiming for. We're probably inside a couple of years now to get there."

• Alastair Borthwick, CFO (Q2 2026 concall, 14 July 2026)

• Full-year 2026 operating leverage guidance raised from 200 basis points to 300-400 basis points. Guided at approximately 200 basis points on the Q4 2025 call (14 Jan 2026); H1 2026 delivered over 450 basis points; raised to 300-400 basis points on the Q2 2026 call (14 Jul 2026). Upgraded.

• BofA Rewards relaunch, targeting membership growth from ~11 million toward 30 million. Relaunched late May 2026; over 2 million enrolments reported on the Q2 2026 call (14 Jul 2026), with the company subsequently disclosing over 3 million in the first seven weeks and total enrolment near 14 million. Roughly half of weekly new members are new to Bank of America.

• More than 165 new financial centres across 63 markets by end-2026, including entry into nine new markets and four new states. Described on the Q2 2026 call (14 Jul 2026) as "rational, full market build-outs in new and growth markets rather than one-at-a-time locations", with continued investment in centres, marketing and rewards flagged as the Consumer Banking growth programme.

• Credit card balance growth progressing toward a 5% target. Balances were growing at 4% in Q2 2026, up from 1% earlier in the cycle (Q2 2026 concall, 14 Jul 2026).

• Merrill AUM growth target of 5% over three to five years. Management was seven months into the plan at the Q2 2026 call (14 Jul 2026) and described it as "off to a good start", with advisor attrition at near historic lows.

• AI deployment scaling from pilots to production, with an explicit headcount consequence. Q4 2025 call (14 Jan 2026): AI reduced coding work by about 30%, equivalent to roughly 2,000 positions. Q1 2026 call (15 Apr 2026): 90 installations, all 200,000-plus employees with access, headcount down 1,070 from year-end 2025. Q2 2026 call (14 Jul 2026): 300 approved use cases, 114 live generative AI implementations, 34 fully deployed, 400,000-plus daily prompts. Repeated across all six calls in escalating specificity.

"We can just make decisions not to hire and let the headcount drift down."

• Brian Moynihan (Q4 2025 concall, 14 January 2026)

• A September 2026 rate hike is embedded in guidance, with quantified upside. Management's Q2 2026 NII guidance assumes one 25 basis point increase in September, and a 100 basis point increase would add approximately $1 billion to NII over 12 months (Q2 2026 concall, 14 Jul 2026).

• Commercial loan growth described as broad-based rather than AI-concentrated. Commercial lending grew 11% year over year with middle market around 8%, spread across business banking, the commercial bank and the corporate bank (Q2 2026 concall, 14 Jul 2026).

• International expansion, with Asia-Pacific called out. International revenue grew 38% year over year in Q2 2026, with Asia-Pacific identified as a standout (Q2 2026 concall, 14 Jul 2026).

• Capital return continues under the $40 billion authorisation. Authorised effective 1 August 2025; approximately $17 billion remained available as at 30 June 2026. The dividend was raised 14% to $0.32 per share in July 2026.

• Medium-term targets set at the 5 November 2025 Investor Day. ROTCE of 15% near term rising to 16%-18% medium term (up from the prior 12%-15% framework), efficiency ratio of 55%-59%, operating leverage of 200-300 basis points or more, deposit growth of 4%+, loan growth of 5%+, NII growth of 5%-7% CAGR, and EPS growth of 12%+. Reaffirmed in the 2025 shareholder letter as goals the team is "committed to" and "making good progress" against in 2026.

Trigger summary

8. Key risks

1. The held-to-maturity securities portfolio carries $82.1 billion of unrealised losses

The mechanism. During the zero-rate era Bank of America invested a very large share of its deposit inflow into long-dated mortgage-backed securities and Treasuries at low yields, and classified most of it as held-to-maturity. HTM securities are carried at amortised cost, so when rates rose the market value fell but the balance sheet and regulatory capital did not reflect it.

As at 30 June 2026 the HTM book had an amortised cost of $505.8 billion against a fair value of $423.7 billion, a gross unrealised loss of $82.1 billion, of which $68.3 billion sits in agency MBS and $13.0 billion in Treasuries and government agencies. The available-for-sale book, which is marked, carried only $3.2 billion of gross unrealised losses on $365.8 billion of amortised cost (Q2 2026 10-Q).

Why it matters even though it is not a capital hit. Three ways. First, it is a drag on earnings: that capital is locked into low-yielding assets, which is precisely why the $10-15 billion quarterly roll-off into 150-200 basis point higher yields is such a powerful NII tailwind. The unrealised loss and the NII tailwind are the same fact viewed from two directions. Second, it constrains flexibility: the bank cannot sell these securities to fund loan growth or reposition without crystallising the loss, so it must wait for maturity. Third, and this is the tail risk, if deposits ran, the bank would have to monetise assets, and this is the portfolio it would reach for. That is the mechanism that destroyed Silicon Valley Bank in 2023.

Calibration. Low probability of the catastrophic version, high certainty of the earnings-drag version. Bank of America's deposit base is granular, insured to a far greater degree than SVB's was, and overwhelmingly operational rather than rate-shopping. The company also holds close to $1 trillion of liquidity. The realistic risk is not a run; it is that the repricing tailwind everyone is counting on takes years and can be reversed by falling rates.

2. Rate sensitivity cuts both ways, and the guidance now assumes a hike

The mechanism. Bank of America is asset-sensitive. Management has quantified it repeatedly and the numbers are large. On the Q3 2025 call: an instantaneous 100 basis point decline would reduce NII by $2.2 billion over 12 months, with roughly 80% of the sensitivity on the short end. On the Q1 2026 call, Borthwick was specific about the downside scenario:

"an additional 100 basis point decline in rates beyond the forward curve, would reduce NII over the next 12 months by $2 billion."

• Alastair Borthwick, CFO (Q1 2026 concall, 15 April 2026)

The Q2 2026 guidance for NII growth at the upper end of 6%-8% assumes one 25 basis point hike in September 2026. That is a defensible assumption given the July 2026 FOMC saw three dissents in favour of a hike, but it is an assumption. The Fed has held at 3.50%-3.75% since December 2025, and if the energy-driven inflation impulse fades and the Fed resumes cutting instead, the mechanism runs in reverse.

Calibration. High probability of moderate impact. This is the most likely single source of a guidance miss over the next four quarters.

3. Two of the most cyclical revenue lines carried the Q2 2026 result

The mechanism. Q2 2026 revenue grew 15% and net income 27%. Look at where it came from: investment banking fees up 50% to $2.1 billion, Global Markets net income up 72% to $2.6 billion, equities revenue up 70% to a record $3.6 billion. Consumer Banking, the largest and most stable segment, grew net income 10%.

Investment banking fees and equities trading are among the most volatile lines in finance. A 50% year-over-year fee increase reflects an open capital markets window, not a structural share shift of that magnitude. Management said so themselves on the call: second-half comparisons are difficult because prior-year markets performance was elevated, and IB fees are "unlikely to repeat" the pace of that lift.

Calibration. High probability of moderate impact. Not a threat to the business model, but a real threat to the earnings trajectory that the current 17.0% ROTCE implies. A reader should mentally separate the durable half of the Q2 2026 beat (NII, operating leverage, cost) from the cyclical half.

4. Consumer credit is running at the good end of management's own normalised range

The mechanism. The FY2025 net charge-off ratio was 0.50% and Q1 2026 was 0.48%, against management's own stated through-cycle normalised expectation of 50-55 basis points (Q4 2025 concall, 14 Jan 2026). The credit card charge-off rate was 3.40% in Q4 2025 and 3.55% in Q2 2026 - a modest deterioration.

Industry data is more mixed than the headline suggests. The 30-day delinquency rate dipped to 2.92% in Q1 2026, but the share of card balances 90 or more days delinquent rose from 7.6% in Q3 2022 to 12.8% in Q1 2026. If unemployment rises from the 4.2% level Moynihan cited on the Q2 2026 call, provisions would rise quickly, and provisions have been running roughly flat to net charge-offs with modest reserve releases - meaning there is little cushion built in.

Calibration. Moderate probability, moderate-to-high impact. The mechanism is well understood and the loss content is knowable; the timing is not.

5. AI-linked and data-centre credit exposure

The mechanism. Commercial loan growth of 11% year over year is strong, and part of the industry-wide growth in commercial lending has been financing for data centres and AI infrastructure. Bank of America's own research team publicly warned in December 2025 of a potential "air pocket" in 2026 data centre debt, noting hyperscaler capex funded by operating cash flow was running out and increasingly being funded with debt.

An analyst put the second-derivative question directly to management on the Q2 2026 call. Moynihan's answer was that the company factors AI impact into underwriting across the portfolio, focusing on underlying tenant credit quality and earnings power, and stressed that commercial loan growth was broad-based rather than concentrated in that theme.

Calibration. Low-to-moderate probability, moderate impact. Management is aware of it and has explicitly disclaimed concentration, which is somewhat reassuring, but the disclosure is qualitative rather than quantified.

6. Regulatory and litigation overhang

The mechanism. Several live matters. Bank of America is a named defendant alongside JPMorgan Chase and Wells Fargo in the CFPB's action over fraud on the Zelle network, where the CFPB asserted customers of the three banks lost more than $870 million over seven years. The company has also disclosed potential exposure relating to its processing of unemployment benefits, automatic credit card payment service features, and its anti-money laundering and economic sanctions programmes, and has said it is in contact with several federal regulators on the BSA/AML and sanctions programmes. A separate lawsuit alleges the bank profited from dealings with Jeffrey Epstein.

Calibration. High probability of moderate cost, low probability of a large one. For a bank this size, litigation is an operating expense rather than an existential event. The tail risk is a consent order that constrains growth, which is what happened to Wells Fargo for the better part of a decade.

7. The efficiency ratio is already at the top of the target band

The mechanism. Investor Day set a medium-term efficiency ratio target of 55%-59%. Q2 2026 came in at 59%, the top of the range, helped materially by a 15% revenue quarter. FY2025 was 61%. If revenue growth decelerates as management themselves flag for H2 2026, the ratio moves the wrong way unless expenses fall in absolute terms.

More notable is what happened to disclosure. On the Q4 2025 call management gave specific expense guidance (Q1 2026 expenses up about 4% year over year). By Q2 2026 the company had moved away from specific expense guidance in favour of communicating through operating leverage, citing variable revenue-linked costs. That is a defensible reason, and it is also a reduction in accountability at exactly the point where expenses become the harder half of the story.

Calibration. Moderate probability, moderate impact. Worth watching precisely because the disclosure got looser as the metric got harder.

8. Deposit competition and the non-interest-bearing mix

The mechanism. Everything in this business rests on funding $2 trillion cheaply. Non-interest-bearing deposits have grown for seven consecutive quarters, up 4% year over year in Q2 2026, and the rate paid on total deposits fell 32 basis points year over year in Q3 2025. Borthwick has been explicit that the strategy is to compete on relationship value - digital capability, security, rewards - rather than on rate.

That strategy works while deposit competition is rational. If rates rise materially, or if a large competitor decides to buy deposits, or if money market funds again become dramatically more attractive, the non-interest-bearing balance is the first thing to move, and it is the single most valuable liability on the balance sheet.

Calibration. Low probability in the near term, very high impact. This is the risk that would genuinely change the business, and it is the one with the least visible early warning.

9. CEO succession

The mechanism. Moynihan has run the company since 1 January 2010 and has indicated he plans to stay through the end of the decade. On 12 September 2025 the company appointed Dean C. Athanasia and James P. DeMare as Co-Presidents and elevated Alastair Borthwick to Executive Vice President and CFO, which is widely read as lining up three potential successors. Per the 12 September 2025 announcement, Athanasia had served as President, Regional Banking since October 2021, President of Retail and Preferred and Small Business Banking from January 2019 to October 2021, Co-Head of Consumer Banking from September 2014 to January 2019, and Preferred and Small Business Banking Executive from April 2011 to September 2014; DeMare had been President, Global Markets and joined Bank of America in 2008.

Orderly succession planning is a positive. The risk is the standard one: a multi-year internal contest can produce departures among the executives who do not win, and the current strategy is unusually identified with one person.

Calibration. Low probability of near-term disruption, moderate impact.

9. Walk the talk

The six concalls used: Q1 2025 (15 April 2025), Q2 2025 (16 July 2025), Q3 2025 (15 October 2025), Q4 2025 (14 January 2026), Q1 2026 (15 April 2026), Q2 2026 (14 July 2026). The most recent is 58 days before this report date.

The starting point: a guidance number set in January 2025 and defended for a year

The commitment that anchors this whole period was made before the review window opened and then repeated inside it. In January 2025 and again on the Q1 2025 call (15 April 2025), management said they expected to exit Q4 2025 with net interest income on a fully taxable equivalent basis in a range of $15.5 billion to $15.7 billion, with growth accelerating in the second half of 2025.

This is a good test case because it is precise, dated, falsifiable and far enough out to be genuinely uncertain when made.

At Q1 2025 the company reported NII of $14.6 billion FTE, up 3% year over year, which management noted finished at the higher end of the expected range for the quarter. Revenue was $27.5 billion, up 6%, net income $7.4 billion, EPS $0.90, up 18%.

Three months later, on the Q2 2025 call (16 July 2025), the temptation to hedge would have been real: revenue growth had slowed to 4% and EPS growth to 7%, at $0.89. Borthwick did not hedge.

"Our expectation for the exit rate of NII in the fourth quarter remains unchanged."

• Alastair Borthwick, CFO (Q2 2025 concall, 16 July 2025)

By the Q3 2025 call (15 October 2025) the guidance moved up, not down: Q4 2025 NII was now expected at $15.6 billion or higher, roughly 8% growth on Q4 2024. The quarter itself was strong - revenue $28.2 billion up 11%, EPS $1.06 up 31%, ROTCE 15.4%.

The Q4 2025 result, reported 14 January 2026, delivered NII of $15.9 billion FTE. The target was $15.5-15.7 billion, set twelve months earlier and defended without revision through a soft second quarter. It was beaten. That is the single most credible thing in this record.

The second cycle: 2026 guidance, raised twice

Having beaten the 2025 exit rate, management set the 2026 bar deliberately low. On the Q3 2025 call (15 October 2025) they framed 2026 NII growth as "something like 5% to 7%", explaining the mechanics openly: $10-15 billion of MBS and mortgage loans rolling off each quarter and being replaced at 150-200 basis points higher yields. The Q4 2025 call (14 January 2026) restated 5%-7% with mid-single-digit loan growth, approximately 200 basis points of operating leverage and a 20% effective tax rate.

Then it moved. On the Q1 2026 call (15 April 2026) the NII range went to 6%-8%. On the Q2 2026 call (14 July 2026) it went to the upper end of 6%-8%. Operating leverage guidance went from "about 200 basis points" to 300-400 basis points, after H1 2026 delivered over 450 basis points.

The pattern across two full cycles is now legible: set a range you are confident in, refuse to lower it when a quarter disappoints, raise it when the evidence supports raising. That is the behaviour of management that guides conservatively.

The Investor Day commitments, and whether they were sandbagged

On 5 November 2025, Bank of America held its first Investor Day in roughly 14-15 years and reset the medium-term framework: ROTCE of 15% near term and 16%-18% medium term (up from the prior 12%-15%), efficiency ratio of 55%-59%, operating leverage of 200-300 basis points or more, deposit growth of 4%+, loan growth of 5%+, NII growth of 5%-7% CAGR, and EPS growth of 12%+.

The market's reaction was mild scepticism. The shares fell 1.8%, and the published analyst commentary was that the ROTCE target was "largely in line with expectations" but that "Bank of America will need to earn some credibility on the path and timing towards its target."

Eight months later the numbers look like this:

Five of six are met or exceeded. Deposit growth at 2% is the one that lags, and it is worth noting rather than glossing: the deposit franchise is the foundation of everything, and it is growing at half the target rate even as loans grow at 8%. That is a loan-to-deposit ratio drifting the wrong way over time, and it is the item to press management on.

The medium-term ROTCE range was hit in eight months rather than several years. That reads as a target set with room to spare, which is consistent with the guidance behaviour above but is also a reason not to extrapolate the 17.0%: it was delivered in a quarter where markets and IB revenue were exceptional.

The NIM target: pulled forward, which is rarer than it sounds

The 2.30% net interest margin target was set at Investor Day on roughly a three-year horizon. On the Q2 2026 call, with NIM at 2.08%, Borthwick moved it in:

"We still feel good about that 230 number that we're aiming for. We're probably inside a couple of years now to get there."

• Alastair Borthwick, CFO (Q2 2026 concall, 14 July 2026)

Companies routinely extend timelines. Pulling one in, eight months after setting it, in public, is uncommon and worth crediting.

The headcount commitment, which is the hardest one to fake

Cost discipline is where bank managements most often say one thing and do another, because expenses are the easiest place to accommodate a business case. Here is the actual series across the six calls.

At Q3 2025 (15 October 2025), management noted headcount had come down from 217,000 three years earlier to 213,000, and guided Q4 2025 expenses roughly in line with Q3 with headcount relatively flat. At Q4 2025 (14 January 2026) the company reported 213,207, essentially flat year over year - and did so having hired 17,000 new teammates and 2,000 college graduates during the year, meaning the flat number was achieved through genuine attrition management rather than a hiring freeze. At Q1 2026 (15 April 2026) headcount was 212,134, down 1,070 from year-end.

Moynihan's own framing, in Q1 2026:

"the application of technology, the process and the customer utilization of our technology has led us basically run the company 19 years later on less people."

• Brian Moynihan (Q1 2026 concall, 15 April 2026)

That claim is checkable against the shareholder letter, which puts 2007 headcount at approximately 210,000 before 76,000 acquired employees arrived in 2008. Today's 212,000 against a pro-forma 286,000. The claim holds.

The one that was quietly dropped

Not everything was delivered. On the Q4 2025 call (14 January 2026) management gave specific expense guidance: Q1 2026 expenses expected to rise about 4% year over year, within a 200 basis point operating leverage frame. By the Q2 2026 call, the company had moved away from specific expense guidance altogether, preferring to communicate through operating leverage because a growing share of costs are revenue-linked and variable.

The reason is legitimate. In a quarter where revenue rises 15% and markets revenue rises 33%, incentive compensation rises mechanically, and an absolute expense number becomes a poor measure of discipline. But the effect is that the company now discloses a ratio where it used to disclose a level, and a ratio flatters in a strong revenue quarter. This is the one place in the record where accountability got looser rather than tighter, and it happened without much comment.

Promise versus outcome

Assessment

This is management that does what it says, and specifically management that guides conservatively and then raises. The defining evidence is the 2025 NII exit-rate commitment: a precise number, set twelve months ahead, defended without revision through a quarter where the temptation to soften was real, and then beaten.

Two honest qualifications. First, the targets appear to have been set with room to spare - hitting a medium-term ROTCE range in eight months is a comment on the target as much as on the performance, and the quarter that got them there leaned on the two most cyclical lines in the business. Second, the expense guidance discontinuation is a small but real reduction in accountability, arriving at the moment when expense discipline becomes the harder half of the story. Neither undermines the record. Both are worth remembering when reading the next set of raises.

10. Shareholder friendliness index

Dividends. Declared dividends per share were $0.88 in 2023, $0.98 in 2024 and $1.06 in 2025 (dividend history), a rise of about 20% across the three years with an increase in every single year. The increases come once a year, in the third quarter, following the Federal Reserve's stress test results: 8% to $0.26 in July 2024, 8% to $0.28 in July 2025, and 14% to $0.32 in July 2026 (July 2026 dividend release). The July 2026 step is the largest in the sequence and marks a twelfth consecutive year of growth. The payout ratio of roughly 30% is unremarkable for a large bank and is the reason the dividend has never had to be cut in this period: there is ample room beneath it.

Buybacks and dilution. Two windows, stated separately. Last ~90 days: Bank of America repurchased $6 billion of stock in Q2 2026 alongside $2 billion of dividends, for $8 billion returned in the quarter, and approximately $17 billion remained available under the authorisation as at 30 June 2026 (Q2 2026 concall, 14 July 2026). Older than 90 days: the Board authorised a $25 billion programme effective 1 August 2024 and replaced it with a $40 billion programme effective 1 August 2025, announced 23 July 2025, at which point roughly $9.1 billion remained on the prior authorisation ($40bn repurchase authorisation). Execution has accelerated hard: repurchases were roughly $4.6 billion in 2023, $13.1 billion in 2024 and $24.1 billion in 2025, with total capital returned in 2025 of over $30 billion, up 41% on 2024 (Q4 2025 concall, 14 January 2026). Quarterly buybacks ran $5.3 billion in Q2 2025, $5.3 billion in Q3 2025, $6.3 billion in Q4 2025, $7.2 billion in Q1 2026 and $6.0 billion in Q2 2026. The share count is genuinely shrinking, not merely offsetting compensation: common shares outstanding fell from 7.61 billion at Q4 2024 to 7.21 billion at Q4 2025 and to roughly 6.99 billion currently, against a basic average of about 8.03 billion in 2023 - a reduction of roughly 13% over three years, and 3.7% in the last twelve months alone.

Verdict: Returns Capital. Twelve consecutive years of dividend growth capped by a 14% raise, a share count down roughly 13% in three years, and buyback execution that tripled in size between 2023 and 2025 while the payout ratio stayed near 30%.

11. Insider activities

Bank of America is US-listed, so the primary source is SEC Form 4 filings via EDGAR. No MoatMap database block was supplied with this request, so the transactions below are drawn from EDGAR Form 4 filings and the Form 4 aggregators that index them (secform4 BAC filings, SEC EDGAR CIK 0000070858). Where a filing footnote explains the transaction, that explanation is used; where it does not, the reason is stated as not disclosed.

Recent transactions

Buys - reading the signal

There were no open-market purchases by any director or executive officer of Bank of America in the twelve months to 10 September 2026 that I could locate in the Form 4 record. Every acquisition in the period was either an equity award (the 4 May 2026 annual director option grants), a performance share unit vesting, or a restricted stock unit settlement. None of these involves an insider committing personal after-tax cash at the prevailing market price, which is the only kind of buying that carries an information signal.

This matters for how the sells below should be read, but it is also worth putting in context. Insider buying is rare at very large US banks generally: executives are already paid overwhelmingly in stock, hold large positions by mandate under share ownership guidelines, and are heavily constrained by blackout windows around quarterly earnings and around every material transaction the firm advises on. The absence of buying at Bank of America is normal for the peer group, not a deviation from it. It is still an absence, and it means there is no bullish conviction signal in this section.

Sells - working out the why

The single most important thing to get right here is that the largest and most frequent line in the Form 4 record is not a sale at all.

Moynihan's recurring 18,083-share transactions on the 15th of each month are settlements of cash-settled restricted stock units - instruments economically equivalent to one share each, which on vesting are converted and the underlying shares disposed of back to the issuer, not into the market. They are automatic, they occur on a fixed calendar date, and they represent the payout of deferred compensation awarded in prior years. Reading them as the CEO selling $1 million of stock a month misreads the instrument. His reported holdings across these filings remained in the 2.9-3.0 million share range throughout, direct and indirect, which is the confirming detail: the position is not being run down. The rising per-share settlement prices across the sequence ($52.59 in January, $49.77 in May, $55.87 in June, $61.59 in July, $64.49 in August) simply track the share price through the year and tell you nothing about intent.

The genuine sales are these:

Geoffrey S. Greener, Chief Risk Officer, 126,756 shares at $53.01 on 5 May 2026 (~$6.7 million). This is the largest single insider disposition of the period and the one most worth flagging. The reason is not disclosed in the filing and is not inferable from the surrounding record. It was executed three weeks after the Q1 2026 results (15 April 2026), which places it in a normal post-earnings open window rather than anywhere unusual. I could not locate a Rule 10b5-1 plan reference for this transaction, so it is recorded here as reason not disclosed rather than assigned one.

The March 2026 cluster. Alastair Borthwick (CFO), James P. DeMare (Co-President), Sheri B. Bronstein (Chief People Officer) and Thomas M. Scrivener (Chief Operations Executive) all sold in the same window around 1 March 2026. This looks like a cluster and is not one. The mechanism is visible in Borthwick's filing, which is the most fully described: 79,101 performance restricted stock units from the 2023 grant vested, 43,743 shares were withheld at $49.83 to satisfy tax withholding, and 68,000 were sold in the open market at $50.24. Bank of America's PRSUs vest on a common annual date, so the entire senior team's awards convert at once and the entire senior team faces a tax bill at once. Simultaneous selling driven by a simultaneous vesting event is a compensation calendar artefact, not a coordinated view. The tell that distinguishes the two is that these sales cluster on the vesting date rather than on any news date.

Borthwick's is nonetheless a real sale beyond the tax obligation: 68,000 shares sold on top of 43,743 withheld. That is post-vest diversification, the most common and least informative reason an executive sells, but it is a choice rather than an automatic mechanic.

Thomas D. Woods, Director, 2,473 shares at $52.19 on 6 May 2026. Tax withholding on a vested director award. Sub-threshold housekeeping, listed only because it sits inside the same week as the board grant.

The largest seller is not an insider at all

The most consequential selling in Bank of America stock over this period sits outside the Form 4 record entirely. Berkshire Hathaway, historically the largest shareholder, has been reducing continuously. It sold 464,781,994 shares - approximately 45% of the position - between 17 July 2024 and 30 September 2025, and continued into 2026, selling roughly 30.2 million shares for an estimated $1.6 billion in Q2 2026, its eighth consecutive quarter of reduction. Bank of America nonetheless remains among Berkshire's largest holdings at around 8-9% of the equity portfolio.

This is disclosed through 13F filings rather than Form 4, and it belongs in a different mental category from officer selling. Berkshire is a portfolio allocator reducing a concentrated position across a leadership transition, and published commentary has linked the decision to the Fed's easing cycle and its implications for bank net interest income - a thesis that, on the evidence in Section 9, has not played out as feared, with NII guidance raised twice and a rate hike now the base case. A seller of that size creates a persistent supply overhang regardless of the reason, and it is worth knowing about. It is not a statement by management about the business.

Net assessment

Insiders are net sellers, but almost none of it carries information. Strip out the cash-settled RSU settlements that are automatic, the annual board option grants that are compensation, and the tax withholding on vested awards, and what remains is Greener's $6.7 million disposition in May 2026 and Borthwick's 68,000-share open-market leg in March 2026, plus three undescribed March sales that sit on the same PRSU vesting date. Activity is broad-based across the senior team rather than concentrated in one person, which is consistent with a compensation calendar rather than a view.

Nothing has changed recently in either direction: no executive has bought for the first time in years, and no one has liquidated a position. Moynihan's holdings sat around 2.9-3.0 million shares throughout.

Plain read: neutral. There is no bullish signal here because there is no open-market buying, and there is no bearish signal because nearly all the selling is mechanically explained by vesting and settlement schedules. The one item that deserves a second look rather than dismissal is the Chief Risk Officer's $6.7 million sale with no disclosed reason, which is large in absolute terms and comes from the one seat in the company with the clearest early view of credit deterioration. On its own, and against a benign credit backdrop that management has repeatedly described as strong, it is not enough to move the assessment off neutral. It is enough to watch whether it repeats.

12. Scenarios

Bull case

The repricing engine runs its full course and nobody has to do anything clever for it to work. Every quarter, $12-15 billion of mortgage-backed securities and mortgage loans bought during the zero-rate era mature and get reinvested 150 to 200 basis points higher. That is a contracted, arithmetic tailwind that continues whether the Fed hikes, holds or cuts, and it carries the net interest margin from 2.08% toward the 2.30% target that management has already pulled a year forward. The Fed hikes in September and holds through 2027 as energy-driven inflation proves stickier than expected, and the asset-sensitive balance sheet catches roughly a billion dollars of additional net interest income for every hundred basis points.

Meanwhile the cost line does something banks have historically been unable to do: it goes down while revenue goes up. The AI programme moves from 114 live implementations to hundreds, the coding productivity gains extend into operations, servicing and compliance, and headcount drifts from 212,000 toward 200,000 without a redundancy programme, purely through attrition and hiring restraint. The efficiency ratio drops through 59% into the middle of the 55-59% band, and the operating leverage that ran at 450 basis points in the first half of 2026 does not decay as fast as management's cautious second-half framing implies.

On the growth side, the BofA Rewards relaunch does what it was designed to do. Membership climbs from 14 million toward the 30 million target, half of new enrolments are genuinely new to the bank, and the multiplier structure pulls card spend and deposit balances out of Capital One and Chase. Deposit growth finally catches up to the 4% Investor Day target, which fixes the one metric currently lagging. Merrill's 5% AUM growth plan compounds against near-historic-low advisor attrition, so the wealth business stops being the slowest grower and becomes the most predictable one. The 165 new financial centres across 63 markets mature into their local small-business franchises, and the four new states start contributing primary relationships rather than costs.

The regulatory environment keeps easing. The March 2026 capital re-proposals are finalised broadly as drafted, large-bank requirements come down modestly, and the GSIB surcharge recalibration frees capital that goes straight into buybacks. The share count, already down 13% in three years, keeps shrinking at 3-4% a year. Two or three years out, this is a company earning at the upper half of a 16-18% return on tangible common equity range, on fewer shares, with a cost base that is smaller in absolute terms than it is today, and with a deposit franchise that no competitor has managed to dent.

Base case

Most of the above happens more slowly and less completely, and that is still a good outcome.

The repricing tailwind is real and mechanical, so net interest income grows in the 5-7% range that management has guided as the medium-term CAGR, landing at the upper end of 6-8% for 2026 as promised and then settling back toward the middle of the range as the easiest vintages of the securities book finish rolling. The Fed's path is genuinely uncertain and roughly cancels out: a September hike helps, a resumption of cuts in 2027 hurts, and the net effect over three years is close to neutral relative to the forward curve embedded in guidance.

Operating leverage moderates from the exceptional first half of 2026 to the 200-300 basis points that Investor Day framed as the steady state. The AI programme delivers, but at the pace of an institution rather than a startup: headcount drifts down by low single-digit percentages a year, the efficiency ratio grinds toward the high 50s rather than the mid 50s, and the savings are partly reinvested into the branch build-out, technology and the rewards programme rather than dropping entirely to the bottom line. That reinvestment is the right call and it is also why the efficiency ratio improves more slowly than the bull case assumes.

Global Markets and investment banking normalise. The 50% year-over-year jump in investment banking fees and the 72% jump in Global Markets net income seen in Q2 2026 were a capital markets window, and the window narrows. Trading revenue stays at a structurally higher level than 2023 because the equities share gains are real, but the growth rate reverts to something in the mid-single digits. That is the arithmetic that turns a 17.0% return on tangible common equity in one quarter into something closer to 15-16% sustained, which is exactly where Investor Day said the near-term target was.

Consumer credit normalises gently rather than sharply. Net charge-offs drift from 0.48% toward the 50-55 basis point through-cycle range management has flagged, card losses tick up from 3.55%, and provisions rise modestly, taking a bite out of earnings growth without threatening the model. Deposit growth stays around 2-3%, short of the 4% target, so the loan-to-deposit ratio creeps up and management has to work harder for funding than it did in 2025.

Capital return continues at roughly the current pace: dividend increases in the high single digits to low teens each July, buybacks of $5-7 billion a quarter, share count down 3-4% a year. The Zelle litigation and the BSA/AML discussions with regulators are settled for sums that are material headlines and immaterial to a balance sheet this size. Moynihan stays through the end of the decade as indicated, and one of the two co-presidents or the CFO succeeds him without disruption.

Bear case

The bear case does not require anything exotic. It requires the four cyclical drivers that are currently all pointing the same way to stop doing so at once.

The inflation impulse that made a September 2026 hike the base case fades quickly, the Fed resumes cutting, and the curve flattens from the front. Management has quantified this precisely: a hundred basis points of cuts beyond the forward curve takes $2 billion out of net interest income over twelve months. Deposit costs do not fall fast enough to compensate because the low-hanging repricing has already been done, and the mechanical asset-repricing tailwind, while still positive, is no longer large enough to carry the whole story. The 2.30% net interest margin target slips back out to the original three-year horizon and then past it.

At the same time the capital markets window closes. Investment banking fees do not repeat a 50% growth year, and the equities trading franchise that grew 70% in a single quarter gives back a meaningful part of that as volatility normalises and share gains prove partly cyclical rather than structural. Because these two lines carried a disproportionate share of the Q2 2026 result, the revenue growth rate falls faster than the market expects. Operating leverage turns negative for a quarter or two, and this is where the discontinued expense guidance becomes uncomfortable: the company now reports a ratio rather than a level, and in a falling-revenue quarter a ratio tells you less than a level would.

Credit is the accelerant. Unemployment rises from 4.2%, the deep-delinquency trend already visible in industry card data (90-day-plus balances up from 7.6% in Q3 2022 to 12.8% in Q1 2026) shows up in Bank of America's book, and net charge-offs move through the 50-55 basis point normalised range rather than stopping inside it. Provisions have been running roughly flat to charge-offs with modest reserve releases, so there is no cushion; the build has to come out of earnings. Layered on top, the data centre and AI-adjacent commercial credit that the bank's own research team warned about in December 2025 produces the "air pocket" they described, and a portion of the 11% commercial loan growth turns out to have been underwritten against tenant cash flows that did not materialise.

The structural squeeze runs underneath all of it. The $82.1 billion of unrealised losses on the held-to-maturity book means a very large slice of the balance sheet is locked into low-yielding assets that cannot be sold without crystallising a loss. If deposits get competitive - a rival buys balances, money market funds get attractive again, or the seven-quarter run of non-interest-bearing deposit growth reverses - the bank cannot fund the gap by selling securities without taking the hit. Deposit growth is already the weakest metric against Investor Day targets at 2% against a 4% goal, while loans grow at 8%. That gap is small today and it is the one that compounds.

The endpoint is not a crisis. It is a company that spent 2025 and 2026 telling the market it could sustainably earn 16-18% on tangible common equity, and then earns 12-13% for a few years while credit costs normalise and the cyclical businesses mean-revert, with the medium-term targets quietly extended rather than missed. Given that the shares re-rated on the strength of a first Investor Day in fifteen years and two consecutive guidance raises, the disappointment would be about the gap between the narrative and the through-cycle reality rather than about anything breaking.