Khan Capitals | October 2026
Key Takeaways
- Six major US banks opened the Q3 2026 season with roughly $43 billion of combined profit, but the dispersion was the story: Goldman Sachs beat consensus by more than 30 per cent while JPMorgan posted a rare headline miss.
- Equities desks paid for the quarter. Goldman’s equities trading revenue jumped 72 per cent to $6.44 billion, JPMorgan’s rose 49 per cent and Morgan Stanley’s 41 per cent to $5.8 billion, the direct dividend of a year of volatility across rates, AI stocks and geopolitics.
- Fixed income desks went the other way, with JPMorgan’s FICC revenue down 4 per cent and Morgan Stanley’s down 7, an odd result in a historic bond rout that suggests client risk appetite in rates products is thinning rather than churning.
- Bank of America was the season’s first casualty, missing on revenue with a $1.2 billion anti-money-laundering settlement charge, net income falling to $6.9 billion, and the shares down around 5 per cent.
- The lending books are quiet, for now. Net interest income ground higher across the group and executives warned in unison about higher rates and sticky inflation, while the credit costs of a 4 per cent Fed have yet to appear in the provisions line.
The Mirror the Banks Hold Up
Bank earnings Q3 2026 arrived in the middle of the most volatile week of the autumn, and the results read like a mirror of the markets that produced them. Every force that whipsawed investors this year, the Fed’s return to hiking, the AI boom and its October credit scare, the bond rout, the tariff economy, showed up somewhere in the six big P&Ls that reported between Tuesday and Thursday. Volatility made the trading floors rich. The same volatility has started, at the margin, to make the lending businesses nervous.
The aggregate numbers were strong: roughly $43 billion of combined net income across JPMorgan, Goldman Sachs, Morgan Stanley, Citigroup, Wells Fargo and Bank of America. But aggregates hide the week’s real information, which sat in the gaps: between the best beat and the worst miss, between the equities desks and the rates desks, and between what executives earned and what they warned about.
The Scorecard: One Outlier in Each Direction
| Bank | EPS vs consensus | Revenue | The line that mattered |
|---|---|---|---|
| Goldman Sachs | $17.01 vs $13.02 | $18.68bn, +23% | Equities trading +72% to $6.44bn |
| Citigroup | $2.85 vs $2.65 | $23.85bn, +8% | Net income +40% |
| Morgan Stanley | $3.03 vs $2.91 | $20.34bn, +12% | Record $9.07bn wealth management revenue |
| Wells Fargo | $1.87 vs $1.84 | $21.87bn, short of estimates | Loans and deposits both +11% |
| Bank of America | $1.08 adj vs $1.10 | ~$30.0bn vs $30.6bn expected | $1.2bn AML settlement charge; shares -5% |
| JPMorgan | $5.57 vs $5.93 | $50.17bn | Equities +49%; FICC -4%; securities-loss item |
Goldman’s beat deserves its scale noted: $17.01 of earnings per share against a $13.02 consensus is not a beat, it is a repudiation of the estimates. JPMorgan’s miss deserves its asterisk equally: excluding a securities-loss item the gap narrows to $5.75 against $5.93, and the franchise beneath it grew equities trading 49 per cent and investment banking fees 14 per cent. The market’s problem with the quarter was less the arithmetic than the symbolism. JPMorgan does not miss, and in a week when investors were already re-examining every assumption about financial assets, even a technical miss from the industry’s bellwether added to the unease.
Volatility Finally Sent an Invoice
The engine of the quarter was the equities floor. Three firms disclosed growth that would have seemed implausible a year ago: 72 per cent at Goldman, 49 per cent at JPMorgan, 41 per cent at Morgan Stanley. The explanation is the year itself. A Fed that stopped cutting and started hiking, an AI trade that has sold off three times since June, a bond market at 2002 yields, and a tariff war that reprices sectors weekly have kept institutional clients hedging, rotating and re-levering continuously. Every one of those episodes, painful for asset owners, was a fee event for the desks that intermediate them.
The fixed income desks tell the stranger story. JPMorgan’s FICC revenue fell 4 per cent and Morgan Stanley’s 7, with Goldman’s down 24 per cent from the second quarter, in a period when Treasury yields hit quarter-century highs and the Treasury itself intervened with buybacks. A bond rout should, in theory, be a rates desk’s harvest. That it was not suggests the flow is one-directional: clients reducing duration and stepping back from rates risk rather than trading it, consistent with the buyers’ strike this publication examined when the 10-year broke 5 per cent. Dealers monetise two-way markets; October’s bond market has mostly had one side.
Bank of America’s $1.2 Billion Reminder
Wednesday belonged, unhappily, to Bank of America. Revenue of roughly $30 billion missed estimates, net income fell to $6.9 billion from $8.3 billion a year earlier, and the quarter absorbed a $1.2 billion regulatory settlement charge tied to anti-money-laundering compliance. The shares fell around 5 per cent into a tape that was already selling financial complexity of every kind. Beneath the charge, the franchise looked steadier than the headline: net interest income rose 8 per cent to about $16.5 billion and deposits held above $2 trillion. But the episode illustrated how little tolerance this market has for negative surprises, and how quickly a one-off charge becomes a referendum on management in a nervous week.
What the Lending Books Are Not Saying Yet
The quietest line in every release was the one investors should watch hardest: credit costs. A Federal Reserve at 3.75 to 4 per cent and promising more, a consumer whose confidence sits at twelve-year lows, and a corporate sector carrying half a trillion dollars of freshly issued AI-related debt would all, in textbook terms, argue for provisions building. They are not, much, yet. Executives chose words instead, warning in near-unison about higher rates and sticky inflation while loan growth at Wells Fargo ran at 11 per cent.
The AI financing question touches the banks directly, and this season offered one concrete data point: JPMorgan is among the lenders that have spent more than six months distributing $38 billion of construction debt tied to an Oracle-leased data centre project, the overhang examined in this week’s AI debt selloff analysis. Underwriting exposure of that kind sits in warehouses, not provisions, and it surfaces only when syndication fails or marks move. Nothing in the quarter forced that disclosure. The 2007 lesson, that bank P&Ls reveal credit turns late and all at once, is the reason the provisions line deserves more attention than the trading line next quarter.
The Hiking Cycle’s Second Face
Rising rates built this earnings season twice over. Net interest income, the spread between what banks pay for deposits and earn on assets, expanded across the group as the Fed’s September hike repriced loan books faster than funding costs. And the volatility that rate uncertainty creates filled the trading floors. That is the benign face of a tightening cycle for bank shareholders, and it is the one on display this week.
| Phase | P&L expression | Status in Q3 2026 | Typical lag from rate peak |
|---|---|---|---|
| Benign: margins widen | Net interest income expands as loans reprice | Fully visible; NII up across the group | Immediate |
| Benign: volatility pays | Trading and hedging revenue surges | Fully visible; equities desks +41 to +72% | Concurrent |
| Lagged: credit strains | Delinquencies, CRE refinancing stress, provisions build | Verbal warnings only; provisions quiet | 12 to 18 months |
| Lagged: losses realised | Charge-offs, warehouse marks, syndication losses | Not visible; $38bn AI construction debt still unsold | 18 months plus |
The second face arrives later. Historically, the credit costs of a hiking cycle lag the rate peak by twelve to eighteen months, surfacing first in consumer delinquencies and commercial property refinancings, then in corporate defaults. With markets pricing a December hike at over 90 per cent odds and no cuts in sight, the current cycle’s peak is still ahead, which means its credit bill has not yet been dated, let alone delivered. The banks’ own guidance, cautious words over cautious numbers, reads like management teams positioning for that sequence without wanting to provision for it prematurely.
The Quiet Winners: Citi’s Rebuild and Wells Fargo’s Uncapped Year
Away from the extremes, two franchises quietly had their best quarters in years. Citigroup grew revenue 8 per cent and net income 40 per cent, with earnings per share up 27 per cent even after stripping the prior year’s goodwill impairment from the comparison. For a bank that has spent half a decade in restructuring, a clean double-digit beat with every division contributing is the first evidence the simplification programme is compounding rather than merely shrinking. Wells Fargo, in its second year free of the asset cap, grew loans and deposits 11 per cent apiece, growth that was structurally impossible for it as recently as 2024. Its revenue shortfall against estimates cost the stock on the day, but an 11 per cent loan growth rate into a 4 per cent Fed is the clearest expression among the six of a bank actually using the cycle rather than hedging against it. Whether that proves brave or early depends entirely on the credit chapter still to come.
Investor Implications
Equities. Bank shares are increasingly a dispersion trade rather than a sector trade. Firms weighted to markets businesses, Goldman and Morgan Stanley this quarter, are compounding volatility into earnings, while balance-sheet-heavy franchises carry the unpriced credit question. The season’s reaction function was also informative: beats were rewarded modestly, misses punished hard, the same asymmetric grading applied to AI results. In both sectors, the market is paying for resilience, not for growth.
Fixed income. The FICC slowdown corroborates what Treasury auction tails already suggested: genuine two-way liquidity in rates markets is thinning at these yield levels. For credit investors, the banks’ reluctance to build provisions while warning verbally is itself a signal, and bank senior paper continues to price a more benign credit path than the CDS of their most levered borrowers. One of those markets is wrong.
Cross-asset. The banks are now the clearest public window into the AI financing chain, from underwriting warehouses to construction lending. Disclosure on that exposure is minimal; the $38 billion syndication overhang surfaced through reporting, not filings. Until it improves, every bank earnings call doubles as a proxy hearing on the credit quality of the AI buildout, and the questions analysts asked this week suggest they know it.
What to Watch
- 27 to 28 October: the FOMC decision; a hold is priced, but the statement’s language on credit conditions will matter more to bank investors than the rate itself.
- Through November: any clearing of the $38 billion Oracle-linked construction syndication, the most visible test of bank appetite for AI credit.
- December FOMC, priced above 90 per cent for a hike: each additional 25 basis points extends NII tailwinds and advances the credit-cost clock simultaneously.
- January: Q4 reports bring full-year provisioning decisions; the first meaningful reserve builds, if they come, traditionally arrive with annual results.
Conclusion
The opening week of bank earnings season delivered a paradox that is easy to misread as strength. Six franchises earned roughly $43 billion by intermediating everyone else’s anxiety: the hedging, rotating and de-risking that a hiking cycle, a bond rout and an AI credit scare generate. That is real profit, and the equities numbers were genuinely historic. But the other half of the banking business, bearing credit risk through a tightening cycle, has not yet presented its bill. The desks have been paid for the volatility. The books have not yet paid for what the volatility is telling everyone about where credit goes next. Investors got the first invoice this week; the second is still in the post.
Sources: CNBC, bank earnings live coverage; Investing.com, Q3 2026 bank earnings analysis; Fortune, banks warn on rates and inflation; Forbes, lenders steady, Wall Street desks surge; TradingView News, Bank of America misses Q3 revenue estimates; Benzinga, analysts cut Bank of America forecasts.
Related Reading: The credit scare that framed the week is analysed in the AI debt selloff of October 2026, and the policy backdrop in the Fed’s first hike since 2023. The bond market conditions behind the FICC puzzle are set out in the arrival of the 5 per cent world, and the state of the US consumer in confidence at a twelve-year low. For the fundamentals, start with how earnings season works and what bank capital actually is.
Leave a Reply