Q1-2026 Bank Earnings: Capital Relief, NIM Divergence, and the AI Efficiency Wager

The paragraph most coverage missed this quarter was not in any prepared remarks about NIM or buybacks. It was in the regulatory commentary sections, where nearly every institution — from JPMorgan Chase to Truist to Ally Financial — disclosed preliminary estimates of Basel III endgame RWA relief ranging from roughly 90 basis points to more than 100 basis points of CET1 benefit. Read alongside the buyback figures — JPM at $8.33B, BAC at $7.20B, Citi at $6.30B in a single quarter — and the implication is plain: management teams are already pricing in capital distributions predicated on rules that have not yet been finalized. The public comment period on the Basel III NPR closes in mid-June, per Bank of America's prepared remarks. The gap between preliminary estimates and final rulemaking is where exam-cycle risk lives.

The second thread the news cycle underweighted was the structural divergence in NIM between G-SIBs and regionals. Wells Fargo's NIM compressed 13 basis points quarter-over-quarter to 2.47% — the sharpest sequential decline among the large-cap reporters in the data — while PNC expanded 11 basis points to 2.95% and Huntington expanded 9 basis points to 3.24%. Those are not random outcomes. They reflect fundamentally different liability structures: consumer-deposit-heavy G-SIBs repricing their back books downward as promotional deposit rates normalize, while asset-sensitive regionals with fixed-rate loan portfolios rolling into higher yields finally see the payoff from years of balance sheet patience. The mechanism is deposit beta in reverse: institutions that paid up aggressively to retain retail deposits in 2024 and early 2025 are now watching those costs compress more slowly than asset yields.

The third story — the one that will occupy examiners more than investors — is the uniform language around qualitative reserve builds tied to Middle East energy risk. Citi disclosed an eight-quarter weighted average unemployment rate of approximately 5.4% embedded in its reserve model, including a downside scenario average of nearly 7%. Regions attributed approximately $17 million of allowance growth to macro uncertainty. KeyCorp said it had added to "already elevated qualitative loan loss reserves." None of these disclosures suggests a credit crisis; all of them suggest that the CECL overlay conversation — how much qualitative adjustment is defensible, how it is documented, and how examiners will evaluate it in the next supervisory cycle — is already live.

The fourth thread is the stablecoin and tokenization disclosures, which have moved from aspirational to operational language with unusual speed. SoFi disclosed that it began minting SoFiUSD in Q1, with settlement capabilities across Mastercard's global payments network. Morgan Stanley launched a digital asset pilot through a partnership with Zero Hash enabling select clients to trade major digital currencies via eTrade. Robinhood launched the public testnet for Robinhood Chain, an Ethereum Layer 2 designed to support tokenized real-world assets, which processed over 100 million transactions. State Street is preparing to launch the State Street Galaxy Onchain Liquidity Sweep Fund. JPMorgan's Kinexys continues to expand programmable money and tokenized deposit functionality in wholesale payments. The question for bank counsel is no longer whether digital assets are coming; it is whether the supervisory framework will arrive before the business lines do.

The cross-quarter signal

The dominant cross-bucket theme this quarter is the Basel III capital relief trade, and it deserves more forensic attention than it has received. The proposed rules, as read by management teams, would deliver RWA reductions across the standardized approach — primarily through better-calibrated credit risk weights for residential mortgages, investment-grade corporates, and well-secured commercial loans. TFC estimates a 9% RWA decline under the revised standardized approach and 11% under ERBA. RF estimates a 10% RWA reduction contributing approximately 100 basis points of CET1 improvement. PNC estimates approximately $45-50 billion of RWA relief, also approximately 10%. MTB sees roughly 90 basis points of CET1 benefit. KEY's preliminary estimate shows a 100-plus basis point benefit to its marked CET1 ratio. The consistency of these estimates — clustering around 90-110 basis points of CET1 improvement — is itself a signal: banks with standardized-approach portfolios dominated by consumer mortgages and relationship commercial lending are the primary beneficiaries of the repricing, and management teams have been doing the math for months.

For JPM, the math runs in a different direction. JPM disclosed that under the Basel III proposals, its G-SIB surcharge is estimated at 5.2% in 2028, a 70 basis point increase from the current 4.5%, requiring approximately $20 billion of additional G-SIB capital. JPM's CET1 would increase approximately 4% under the proposals while the Federal Reserve estimates roughly a 5% reduction for large banks in aggregate — meaning JPM would absorb disproportionate capital requirements relative to the peer group. That asymmetry explains why JPM's prepared remarks were notably more critical of the methodology than those of regional peers celebrating projected capital relief.

NIM trajectory deserves the cross-bucket treatment it rarely receives. Among G-SIBs, BAC's NIM declined 0.01 percentage points to 2.07%, JPM's declined 4 basis points to 2.50%, C's declined 3 basis points to 2.46%, and WFC's declined 13 basis points to 2.47%. BK's NIM held flat at 1.38% — structurally low because BK is a custody bank whose NIM economics are not comparable to lending institutions. STT expanded 6 basis points to 1.16%, also a custody-bank outlier. Among regionals, the picture inverts: HBAN expanded 9 basis points to 3.24%, PNC expanded 11 basis points to 2.95%, KEY expanded 5 basis points to 2.87%, MTB expanded 2 basis points to 3.71%, and FITB expanded 17 basis points to 3.30%. The funding-mix mechanism is straightforward: regionals carry a higher proportion of commercial and relationship deposits that repriced aggressively in the rate cycle but are now normalizing, while their fixed-rate commercial loan books are rolling at higher yields. Consumer-facing G-SIBs, by contrast, hold larger shares of retail savings and money market deposits that repriced late and are now a drag on liability cost reduction.

The AI efficiency narrative is the quarter's most repeated and least verified claim. BAC described using AI to reduce FTEs through "detailed process reengineering." WFC reported that Fargo, its AI-powered virtual assistant, reached over one billion customer interactions less than three years since launch, and cited 23 consecutive quarters of headcount reductions. BK has over 200 AI solutions in production and described moving from point solutions to end-to-end process improvements. STT has 70 AI use cases live and expects "tangible business impact to begin emerging in the back half of 2026" — a notably precise caveat. LC reported a record automation rate above 90% for issued loans and over 60 active AI initiatives. Block disclosed that Builderbot is making 15% of production code changes nearly fully autonomously, with production code changes per engineer up more than 2.5x compared to January. COIN reported a 78% year-over-year increase in pull requests per engineer. The pattern across banks and fintechs is consistent: AI is measurably improving engineering and operations throughput; the translation to expense ratios remains a next-quarter and next-year claim rather than a current-period fact. CROs and CCOs should note that the qualitative risk disclosures around AI are also hardening — JPM called cyber risk the firm's largest risk and noted AI has made it worse, BK cited AI as a tool for both productivity and adversarial attack, and WAL's regulatory outlook section flagged "emerging external focus among regulators and other officials related to risks in connection with the development and use of artificial intelligence."

Credit quality remained well-behaved in aggregate but the details warrant attention. NCOs at Ally Financial spiked 63 basis points quarter-over-quarter to 1.97%, driven by auto credit seasoning — though Ally attributed the sequential move to portfolio dynamics rather than systemic deterioration, and the firm emphasized auto applications were up 16% year-over-year even as it moderated origination volumes. LC's NCO rate of 3.50% is structurally elevated for unsecured consumer lending but improved 20 basis points sequentially. SOFI's NCO of 2.04% rose 9 basis points. Among regionals, TFC's NCO ratio rose 4 basis points to 0.61%, with management specifically noting the increase in indirect auto nonperforming loans reflected a change in nonaccrual criteria effective January 1, 2026 rather than underlying credit deterioration — a disclosure that examiners reviewing NPL trend data should weight carefully. HBAN's theme flags carry both reserve_build and credit_deterioration simultaneously. The picture at the G-SIB level is benign: BAC at 0.48%, WFC at 0.45%, JPM not reported in the NCO field but with management commentary flagging private credit underwriting weakening. GS reported zero NCO for the quarter.

The private credit commentary is worth isolating as a cross-institution signal. JPM estimated that approximately half of private credit market share was regulatory arbitrage that banks could potentially recapture, and described willingness to compete side-by-side on investment-grade direct lending. MS noted that private credit markets "experienced a learning moment with both lenders and borrowers being scrutinized carefully." RF disclosed private credit exposure of less than 2% of total loans, characterizing it as limited. FITB noted that private credit exposures combined with BDCs represent less than 1% of total loans. The convergence of these disclosures — banks simultaneously minimizing their own private credit exposure while signaling intent to recapture lending share — suggests the competitive dynamic between bank lending and private credit is entering an active renegotiation phase, likely to accelerate as Basel III capital relief materializes and increases bank lending capacity relative to private fund competitors.

Themes by frequency

G-SIBs (8 reports)

Regional Banks (9 reports)

Neo / Digital Banks (6 reports)

Emerging Tech / Fintech (3 reports)

Sources

• BAC (Bank of America) — Press release · Earnings call
• GS (Goldman Sachs) — Press release · Earnings call
• JPM (JPMorgan Chase) — Press release · Earnings call
• C (Citigroup) — Press release · Earnings call
• BK (BNY Mellon) — Press release · Earnings call
• WFC (Wells Fargo) — Press release · Earnings call
• STT (State Street) — Press release · Earnings call
• MS (Morgan Stanley) — Press release · Earnings call
• TFC (Truist Financial) — Press release · Earnings call
• RF (Regions Financial) — Press release · Earnings call
• FITB (Fifth Third Bancorp) — Press release · Earnings call
• USB (US Bancorp) — Press release · Earnings call
• ZION (Zions Bancorp) — Press release · Earnings call
• KEY (KeyCorp) — Press release · Earnings call
• MTB (M&T Bank) — Press release · Earnings call
• HBAN (Huntington Bancshares) — Press release · Earnings call
• PNC (PNC Financial) — Press release · Earnings call
• ALLY (Ally Financial) — Press release · Earnings call
• LOB (Live Oak Bancshares) — Press release · Earnings call
• CASH (Pathward Financial) — Press release · Earnings call
• AX (Axos Financial) — Press release · Earnings call
• CUBI (Customers Bancorp) — Press release · Earnings call
• LC (LendingClub) — Press release · Earnings call
• SOFI (SoFi Technologies) — Press release · Earnings call
• WAL (Western Alliance) — Press release · Earnings call
• XYZ (Block) — Press release · Earnings call
• HOOD (Robinhood) — Press release · Earnings call
• PYPL (PayPal) — Press release · Earnings call
• COIN (Coinbase) — Press release · Earnings call