Capital One’s EPS Rebounded to $4.73. Credit Did the Work

Capital One’s Q2 FY2026 results marked a clear recovery in reported profitability. Revenue reached $15.85 billion, up 26.9% from $12.49 billion in Q2 FY2025 and 4.0% sequentially. Reported EPS improved to $4.73 from $3.34 in Q1 FY2026 and a loss of $8.58 a year earlier. The company also reported adjusted EPS of $5.81, excluding acquisition amortization, Discover integration expenses and Brex integration expenses, according to the Capital One Q2 FY2026 earnings release.

The improvement was not solely a revenue story. Provision for credit losses declined by $1.1 billion sequentially to $3.0 billion and included a $662 million loan-reserve release. That reduction more than offset the gap between $3.6 billion of net charge-offs and the reported provision. Credit therefore contributed materially to the EPS rebound, and the reserve release should be separated from recurring operating earnings.

Underlying spread economics also improved. Net interest margin rose 14 basis points to 8.01%, while pre-provision earnings increased 1% to $6.8 billion. Period-end loans held for investment grew $9.4 billion, or 2%, to $457.2 billion. Credit card loans increased 2%, auto loans increased 4%, and commercial loans increased 1%. These figures show broad, if uneven, balance-sheet growth rather than dependence on a single lending category.

Costs were the main offset. Non-interest expense increased 7% sequentially to $9.0 billion, including an 11% increase in marketing and a 6% increase in operating expenses. The quarter contained $494 million of acquisition amortization, $298 million of Discover integration expense and $96 million of Brex integration expense. Capital One’s reported efficiency ratio was 57.05%, compared with an adjusted ratio of 51.38%.

The result was a 24.1% pre-tax margin, up from 17.7% in Q1 FY2026 and negative 47.5% in Q2 FY2025, based on the supplied quarterly financial table and the Capital One Q2 FY2026 Form 10-Q. Gross margin and free cash flow are not useful measures for this business because deposits and lending assets are operating inputs rather than financing and working-capital items in the industrial-company sense.

The quarter strengthens the case that Capital One’s post-transaction earnings power is emerging, but it does not yet establish a clean recurring run rate. Revenue growth, a higher net interest margin and expanding loan balances are favorable because they demonstrate that the larger franchise can generate more income before credit costs. Management said the results reflected “solid top line growth and strong credit performance” and that the Discover integration was going well 14 months into the process.

The distinction between reported and adjusted EPS remains important. The $1.08 difference between reported EPS of $4.73 and adjusted EPS of $5.81 reflects costs that management excludes, but acquisition amortization and integration spending are still real claims on current-period earnings. Some integration expense should eventually diminish, while acquisition amortization may persist longer. Capitalizing the adjusted result without allowing for either item would overstate normalized earnings.

Credit is the more consequential uncertainty. A $662 million reserve release helped the current quarter, whereas reserve additions would reduce earnings in a weakening consumer environment. Net charge-offs remained $3.6 billion, so the quarter did not show an absence of credit costs; it showed that the provision declined relative to the prior quarter. The bullish interpretation is that reserves are adequate and performance is holding. The bearish interpretation is that current EPS benefited from a credit adjustment that cannot recur indefinitely.

Capital provides a meaningful counterweight. The preliminary Common Equity Tier 1 ratio was 13.7% at June 30, 2026. That level indicates a capital buffer while the company integrates Discover and carries $457.2 billion of loans held for investment. Period-end deposits declined 1% to $484.3 billion, but average deposits increased 1% to $486.8 billion, and the rate paid on interest-bearing deposits fell nine basis points to 2.91%. These movements support funding economics, though deposit stability remains central to the thesis.

Fair value remains $225 per share. The quarter was better, but not sufficiently clean to justify changing the prior valuation. The shares trade at 12.4 times supplied trailing EPS of $16.70. Our bank valuation uses normalized EPS rather than free cash flow or EV/EBITDA because those industrial-company measures do not describe a deposit-funded lender appropriately.

The bear case is $160, based on assumed normalized EPS of $16.00 and a 10.0 times multiple. It represents weaker consumer credit, persistent integration costs and limited operating leverage. The base case is $225, calculated from assumed normalized EPS of $18.75 and a 12.0 times multiple. It assumes integration costs moderate, revenue growth converts into earnings and credit losses remain manageable without relying on repeated reserve releases. The bull case is $273, based on $21.00 of normalized EPS and a 13.0 times multiple, reflecting successful integration, durable net interest margin and stronger efficiency.

We assign probabilities of 25% to the bear case, 50% to the base case and 25% to the bull case. The base-case value is used as the published fair value, consistent with the stated methodology. At the current price of $207.11, $225 represents 8.6% upside. That falls short of the 15% threshold required for an Undervalued verdict, leaving Capital One Fairly Valued.

The assumptions are intentionally below a simple annualization of adjusted Q2 EPS. Annualizing $5.81 would ignore the quarterly variability of provisions, reserve releases and integration costs. The valuation instead requires evidence that improved revenue and margins can persist through a normalized credit cycle.

A sharper deterioration in consumer credit is the clearest threat. Capital One has substantial exposure to credit cards and auto lending, where losses can react quickly to unemployment, borrower stress and weaker used-vehicle economics. If net charge-offs rise faster than revenue and reserves must be rebuilt, normalized EPS could fall toward or below the bear-case assumption.

Integration execution is the second risk. Management said Discover integration was going well, but the quarter still included $298 million of Discover integration expense and $494 million of acquisition amortization. The thesis would weaken if costs remain elevated without corresponding improvements in efficiency, payment-network economics or customer growth.

Expense discipline also needs to improve. Non-interest expense grew 7% sequentially while pre-provision earnings advanced only 1%. Marketing can support future growth, but sustained expense growth above revenue growth would prevent the enlarged platform from producing the anticipated operating leverage.

Finally, funding and capital could become constraints. Period-end deposits fell $4.8 billion during the quarter even though average deposits increased. A sustained deposit decline, higher funding costs or a lower regulatory capital ratio could limit loan growth and capital returns. The next confirmation point is not another adjusted EPS figure alone; it is evidence that provision trends, expenses and integration costs are moving toward a durable post-acquisition earnings base.