Capital One Reports $3 Billion Second-Quarter Profit as Credit Losses Fall

Capital One's Rebound Quarter

Capital One Financial said on July 21, 2026, that it earned $3.0 billion in the second quarter, or $4.73 per diluted share^[1][3]. That is a sharp turnaround from a year earlier, when the bank posted a $4.3 billion net loss^[1][3][11]. Back then, a one-time accounting charge tied to buying Discover had dragged results down^[11].

On an adjusted basis, which strips out one-time items, Capital One earned $5.81 per share^[1][3]. Wall Street analysts had expected about $4.69, so the bank beat that estimate by a wide margin^[3][4]. The news landed about 14 months after Capital One finished buying Discover Financial Services in May 2025^[3][11].

A big driver of the higher profit was a smaller "provision for credit losses." That is money banks set aside now to cover loans they expect to go bad later^[1]. This quarter, the provision fell about $1.1 billion from the prior quarter, to $3.0 billion^[1]. Part of that drop came from a $662 million "reserve release," meaning the bank decided it no longer needed some money it had earlier stockpiled for losses^[1][6].

What Both Sides Agree On

Some numbers here are not in dispute. Card loans grew to $275.4 billion and auto loans to $89.3 billion^[9]. Total revenue rose 4% from the first quarter, to $15.9 billion^[1][4].

Borrowers also paid a bit more reliably. The share of card loans the bank wrote off as losses fell 22 basis points from the prior quarter, to 3.23% overall^[9]. The domestic card charge-off rate was 4.71%, down 54 basis points from a year earlier^[9]. Late payments improved too: the share of accounts more than 30 days delinquent fell to 3.39%^[9].

Net interest margin, a core measure of how profitable a bank's lending is, rose to 8.01%^[1][4]. It measures the gap between what a bank earns on loans and what it pays out on deposits. It is separate from the reserve accounting that boosted this quarter's profit, so both the company and bullish analysts point to it as evidence the improvement is not just bookkeeping^[1][4].

And everyone agrees on the deal that reshaped the company. Capital One closed its purchase of Discover on May 18, 2025, becoming the largest U.S. credit card issuer and, notably, the only major one that also owns its own payment network^[11][15]. That last part matters: most card issuers rely on Visa or Mastercard to process transactions and pay them a fee for it. Capital One now owns that plumbing itself through Discover's network, so it can keep those fees rather than pay them out^[1][7].

The Reserve Lever

Understanding the skeptics' concern requires understanding how loan-loss reserves work. Banks are required to estimate, quarter by quarter, how much of their current loan book will eventually go unpaid. That estimate is a judgment call, not a hard fact^[1][6].

When a bank thinks losses will rise, it "builds" reserves, setting aside more money now, which lowers reported profit. When a bank thinks losses will fall, it can "release" reserves, and that money flows straight into profit even though no new customer paid a cent^[1][6]. This gives management real, and completely legal, discretion over how a given quarter's earnings look.

That is exactly what happened here. The $662 million reserve release was a management judgment that future losses will be lower, and it padded the headline profit number^[1][6]. Skeptics are not accusing anyone of wrongdoing. They are simply asking whether a good quarter built partly on a shifted estimate reflects the same kind of strength as one built purely on new revenue^[6].

Bulls, Skeptics, and the Company's Case

Capital One's management frames the quarter as proof that two things are working at once: a resilient U.S. consumer and a Discover deal that is paying off^[1][3]. CEO Richard Fairbank said consumers and the economy "remain resilient^[3]." The company points to falling charge-offs, falling delinquencies, and an integration on track to deliver $2.5 billion in cost savings by 2027^[1].

Capital One also argues the merger helps merchants, not just the bank, because Discover's network has historically charged merchants lower fees than Visa or Mastercard^[1][7]. The company has an obvious interest in this framing: it paid roughly $50 billion for Discover, and its stock has fallen sharply in 2026, so it needs quarters like this one to demonstrate the deal was worth it^[3][14].

Cautious analysts do not dispute the beat itself, but question what it is made of. They note that the reserve release adds to profit without new cash coming in, and that Capital One's stock did not rally on the news, with some reports showing shares dipping below $210^[6][13]. To them, a 4.71% domestic card charge-off rate, while improved, is still high by historical standards, so calling the consumer fully "resilient" may be premature^[9].

Consumer advocates, including Senator Elizabeth Warren, focus less on this quarter's numbers and more on what the merger structurally created. They argue the combined firm now dominates lending to "nonprime" borrowers, people with credit scores below 660 who have few other places to turn^[8]. Warren has said Discover previously offered such borrowers rates about two percentage points lower than Capital One did, an edge she expects to erode now that the two are one company^[10]. Before the deal closed, she and other lawmakers estimated the combined firm would control more than 30% of the nonprime credit card market, and calculated the merger would push a standard antitrust concentration measure up by roughly 400 points in that segment, a jump regulators typically treat as a warning sign^[16]. Merchants, meanwhile, are watching interchange fees, the charge collected each time a card is swiped. A bigger issuer that also owns a network has more leverage over those fees, though Capital One counters that Discover's fees have historically run lower than rivals'^[1][7].

How the Coverage Split

Financial outlets covered the earnings largely as a straightforward beat. Bloomberg's headline named the driver directly: "Capital One Profit Beats Estimates as Loan-Loss Provisions Drop^[5]." Investing.com leaned more bullish, framing it as a "strong beat" with "credit improves, NIM expands^[4]."

MarketScreener split the difference in a single headline, calling it a quarter that "smashes expectations, but the drop in provisions raises questions^[6]." Quartz stayed descriptive, simply noting the beat and its cause without editorializing on the merger.

Progressive outlets took a different angle entirely. Common Dreams centered its coverage on Warren's push to have the Federal Reserve reconsider its approval of the deal, giving little space to the company's side^[10]. Trade press like Payments Dive covered merger opponents challenging the deal's approval, while still carrying Capital One's rebuttal on merchant fees^[8].

The Bigger Picture

This quarter's profit is not really the end of a story. It is another data point in a fight that started before the merger closed and shows no sign of ending. Capital One's structural position, the largest card issuer, owning its own network, concentrated in nonprime lending, was set the moment the Discover deal closed in May 2025^[11][15].

Every quarter from here forward will be read through that lens. A strong quarter looks, to the company and to bullish analysts, like proof the deal is working as promised^[1][3][4]. The same strong quarter looks, to critics and to Warren, like evidence the combined firm has room to raise fees and squeeze vulnerable borrowers precisely because there is less competition to stop it^[8][10]. Both readings can draw on the same set of facts, which is why the numbers alone are unlikely to settle the argument.