Goldman Sachs Is Handing Apple Card to Chase. The Hard Part Was Never the App

On Friday, August 16, 2019, Goldman Sachs’s board was warned that the system for handling Apple Card disputes was “not fully ready” because of technological issues. That is the wording the Consumer Financial Protection Bureau (CFPB) used later, in an order against the bank. The following Tuesday, August 20, the card launched.

The public record attributes the decision to go ahead to Goldman as an institution. It does not name who made it, and it does not say the system was unusable that day. What it does show is how much dull, hard work sits behind a product that feels effortless on an iPhone: investigating a disputed charge, answering a customer on time, reporting credit data accurately. Someone has to own those jobs, and in a regulated product the law decides who.

On January 7, 2026, Goldman agreed to transfer the Apple Card program, with more than $20 billion in balances, to Chase. The decision closes the most visible chapter of Goldman’s push into consumer finance. It also works as a case study in what executives take on when they build a product with a partner.

Why did Goldman Sachs partner with Apple on a credit card?

Goldman wanted a consumer business that could steady earnings tied to deal markets, and Apple offered customers the bank could not easily reach by itself.

The bank launched its Marcus consumer brand in 2016 and signed the Apple Card agreement in December 2017. David Solomon was then Goldman’s president. He became chief executive in October 2018, which means he inherited the partnership rather than signing it as CEO. He was running the firm when the card launched and when it was later unwound.

At Goldman’s first investor day, in January 2020, executives set five-year targets: consumer deposits from $60 billion to at least $125 billion, and loan and card balances from $7 billion to more than $20 billion. The consumer bank then produced about 2.4 percent of firm revenue, according to Reuters, and Solomon asked investors for patience, saying the firm was planting seeds that would take time to grow. Those were ambitions, not results, and they should be read that way. By the end of 2020, deposits had reached $97 billion, so the funding side of the plan looked well on its way. Lending, where credit risk lives, was the harder half.

Apple Card was the fast route to the lending side. According to the CFPB, Goldman agreed to extend the credit and investigate customer disputes. Apple designed the interfaces customers used to manage their accounts and created the advertising. Mastercard ran the payment network. To a cardholder this was one product. Inside, it was two companies with a handoff in the middle.

The contract also carried an incentive. According to the CFPB’s order, it gave Apple the right to charge Goldman $25 million for each 90-day delay in launch that Goldman caused, and the bureau described that as an incentive to introduce the card early. On a call with reporters, CFPB Director Rohit Chopra went further, saying Goldman made the decision to proceed “since Apple negotiated” the $25 million delay provision. That was the director’s public explanation, not a separate finding in the consent order, and it did not identify an individual decision-maker. Nothing public shows that Apple ever collected the penalty. Goldman has said it worked to resolve the technological and operational challenges that followed the launch. Clauses like this are routine and often reasonable. The harder question is whether anything in the contract protects against the opposite risk, launching too soon.

What did the CFPB find about Apple Card?

On October 23, 2024, the CFPB issued separate orders against Goldman Sachs Bank USA and Apple. The bureau said the breakdowns affected hundreds of thousands of Apple Card users, which put the problem well beyond a handful of unlucky customers.

For Goldman, the bureau found failures in how disputes were handled: acknowledgements and resolution letters that were late or missing, investigations that were not reasonable, and adverse reports to credit bureaus about amounts still in dispute. It also said Goldman misled some customers about how refunds were applied. For Apple, the bureau found that beginning in June 2020, in tens of thousands of instances, disputes started in Apple’s Wallet app never reached Goldman. Both companies were cited over how interest-free installment financing for Apple devices was presented.

Goldman was ordered to pay a $45 million civil penalty and at least $19.8 million in redress. Apple was ordered to pay $25 million. Goldman was also barred from launching a new credit card unless it gave the regulator a compliance plan 90 days ahead. The bank agreed to the order without admitting or denying the findings, and said it had worked to fix technological and operational problems that followed the launch. The bureau’s director at the time, Rohit Chopra, called the program’s execution “a mess.”

The pattern matters more than any single figure. Apple’s form did not pass disputes along. Goldman’s team did not investigate the ones it received to the legal standard. Each company could point to a part that worked, while the customer with money in dispute experienced one failure. In a partnership, the seams between firms are where a product tends to break, and they are also where accountability gets blurry.

Why did Goldman decide to leave?

The business was not earning its place. Platform Solutions, the segment that housed the card, lost money in every year from 2021 through 2024. On Goldman’s comparable segment figures, its pre-tax losses were $1.07 billion in 2021, $2.06 billion in 2022, $2.02 billion in 2023 and $997 million in 2024, about $6.2 billion over four years. The segment also held businesses Goldman has since sold or exited, including GreenSky and the General Motors card, so that total is not an Apple Card loss.

On January 17, 2023, Solomon told analysts that Goldman had “tried to do too much too quickly” in consumer finance. A retreat followed in steps. Goldman booked a $470 million loss on the sale of some Marcus loans in the first quarter of 2023, largely offset by a $440 million release of loan-loss reserves. It then sold GreenSky and moved the GM card to Barclays. Apple Card, the biggest and most public of the partnerships, came last. Reuters reported that Apple talked to other issuers, including Barclays and Synchrony, before JPMorgan emerged as the buyer, and that talks with JPMorgan began in 2024. Replacing a bank partner, in other words, is slow work even when both sides want it.

By January 15, 2025, Solomon was saying in public what many had assumed. The contract with Apple ran until 2030, he told analysts, “although there’s some possibility that it won’t continue until that time frame.” He said Platform Solutions had weighed on the firm’s return on equity by 75 to 100 basis points in 2024, with improvement expected, and, according to a transcript of the call, that the card would not be a long-term business for Goldman.

Set that against the rest of the firm and the logic is easier to follow. In the third quarter of 2026, Goldman reported net revenues of $18.68 billion and an annualized return on equity of 18.7 percent. A drag of 75 to 100 basis points on returns is a real cost for a firm whose strength lies in advice, trading and asset management. That is an interpretation, but it is the one the numbers support.

What does the exit cost Goldman?

When it announced the Chase agreement, Goldman quoted Solomon saying the deal “substantially completes the narrowing of our focus in our consumer business.” The accounting behind it is easy to misread.

Goldman moved the Apple Card loans to held for sale in December 2025. According to its annual report, the portfolio was $21.26 billion, or $19.74 billion after markdowns. In the fourth quarter the firm recorded a $2.48 billion reduction in loan-loss reserves, which sits in the provision for credit losses, and a separate $2.26 billion reduction in net revenues from markdowns and contract termination obligations, plus $38 million of operating expenses. Goldman said the combined effect would add $0.46 to fourth-quarter earnings per share.

These are accounting entries tied to a single event, and they are easy to misread. They are not cash proceeds from a sale, because the transition has not closed. The $2.48 billion release is not a profit of that size, and the $2.26 billion reduction is not a cash loss of that size. Goldman has not published how the $2.26 billion divides between markdowns and termination obligations, so any split would be a guess. JPMorgan said it would record a $2.2 billion provision for credit losses tied to the portfolio; that is its own accounting for its own purchase, not a price.

What the entries do show is that Goldman’s books absorbed the transition in a single quarter, with a small net gain. They do not measure the full cost of the consumer effort, which built up over years in losses, penalties, expenses and management attention. That is analysis, not a figure Goldman has disclosed.

The results since the announcement show a business shrinking in place. In the third quarter of 2026, Platform Solutions revenue was $170 million, 72 percent lower than a year earlier, mainly because of Apple Card markdowns, and the segment lost $76 million. That is one segment. The firm as a whole earned $5.29 billion.

What is Goldman betting on? That capital and attention freed from a low-return consumer business will earn more in the franchises where it already has an edge. Management has said as much: Solomon pointed to Global Banking & Markets and Asset & Wealth Management as the focus. The bet carries its own risk. Goldman gave up a direct card relationship with millions of households, and it will not rebuild one quickly if the strategy changes again.

The exit is also unfinished. The transition is expected to take about 24 months, and until it is complete Goldman has said it will keep operating the program and recording its ongoing business results. Cardholders continue to use the card on the Mastercard network in the meantime. A handover on that scale is an operational exercise of the kind that produced the CFPB findings in the first place. Nor is Goldman leaving consumer banking altogether: it still raises deposits through Marcus, which its annual report places within Asset & Wealth Management.

What can executives learn from the Apple Card partnership?

The first lesson is about ownership. A clean interface makes promises about refunds, disputes and credit reporting, and in a regulated product somebody holds the license and the liability. Apple shaped the experience, while the CFPB’s order assigned much of what happened behind it to Goldman. The brand owner and the licensed operator can carry very different exposure, which is why a contract needs to say plainly who answers for what.

Incentives deserve the same scrutiny. If a contract rewards hitting a date, someone independent should hold real authority to say not yet, and saying it should carry consequences of its own. A third-party readiness warning is worth only what the organization does after receiving it, and Goldman’s board had four days. The CFPB’s account of the launch sets the delay clause next to the board warning, and leaders should assume that a clause written for speed will one day be examined for what it did to judgment.

The handoffs matter as much as the parts. The dispute failures sat at the junction between two companies’ systems, and the CFPB said Apple’s forwarding problem began in June 2020, long after launch. Readiness, on that evidence, is a running condition rather than a launch-day checklist: the path a customer complaint travels needs end-to-end testing and then continuous monitoring.

Finally, the exit belongs in the original negotiation. Goldman’s contract ran until 2030, and Reuters reported that talks with JPMorgan began in 2024. Leaving took years, and on Goldman’s own accounting it produced a $2.26 billion reduction in revenue alongside reserves the firm released. Terms covering servicing during a transition, data transfer and who bears credit losses decide whether a clean break is possible.

A product can succeed while the operation behind it struggles

Apple Card did win customers. Apple said roughly 12 million people held it as of early 2024, and Goldman called it an award-winning program. That is what makes the case useful. A product can earn loyalty and still run on an operation that falls short of what regulators, and customers, expect.

The next test belongs to Chase and Apple, who now have to deliver the experience on a bank’s infrastructure, and to Goldman, which has to hand over millions of accounts cleanly while still responsible for them. Whether the transition goes smoothly will say more about the lessons of this partnership than any announcement.