Goldman Sachs Didn't Collapse. It Tried to Become a Normal Bank — and Bled $3.8 Billion Finding Out Why It Shouldn't.

In January 2026, Goldman Sachs handed the Apple Card — the splashiest consumer product it ever built, a roughly $20 billion portfolio of credit it had partnered with the most valuable company on earth to issue — over to JPMorgan Chase, and the move was framed not as a loss but as a relief: a projected $0.46-per-share gain in the quarter from releasing reserves it no longer needed.⁷ That is what a strategic surrender looks like when you do it carefully. A bank that supposedly 'collapsed' was, in the same breath, booking a tidy accounting win on the way out the door.

The story you've seen is that Goldman Sachs collapsed. It didn't. The firm posted full-year 2024 net revenues of $53.51 billion, net earnings of $14.28 billion, and diluted EPS up 77% to $40.54.¹ There is no insolvency, no takeover, no smoking crater. The collapse is a myth. But underneath the myth is a real fall — slower, quieter, and far more instructive than a crash.

The thing that actually died had a friendly first name

Here is the thesis, plainly: Goldman didn't collapse — it spent the better part of a decade trying to stop being Goldman, lost roughly $3.8 billion learning that it couldn't, and retreated. The casualty wasn't the firm. It was Marcus — the consumer brand named, with calculated warmth, after one of the firm's own founders, built to lend to ordinary people the way the white-shoe investment bank never had. Launched in 2016, before David Solomon ran the place, with Harvey Schwartz — the man who would lose the CEO race to Solomon — among its original backers.⁴ Solomon inherited the dream, championed it, and then was forced to bury it.

The mechanism of the failure is the interesting part, because the losses didn't come from where the brand name pointed. Marcus savings accounts and CDs — the gentle, deposit-gathering business — were not the bleeding wound. The damage concentrated in the segment Goldman called Platform Solutions: the Apple Card partnership and the GreenSky point-of-sale lender it had bought. That segment lost more than $3 billion from the start of 2020, per an SEC filing — $783 million in 2020, $1.05 billion in 2021, and $1.21 billion in just the first nine months of 2022.² Add the rest of the consumer push and the combined pre-tax losses from 2020 to 2023 reached roughly $3.8 billion.³ A firm built to advise on billion-dollar deals had wandered into a business where the unit economics are won or lost a few hundred dollars of credit-card delinquency at a time — and it was losing them.

Why a wholesale bank can't simply put on a retail coat

The deeper why is structural, not managerial. Goldman's entire machine — its talent, its incentives, its risk culture — was tuned for transactions that are large, infrequent, and negotiated. Consumer credit is the inverse: tiny, constant, and statistical. You don't out-negotiate a credit-card book; you grind it down with underwriting models, servicing infrastructure, and the patience to lose money for years while the franchise compounds. Goldman had the capital to fund that patience, but not the muscle memory — and the cost of building the muscle showed up as red ink. Worse, the consumer business invited a kind of attention Goldman's deal-making had always operated above. By 2022 the Federal Reserve had opened a review of the money-losing Marcus unit, and the CFPB was investigating the firm's credit-card account-management practices.⁵ A wholesale bank answers to sophisticated counterparties. A retail bank answers to regulators who exist to protect the unsophisticated — and that scrutiny is not a tax you can pay your way out of.

A retreat conducted like a deal, not a rout

What distinguishes a fall from a collapse is whether you choose the exits or have them chosen for you, and Goldman chose. The dismantling was phased and deliberate, sold off piece by piece to buyers who wanted the parts. GreenSky went to a consortium led by the private-equity firm Sixth Street in October 2023. Marcus Invest went to Betterment in 2024. The GM credit-card business went to Barclays in the third quarter of 2025.⁶ And the Apple Card, the crown jewel of the whole misadventure, is moving to JPMorgan — a transfer announced in January 2026 and expected to take about two years to complete.⁷ Each sale converted a strategic embarrassment into a transaction, the one thing Goldman has never forgotten how to do.

But wasn't Goldman one missed rescue from a real collapse?

The honest objection cuts the other way: maybe 'Goldman never collapses' is itself the myth. In September 2008, the firm converted from an investment bank to a bank holding company on the 21st, and two days later Berkshire Hathaway put $5 billion of preferred stock into it.⁸ Goldman's official line has always been that it was never truly at risk. But the advocacy group Better Markets, citing an internal New York Fed email, argues Goldman was in genuine danger of failure and was rescued by government action.⁸ If that reading is right, the firm's reputation for invincibility is partly a survivorship story it got to narrate afterward. That's the fair steelman — and it sharpens rather than softens the point. A bank that may have been rescued once already had every reason to diversify away from the volatile wholesale business that nearly killed it. The consumer push wasn't hubris for its own sake. It was an attempt to buy the stable, boring deposit base that bank holding companies are supposed to have. The strategy was logical. The execution ran into a wall that capital alone couldn't climb.

Goldman Sachs spent the better part of a decade trying to befriend the everyman — gave the project a warm first name, partnered with Apple, bought a lender, gathered deposits — and discovered that the skills that make you formidable in the boardroom make you ordinary at the kitchen table. It absorbed roughly $3.8 billion in tuition and walked away, selling each piece to someone who wanted it. The headlines called it a collapse. It was the opposite: a firm wealthy enough to be wrong out loud, in public, for years, and still post a record. The fall was real. It just wasn't the firm that fell — it was the fantasy that Goldman could be anything other than Goldman.