Klarna

Executive Summary

Klarna's product is a button. Its business is a loan book that turns over about ten times a year. In the first quarter of 2026 the company originated $33.7 billion of purchases through an average consumer balance of $124, held for an average of 39 days, and lost 0.55 percent of what it originated [1]. Nearly everything that has ever happened to Klarna, good and catastrophic, follows from that one structural fact.

For fourteen straight years from its founding in 2005 the company was profitable, because it grew inside markets whose repayment behaviour it could already predict [2][3]. Then it entered markets it could not predict, and the arithmetic reversed. Losses arrived in 2019, deepened every year to a 2021 net loss of SEK 7.1 billion (about $825 million), and in July 2022 the private market cut the company's value from $45.6 billion to $6.7 billion in twelve months [4][5][6]. The same management team then cut costs, tightened underwriting, and delivered a profitable year in 2024 on lower headcount than 2021 [7][8].

Today Klarna is listed on the New York Stock Exchange, worth about $7.5 billion, roughly half its own IPO price and a sixth of its private peak [9]. It has applied for a US banking licence, and it is deliberately lengthening the duration of its loan book for the first time in twenty years [10][1]. Every era of this case asks the same question in a new costume. When you carry someone else's credit risk in exchange for volume, how fast are you allowed to grow?

---

The Cloakroom Deal, 2004-2008

In December 2004, at a Christmas mixer in Stockholm, a 23-year-old with no company and no product cornered one of the few people in Sweden who wrote angel cheques. Jane Walerud had sold the Erlang software firm Bluetail to Alteon four years earlier and was investing the proceeds. Sebastian Siemiatkowski described an idea for letting Swedes buy online and pay after the parcel arrived. She agreed to put in SEK 600,000, roughly $80,000 at the time, for 10 percent of a company that did not yet exist. In her telling the deal was settled where they were standing, by the coats [11][12]. She also introduced the founders to a team of programmers who built the first platform and took shares instead of salary [13].

That was the second-best thing that happened to the idea in 2004. The best was the rejection. The founders' account, repeated in every retelling since, is that they pitched at a Stockholm School of Economics event in front of a jury of Swedish business grandees, were told the banks would simply do this themselves, and placed last. No contemporaneous record of the evening survives, and one Swedish account has the same company winning a different school competition that season, so the story should be read as founder testimony rather than fact [12][11]. Siemiatkowski's own gloss on it, given to Sequoia Capital years later, is the part that matters.

My reaction wasn't to give up and go home, it was to curl up my fist and say "I'll show you."

Sebastian Siemiatkowski, recounting the pitch competition, Sequoia Capital profile [14]

He was 23, had returned from travelling with no job, and had spent part of 2004 on Swedish social assistance [11]. He had met his co-founder Niklas Adalberth on a Burger King production line, an origin he has told repeatedly and which no employment record has ever confirmed [15]. The third founder, Victor Jacobsson, would leave the company in 2012 and remain one of its largest shareholders, which becomes important two decades later. The holding company was registered in February 2005 under the name Kreditor Europe AB, and the first transaction ran that April at a Stockholm bookshop called Pocketklubben [16].

Sweden made the idea plausible in a way it would not have been elsewhere. Swedish mail-order retail had trained a generation to receive goods first and settle an invoice afterwards, so the behaviour Klarna needed already existed. What did not exist was a business willing to stand between the shopper and the shop.

Here the founders faced the only decision in this case that determines all the others. There were two ways to build the product. The light version was a technical one, routing the shopper to an invoice the merchant still owned, collecting a small fee for the plumbing and carrying no exposure. The heavy version was to buy the receivable outright, pay the merchant on dispatch, and take the loss when the shopper did not pay. The light version needed engineers. The heavy version needed capital, a credit model, and a tolerance for being wrong about strangers.

Nothing in the record shows the three of them debating this in a room, so the alternatives here are reconstructed from what a Swedish payments builder could have chosen in 2005 rather than from a documented argument. What can be verified is that the merchants had a real and expensive problem, which was that shoppers abandoned checkouts rather than hand card details to an unfamiliar site, and that whoever solved it would be paid out of the merchant's conversion rate.

---