The Upstart story: 20x in 10 months, then −97% in the next 14

1. Dave Girouard had been president of Google Enterprise at Google. He left to start his own company and called it Upstart.

The first idea was an odd one. Upstart didn't make ordinary loans. Recent college graduates raised money from backers and, in return, promised them a share of their future income. According to TechCrunch in August 2012, payments were capped at 7% of income, and nothing was owed in years when income was below $30,000.

Silicon Valley liked the idea. The first $1.75 million came from Kleiner Perkins, NEA, Google Ventures, First Round Capital and Mark Cuban. But the "invest in a person's future" model never became a big business.

In May 2014 Upstart pivoted: it began offering ordinary personal loans — first 3-year, then 5-year. One thing survived the pivot, though. When deciding whether to lend to someone, the company didn't look only at the credit score (FICO). It also considered where the person went to school, what they studied and where they worked.

That "different way of looking" is what would later excite Wall Street. But first you need to understand how the model works — because its biggest weakness was hidden there too.

Upstart doesn't call itself a bank. Its model works like this:

• A person applies for a loan on Upstart's site.
• Upstart's AI model assesses the risk and sets the price (the interest rate).
• A partner bank or credit union actually makes the loan.
• Many loans are later sold to large institutional investors.
• Upstart earns a fee on each loan.

The beauty of the model was that Upstart could grow like a software company without putting its own money at risk. In 2017 it became the first company to receive a "no-action letter" from the US Consumer Financial Protection Bureau (CFPB). According to Banking Dive, in the first two years of that period Upstart's model approved 27% more loans than the traditional approach, with average APRs 16% lower.

But this elegant scheme carried one quiet question: who actually puts up the money for the loans? The answer: banks and investors. When they are willing to fund, Upstart thrives. And when they aren't?

At the end of 2020 nobody was asking. Money had never been so cheap.

December 16, 2020. Upstart listed on Nasdaq. The shares were priced at $20 — the low end of the expected range. On the first day they closed at $29.47, up 47%.

Then the race began. On March 18, 2021 the stock jumped 89% in a single day, crossing $100. In August it reached $170. On October 15, 2021 it closed at $390, and intraday it went past $401.

Do the math: IPO price $20 → peak ~$400. Roughly 20x, in just 10 months. Even measured from the first-day close, it was more than 13x. Someone who bought $1,000 of stock at the IPO had about $20,000 ten months later.

This wasn't pure hot air. The numbers really were remarkable. In 2021 Upstart's platform originated 1.3 million loans totaling $11.8 billion — 338% more than the year before. Revenue reached $849 million (+264%) and net income was $135 million (versus just $6 million in 2020). Upstart also moved into auto loans.

But the numbers weren't the only thing lifting the price. The backdrop was perfect: interest rates near zero, "AI" and "fintech" the hottest words on the market, and plenty of money looking for a home. Investors valued Upstart not as a lender, but as a fast-growing technology company.

That's where the first crack appeared. But it was hidden deep in the report — on the balance sheet.

Upstart's promise was "we assess the loans, others fund them." But from late 2021 something else started showing up on the balance sheet.

Loans held on the company's own books were about $78 million at the end of 2020. By the end of 2021 the figure was $252 million, and by the end of March 2022 it was $598 million. In other words, Upstart was keeping part of the loans it had underwritten on its own balance sheet.

Why? Because the wind was shifting. In early 2022 the US Federal Reserve began raising interest rates to fight inflation. When rates rise, two things happen at once:

• Credit gets more expensive for borrowers — fewer loans are made.
• Investors who buy loans become cautious — they demand higher returns or step back altogether.

For Upstart the second one hurt the most. Its entire model depended on others being willing to fund loans. And that willingness had been born in an era of cheap money.

In May 2022 this dependence became visible to everyone.

May 9, 2022. Upstart reported its first-quarter results. The numbers themselves weren't bad: revenue grew 156% to $310 million, and the company posted its seventh consecutive profitable quarter.

But the company cut its full-year revenue forecast from $1.4 billion to $1.25 billion. CFO Sanjay Datta said default rates were "now back to, or in some cases, above pre-pandemic levels."

The next day, May 10, the stock fell 56% and closed at $33.61. In a single day.

In July came another blow. Upstart pre-announced second-quarter revenue of about $228 million instead of $295–305 million. CEO Girouard put it plainly: "Inflation and recession fears have driven interest rates up and put banks and capital markets on cautious footing." The company acknowledged its marketplace had become "funding constrained."

By the end of 2022 the stock had fallen to ~$12. From $390 in October 2021, that's roughly −97%. The $1,000 that had grown to $20,000 at the peak was now worth about $600 — well below the IPO price.

The annual report completed the picture: in 2022 revenue slipped to $842 million, the company posted a $108.7 million net loss, and loans on the balance sheet topped $1 billion. In November 2022 it cut 140 jobs, and in early 2023 another 365 — roughly 20% of staff. One analyst quoted by Banking Dive called reliance on third-party funding "the biggest risk to Upstart."

But the story didn't end there.

Upstart did not go bankrupt. It cut costs and slowly started to recover.

The stock, meanwhile, became a real roller coaster. From ~$12 in May 2023 to above $70 in August — almost 6x in three months. Back to ~$22 in the summer of 2024. Back above $80 at the end of 2024. Each time someone bought thinking "the comeback has started," and someone else lost again.

In 2025 the business delivered: revenue of $1.04 billion (+64%), 1.5 million loans worth $11 billion on the platform, and — after three years of losses — $53.6 million in net income again. The company says it works with more than 100 banks and credit unions.

But notice: even in 2025 the stock never came close to the $390 of 2021 — at its highest it was still roughly four times lower. For an investor who bought in October 2021, the company's recovery did not bring the money back.

Here Upstart closely resembles the Zoom story: the company survived, but the peak price had already priced in the future.

The core lesson of Upstart is simple but painful: a stock can rise 10–20x, and it can give it all back just as fast.

The 20-fold rise in ten months wasn't driven only by the company's success. Much of it came from the backdrop — zero interest rates, cheap money, market enthusiasm. When the backdrop changed, the price left with it.

Upstart's weakness wasn't hidden. It was written in the annual reports, in the list of risk factors: the company depends on banks and capital markets to fund its loans. It's just that in 2021 nobody wanted to read it.

When you look at a stock that has multiplied many times over, ask yourself three questions:

• How much of this rise came from the business, and how much from market mood?
• What outside resource — cheap money, a single customer, a single event — does the company depend on?
• What happens if that resource disappears?

Upstart is a lending-related business. For that reason a halal-minded investor needs to check it separately: see the current status on the UPST page and the criteria in the methodology.