Upstart Holdings announced first-quarter results from its San Mateo headquarters early on May 5, 2026: originations increased by 61 percent, revenue increased by 44 percent, and 425,356 loans were originated in a single quarter. These figures would make nearly any financial technology company that isn’t Stripe or Klarna jealous. That day, the stock dropped 14.8%. Year over year, the adjusted EBITDA margin decreased from 20 to 13%. The operational loss had increased.
The market concluded that the margin compression was the key after examining both sets of data. This is the Upstart experience distilled into a single trading session: a company generating growth numbers that most businesses would consider extraordinary, closely monitored by about one-third of its float sitting short, and trading with a beta of 2.27, which indicates that the stock moves like a leveraged ETF in either direction when the news changes.
The origination engine and its real actions are where the bull case begins. Unlike the few factors that a FICO score takes into account, Upstart’s AI engine examines more than 1,500 variables every loan application, including employment history, education, cash flow patterns, and spending habits. The business claims—and provides evidence to back up this claim—that their model finds creditworthy borrowers who perform poorly on conventional measures and eliminates hazardous borrowers who perform well. If that is the case at scale and the conversion rate improvement from 14 to 19 percent year over year indicates that it is getting better, the model has a competitive edge over traditional underwriting that doesn’t diminish as it expands; rather, it gets stronger as more loan data is added.
In 2025, the company experienced an 86% increase in originations, made its first return to GAAP profitability in a number of years, and is expecting a compound annual growth rate of 35% through 2028. After Paul Gu took over as CEO in May 2026, Dave Girouard called the 2025 results “a ratio any business would die for” when he revealed that personnel had only increased by 18% compared to 86% rise in origination. It’s worth considering that efficiency claim.
The AI model’s effectiveness is not at issue in the bear case. The financial system is the issue. Upstart functions as a marketplace, originating loans on behalf of credit union and bank partners before selling them to institutional investors who provide funding. The corporation does not, in theory, bear the credit risk; instead, it gets origination fees. Theoretically. In actuality, Upstart ended up retaining loans on its own balance sheet when interest rates spiked in 2022 and institutional buyers withdrew from consumer credit, taking on precisely the credit risk that its marketplace architecture was intended to minimize.
This structural vulnerability is directly addressed by Upstart’s Q1 2026 application for a national bank charter: the company becomes far more resilient to rate cycle disruption if it can finance itself with deposits rather than relying on institutional capital market participants. It takes years to apply for a charter. The reliance on third-party capital is still a significant risk in the meantime.
As of April 2026, the short interest rate was 33.83 percent of float, which is very high for a corporation reporting growth rates. Upstart is delivering right now. High short interest rates can be counterproductive because, if the company keeps operating, shorts will eventually cover, which will put additional pressure on buyers.
However, a third of float being short also indicates a sincere institutional belief that something isn’t working, and because the operating loss increased despite a 44% increase in revenue in Q1 2026, it’s likely the margin trajectory. Year over year, adjusted EBITDA margins decreased from 20 to 13 percent. The next three earnings calls will either confirm or refute the company’s full-year guidance of 21 percent adjusted EBITDA margins, which implies significant improvement over the next three quarters.
Upstart’s ability to completely separate its business cycle from macroeconomic credit circumstances is still up for debate, and the bank charter application implies that management is aware that the existing structure is not the final version. Genuine execution is evident in the 2025 numbers, which show $1 billion in sales, GAAP profitability, and fivefold growth in both vehicle and housing originations.
The credit environment, the speed of the charter application, and whether the company’s predicted margin rebound for the second half of 2026 actually occurs will determine whether execution at this level is sufficient to support a route to double market value. Observing the quarterly reports gives the impression that Upstart is placing a genuinely intriguing wager on AI underwriting, one that isn’t plainly correct or incorrect. This is essentially where intriguing wagers reside for a stock with this level of short interest and volatility.
