The Big Idea
Most lending businesses follow a simple rule, your past defines your future. Credit scores, repayment history, and income stability decide whether you get a loan.
Upstart Holdings is trying to change that.
Instead of relying on traditional credit scoring models, Upstart uses artificial intelligence to assess borrowers. The pitch is straightforward: by analyzing more data points, AI can better predict risk, approve more borrowers, and reduce defaults at the same time. It’s a bold idea. And for a while, the market loved it.
From Boom to Reality Check
During the low interest-rate era, Upstart was one of the hottest fintech stocks. Loan volumes surged, revenue skyrocketed, and the company positioned itself as the future of credit underwriting. But then the cycle turned.
As interest rates rose sharply, lending slowed across the economy. Banks became cautious, funding dried up, and Upstart’s model faced its first real stress test. Loan originations dropped, and the company was forced to hold more loans on its own balance sheet something it ideally wanted to avoid.
This shift exposed a key vulnerability: Upstart doesn’t control demand. It depends heavily on lending partners and broader macro conditions.
Signs of a Comeback
Now, the latest results suggest that things may be stabilizing.
Upstart recently reported quarterly revenue of around $135 million, showing sequential improvement after a difficult period. More importantly, loan originations have started picking up again, indicating that lending partners are slowly returning.
Another positive sign is that the company is reducing the amount of loans it holds on its balance sheet. This matters because holding loans increases risk and ties up capital something Upstart wants to minimize as it scales its platform model.
In simple terms, the engine is starting to restart.
The AI Advantage, Still Unproven
At the core of Upstart’s story is its AI model.
The company claims that its models can approve more borrowers at lower default rates compared to traditional methods. If true, this is a massive advantage. It could expand access to credit while improving efficiency for lenders.
But here’s the catch.
This claim has not been fully proven across multiple credit cycles. The recent downturn raised questions about how well the models perform under stress. And in lending, consistency matters more than innovation. Banks don’t just want better models. They want reliable ones.
A Business Model in Transition
Upstart is also evolving how it operates.
Initially, the company positioned itself as a pure platform, connecting borrowers with lenders without taking on credit risk. But during the downturn, it had to step in and hold loans temporarily, blurring that model.
Now, it’s trying to return to its original structure by:
- Partnering with more funding sources
- Reducing balance sheet exposure
- Expanding into new verticals like auto loans and home equity
If successful, this could diversify revenue streams and make the business more resilient.
The Bigger Risk
Despite the improving numbers, risks remain significant.
Upstart is highly sensitive to interest rates and economic conditions. If rates stay elevated or credit demand weakens again, growth could stall. At the same time, competition from traditional banks and fintech players continues to intensify.
And unlike software companies, this is not a high-margin, asset-light business in practice, at least not yet.
The Bottom Line
Upstart is one of the more interesting fintech experiments in the market today.
It combines AI, lending, and platform economics into a single story. The potential is large, but so is the uncertainty. The company is still proving whether its model can work consistently across cycles.