Upstart Just Put Up Its Best Quarter Ever — and the Stock Is Still Down 56%. One Proof Point Is Missing

Upstart Just Put Up Its Best Quarter Ever — and the Stock Is Still Down 56%. One Proof Point Is Missing

Upstart last month reported what its own investors have begun calling its best quarter since the story turned: record loan volume of $4.2 billion, revenue up 42% year over year to $365 million, and $16.5 million of GAAP net income, up 195% from the same quarter a year earlier. Shares jumped more than 10% the day the numbers landed. Read in isolation, that is a company that has pulled off the unthinkable — profitable again, growing again, and doing both at once.

Now read the price. Over the past year the stock is still down roughly 56%, from a 52-week high above $76 to about $30 today, near the bottom of its range. Roughly a third of the available shares are sold short. This is what a market pricing a broken company looks like, and it is sitting on top of what the operating statements say is a working company. That gap — between a record quarter and a 56% decline — is the whole story.

The most recent thing investors are told to worry about is an insider disposing of shares. In June, an insider sold 15,000 shares at $30.41, proceeds of about $456,000, per an SEC filing disclosed on June 11. Set that against what the people closest to the business actually did in 2026. The CEO spent roughly $5 million buying shares this spring rather than selling them, and insiders at last count held about 29 million shares — roughly 30% of the company. Insiders sell for any number of reasons — taxes, diversification, estate planning. They tend to buy for one reason. A single $456,000 disposal tells you nothing; the pattern around it tells you more.

The market is still pricing the old Upstart

To understand the price, you have to reconstruct the story the crowd is still trading. This is the 2022 Upstart: the AI lender that looked brilliant while credit was cheap and nearly broke when the funding market froze. Rates shot up, the market for securitizing loans closed nearly overnight, guidance got pulled, and a stock that had traded far into triple digits fell more than 90%. Once a crowd lives through that, it does not easily go back. It reclassifies the company as a leveraged, cyclical lending business with a fashionable AI wrapper, and it stops giving the income statement credit for anything.

You can see that reclassification in the multiple. A company growing revenue at 42% while generating real GAAP profit trades at about 2.4 times trailing sales — roughly 3.6 times on an enterprise basis once the balance sheet's debt is added back in. Those are the kinds of multiples you pay for a mature lending company, not a 40%-plus grower. The valuation is not arguing with the business; it is arguing with the last credit cycle.

The income statement has already crossed back

What is actually underneath? The engine is personal loans. In the second quarter, unsecured lending produced $3.64 billion of originations, up 46%, at a contribution margin that held steady around 62%. Contribution profit — the revenue left after the direct cost of originating a loan, before the company's overhead — hit an all-time high of $193 million in the quarter, more than half of total revenue. Upstart's pitch, that 91% of its loans are now processed end to end with no human involvement, is the moat claim: the more data it collects, the better its models price risk, and the cheaper its cost of lending gets.

The genuinely interesting part is the trajectory of the parts that still lose money. The newer products — auto, small business, and home equity — are being carried at a negative contribution margin, but the losses have narrowed from minus 176% a year ago to minus 35% now. That is a pair of money-losing machines pointed at breakeven, improving for two consecutive quarters. Management guides to roughly $1.4 billion of revenue and about $294 million of adjusted EBITDA for 2026. Adjusted EBITDA is a profitability measure that strips out stock-based compensation and other non-cash charges; that guidance implies an adjusted margin in the low 20s. None of this is the 2022 story. It is a company compounding revenue at 40%-plus while its two drags inch toward profitability.

Be fair about the other side of the ledger: the profit is real but thin, and it has been lumpy. Per-share profit in the first quarter of 2026 was roughly double the second quarter's, because the company is consciously spending to scale those newer products instead of banking every dollar of the growth. That is a bet on future margin, not current margin. Directionally it supports the turnaround; it does not yet prove the final shape.

The proof point that's still missing

Here is the part that has to stay honest. The one number this company has not yet delivered is free cash flow — cash left over after paying for the growth and the interest. Over the trailing year, free cash flow was about negative $305 million. Operating cash flow consumed roughly $270 million in the first half of 2026 alone. Cash on hand is about $456 million against $2.37 billion of debt, leaving about $1.55 billion of net debt and a debt-to-equity ratio around 2.5.

There is a legitimate mechanical explanation, and it matters that you understand it. When Upstart originates a loan and keeps it on its own books briefly before selling it or bundling it into securities, the cash goes out the door immediately while the revenue shows up later. The loans it holds are financed through warehouse credit lines — short-term borrowing against the loan portfolio. So rising originations create a cash drag by construction. That is why the income statement can look excellent while the cash-flow statement looks alarming: growth itself is consuming cash.

But "explained by the model" is not the same as "proved by the model." The company says it is funding this growth without adding equity capital — which is a polite way of saying it is funding growth with borrowed money. And that funding mechanism is exactly what broke the old story in 2022: when the funding market froze, the growth machine stalled. The income statement has now proven the underwriting is back. The cash-flow statement has not yet proven the funding model is durable. That is the one bridge still under construction, and it is the reason I will not dress this up as a finished thesis.

What would prove it, and what would break it

Some nuance on the crowd, because it is not as clean as the "everyone gave up" setup. The price has capitulated — down 56% with shorts parked at a third of the float. But the ratings tape has not. AInvest's aggregate signal still labels the stock a Buy, with a composite score around 5.75 out of 10, and as of midyear the average sell-side price target clustered near $40, well above the current quote. So the crowd is divided, not surrendered: shorts pressing the old disaster story on one side, analysts still carrying the old growth story on the other, and a price stuck between them while the company prints records. The division is the opportunity. It is also why the setup is not clean yet.

Here is the bridge that would make this thesis concrete. Over the next two quarters: full-year 2026 lands near the guided $1.4 billion of revenue and $294 million of adjusted EBITDA; originations keep compounding; the secured products cross toward breakeven. Then the decisive one — free cash flow turns positive on a trailing-12-month basis, or at the very least net debt stops climbing while growth continues. If this business can compound at 40%-plus and fund that growth with its own cash, then the current multiple is pricing the old, broken lender, and the gap between the price and the operation is enormous. The 12 months ahead are improving; the market, on the price and on the short book, is still betting they are not.

And the tripwire, stated plainly so it can be checked. If net debt keeps climbing toward $2 billion while free cash flow stays deeply negative into 2027, then this growth is leverage wearing a growth costume, not compounding — and the crowd shorting a third of the float will have been right all along. A funding-market freeze would break it the same way and faster. Those are the two things that should make you care less about a $456,000 insider sale and more about the direction of the balance sheet.

Watch the next two quarters as the proof window. The stock's collapse and the token insider disposal are the old story's noise; the record quarter is the new story's signal. But a signal is not proof. Until the cash-flow statement confirms what the income statement claims, Upstart is a setup to track with the bridge and the tripwire named above — not a price target waiting to be minted from hope.