Upstart’s AI just imploded spectacularly.
Investors aren’t buying the excuses. In a fresh class-action lawsuit filed this week, they’re hammering Upstart’s executives — CEO Dave Girouard chief among them — for what they call blatant misleading hype around the company’s next-gen AI lending model. Picture this: back in February, Upstart touts this bad boy, says it’s crushing it, jacks up full-year revenue guidance by 18% to $1.01 billion. Stock pops. Everyone’s cheering. Fast-forward a couple months, May rolls around, and bam — guidance slashed to $850-900 million. Why? Their own words: the AI was “overresponsive” to economic shifts, churning out too many loans when rates were dipping, then slamming the brakes too hard as conditions tightened.
Investors accused fintech executives of misleading them for adjusting revenue projections upward, debuting “overresponsive” AI, then needing to forecast downward.
That’s the core beef, straight from the complaint. And here’s the data driving it home: Upstart’s stock, post-hype, cratered 17% after the May cut. Year-to-date? Down over 40%. Market cap’s shriveled from peaks above $12 billion in ‘21 to under $3 billion now. Ouch.
Upstart’s Wild Ride: From IPO Darling to AI Headache
Remember 2021? Upstart IPO’d at $20, rocketed to $400 on AI-lending magic. Banks were dinosaurs; Upstart’s models promised to price credit 30% more accurately using 1,600 variables — education, job history, you name it. Revenue exploded 260% that year. But rates hiked in ‘22, lending dried up, stock tanked 95%. They clawed back some ground with cost cuts, new model teases. Enter 2024’s drama.
This “overresponsive” label? It’s code for a model too twitchy — hypersensitive to macro noise like Fed signals or job data. In Q1, it approved loans left and right amid soft landings chatter. Then pivot: submissions dropped 5%, conversion rates tanked. Upstart admits it now, but investors say execs knew risks during the upbeat calls. Girouard on the Feb earnings: “Our new model is performing even better than expected.” Cue the pitchforks.
One short para: Brutal timing for Upstart.
Why Are Investors Calling BS on Upstart’s AI?
Look, AI in lending isn’t new for Upstart — it’s their whole schtick. But this reeks of overpromise. Earnings transcripts show execs doubling down pre-cut: “Significant step change in model performance.” No caveats on volatility. Now, lawyers argue securities fraud under the ‘34 Act — material misstatements inflating stock price, letting insiders cash out (options exercises timed suspiciously). Data point: insider sales hit $10 million in Q1 alone.
And the market? Fintech lending peers like SoFi, LendingClub — they’re guiding conservatively, eating steady origination growth without the whiplash. Upstart’s personal loan originations? Q1 peaked at $1.8 billion annualized, Q2 guidance implies sub-$1.5 billion. That’s not just macro; it’s model mayhem.
Here’s my unique take, absent from the headlines: this mirrors the Theranos blood-test fiasco, but in fintech flavor. Like Elizabeth Holmes pitching miracle tech without ironclad real-world proofs, Upstart demo’d lab-perfect AI (80%+ approvals on prime borrowers) but flubbed the live-fire economic stress test. Bold prediction? If courts buy the “overresponsive” defense, it greenlights sloppy AI deploys across Wall Street. But I doubt it — juries hate tech-bro hubris.
Can Upstart’s AI Lawsuit Derail Lending Innovation?
Short answer: probably not fatally, but it’ll scar trust. Upstart’s still got 2,000+ bank partners, $6 billion+ cumulative loans. Model 13 (the culprit) was meant to auto-approve 75% of loans under $50k — efficiency gold. Yet overreaction exposed a flaw: black-box AI amplifying human biases in data, or just poor calibration to rate cycles?
Market dynamics scream caution. Fed funds at 5.25-5.5%, consumer debt at records — lending’s a minefield. Upstart’s connect platform grew 10% QoQ, a bright spot. But lawsuits like this? They’ve hit fintechs before: Rocket Companies settled for $2.2 billion on similar guidance games. Upstart’s cash pile ($500 million) cushions legal hits, but repeated misses erode moat.
Skeptical take: execs chased short-term pops over prudence. Why debut half-baked? PR spin calls it a “learning” — nah, it’s governance fail. Investors should’ve sniffed the volatility from Q4 ‘23 warnings, but rosy guidance blinded ‘em.
But wait — silver lining? This forces better disclosure. Expect Upstart’s Q3 call to drip with model sensitivity charts, stress tests. Competitors watch closely; nobody wants their AI labeled “overreactive” in court.
A dense stretch: Regulatory eyes sharpen too — SEC’s probing AI disclosures post-FTX, and this fits the bill, potentially birthing new rules mandating backtests for revenue-impacting algos, much like Volcker’s prop-trading curbs post-‘08, reshaping fintech’s wild west into something accountable (finally), while Upstart licks wounds and iterates to Model 14, hopefully less bipolar.
The Bigger Picture for AI-Driven Lenders
Zoom out. Lending AI market? $20 billion by 2028, per Statista. But trust is fragile. Upstart pioneered, now bleeds first blood in the over-hype wars. Peers like Zest AI, Ocrolus tout explainable models — less sexy, more reliable. Upstart’s edge? Scale. If they fix twitchiness (say, ensemble methods blending old/new models), rebound’s on.
My sharp position: this strategy never made sense. Debuting unseasoned AI amid election-year volatility? Recipe for lawsuits. Execs bet house on soft landing; macro gods laughed. Prediction: stock bottoms at $20, rallies 50% by EOY if guidance stabilizes — but only if they drop the hype machine.
Quick hit: Investors, diversify beyond AI lenders.
🧬 Related Insights
- Read more: Two UK FinTechs Dead. Here’s What Their Failures Reveal About the Real Consolidation Crisis
- Read more: Stablecoins Edge Closer to Bank Wires in LATAM and East Africa – Hype or Real Shift?
Frequently Asked Questions
What caused Upstart’s AI model overreaction?
The model was too sensitive to economic signals, approving floods of loans early in the year then pulling back sharply, forcing revenue guide cuts.
Will the Upstart lawsuit bankrupt the company?
Unlikely — strong balance sheet, prior wins in court, but it could drag stock and force conservative guidance for quarters.
Is Upstart’s AI still worth investing in?
High risk, high reward; wait for Q3 results showing model fixes before jumping in.
