💰 Market Cap: $4.5B | 🏢 1.2K Employees | 🌍 United States

👨‍💼 CEO Dave Girouard (ex-Google) | 🎯 AI-Driven Lending | 🇺🇸 San Mateo, CA

$39.88

📉 -$6.37 (-13.8%) Post-Earnings

-59% YTD | Down 59% from $96 Aug Peak

🚀 Needs:

Auto retail scales to 20%+ of originations maintaining 70%+ sequential growth trajectory with dealer rooftops expanding to 500+ locations • Home equity instant approvals reach 30%+ (from 20% in Oct) with multimodal AI document automation • Model calibration improvements deliver 50% reduction in unwanted conversion volatility • 17 new funding partner agreements signed in 2025 ramp to full production • Private credit partnerships expand beyond current 10 partners with 100% retention maintained • Conversion rates recover to 23%+ as UMI stabilizes • Cross-selling and marketing AI optimizations drive 50%+ uplift in incremental originations sustained

⚖️ Needs:

Q4 revenue of $288M achieved with contribution margin stabilizing at 53% • Full year 2025 revenue $1.035B with GAAP net income $50M delivered • Balance sheet loan holdings reduced from $1.2B through closing multi-year funding agreements across all new products • Auto rooftops double again reaching 200+ locations • Home equity instant approvals reach 15-20% with strong credit performance • Model improvements deliver stable 20-22% conversion rates • Seven new bank/credit union partnerships added quarterly maintained • Take rate optimization AI model successfully monetizes borrower value • Holiday Q4 seasonality managed appropriately

⚠️ Risk:

Model conservatism persists despite UMI normalization driving sustained conversion below 18% • Prime segment (720-750 FICO) UMI remains elevated at 1.5+ vs subprime 1.2-1.3 creating approval challenges • Balance sheet holdings stay at $1.2B+ as funding deal timelines extend beyond Q4 • Repayment speed increases reflect refinancing boom away from Upstart rather than credit improvement • Auto fraud concerns lengthen partner diligence timelines delaying rooftop expansion • Competitors grow faster by taking more credit risk that Upstart's conservative models decline • Model "over-responsiveness" continues despite 50% volatility reduction efforts • New product funding agreements face extended legal/bank relationship processes • Private credit partners reduce appetite amid macro uncertainty

💰

What Happened

Revenue: $277M (+71% YoY, +8% QoQ) missed $280M estimate by 6% with fee revenue $259M offset by $19M net interest income from balance sheet

Adj EPS: $0.52 crushed $0.42 estimate with GAAP EPS $0.23 aided by $7.2M gain on convertible debt repurchase

Conversion Drop: 20.6% from 23.9% in Q2 as models responded to UMI rising 0.2 points in July/August before reverting

Applications Surge: 2M+ applications (+30% QoQ) reached highest level in 3+ years despite conservative underwriting

Volume Growth: 428K transactions (+128% YoY, +15% QoQ) with 300K new borrowers and average loan size $6,670 (-12% QoQ)

📈

Why It Matters

Feature Not Bug: CEO Dave Girouard emphasized model conservatism is "a feature, not a bug" proving system working as designed

$50B Milestone: Platform generated $50B+ in AI-powered loans since inception with 98M borrower repayment events providing unmatched training data

Calibration Breakthrough: CTO Paul Gu announced 50% reduction in unwanted month-to-month conversion volatility from model improvements

Funding Strength: 10 active private credit partners with 100% retention, 7 new bank partners (best quarter for new logos), all-time high monthly funding capacity

Fastest Response Time: Models detect changes in borrower behavior faster than traditional backward-looking credit metrics like charge-offs

🎯

What's Next

Q4 Guidance: Revenue $288M (+4% QoQ) with fee revenue $262M, NII $26M, contribution margin 53%, Adj EBITDA $63M, GAAP net income $17M

Full Year 2025: Revenue $1.035B (fees $946M, NII $89M), Adj EBITDA margin 22%, GAAP net income $50M marking return to profitability

Balance Sheet Strategy: Multiple funding agreements across new products expected by year-end to release $1.2B invested capital

Take Rate Moderation: CFO Sanjay Datta announced plan to "dial up forward investment into customer lifetime value" by moderating take rates for higher volumes

Credit Tailwinds: Rising repayment speeds and decelerating consumption historically signal improving consumer health heading into 2026

💡

Bottom Line for Retail Investors

Upstart's 14% post-earnings selloff creates a compelling contrarian opportunity for investors who understand what actually happened. The market punished the stock for choosing credit quality over volume—its AI models detected macro signals (UMI up 0.2 points) and proactively reduced approvals while competitors kept lending aggressively. CEO Girouard's mantra: "the model is always right" and they "don't overrule the model." At $39.88, you're paying just 3.9x forward revenue for a company that just proved its discipline with a crucial technical win: 50% reduction in model volatility. With $50B in AI loans originated, 98M repayment events (vs competitors' limited data), and 100% retention of all 10 private credit partners, this represents maximum pessimism pricing. The auto retail business doubled locations, home equity reached 20% instant approvals using multimodal AI, and applications hit a 3-year high—all while maintaining exceptional credit performance. Base case $60 target (+50% upside) assumes steady execution, while bull case $85 (+113%) requires new products to scale and conversion rates to normalize. If you believe AI models that prioritize credit quality over short-term volume will win the next credit cycle, this 59% drawdown from the August peak is your entry point.

🤖

AI Lending Pioneer

98M Training Events: Platform processed 98M borrower repayment events with 105K new repayments daily creating proprietary dataset

50% Volatility Reduction: New calibration methodology cuts unwanted conversion volatility by half while maintaining precision

30% Latency Improvement: Parallelized loan pricing reduced end-to-end latency 30% enabling larger, more complex models

ML Take Rate Optimization: Launched true machine learning model to capture value in relation to borrower value creation

🚗

Product Diversification

Auto "Breakout Business": Retail doubled rooftops in Q3, expanded to 4 new states, achieved 70%+ sequential transaction growth

Home Equity Automation: Instant approvals grew from <1% at launch to 10% in Sept to 20% in Oct using multimodal AI for document review

Instant Funding Launch: Small-dollar relief loans launched 90-second funding in September for qualified borrowers

17 New Agreements: Signed 17 partner agreements in 2025 (9 in Q3 alone) for new product funding ramping Q4 and into 2026

💪

Profitability Inflection

GAAP Profitability Return: $32M net income with $7.2M gain on debt repurchase, $50M expected for full year 2025

Servicing Revenue Growth: Steady 10% sequential growth rate providing recurring revenue stream visibility

Balance Sheet Returns: Strong spreads above market base rates on $1.2B holdings offsetting fee revenue shortfall

Fixed Cost Discipline: Fixed expenses down 7% QoQ with variable costs up only 11% vs 15% volume growth showing leverage

📊

Demand Acceleration

Record Applications: 2M+ applications in Q3 (+30% QoQ) reached highest level in 3+ years despite conservative underwriting

Private Credit Dominance: 10 active partners with 100% retention since first agreement in 2023, renewed largest partner for second time

ABS Oversubscription: September securitization saw strong demand with oversubscription, spread tightening, 30 investors including 7 first-timers

Marketing AI Breakthrough: Proprietary targeting technique delivered 50% uplift in incremental originations from same marketing spend

📉

Conversion Rate Pressure

UMI Spike Impact: Upstart Macro Index rose 0.2 points in July/August triggering model conservatism before reverting

Approval Rate Mechanics: Conversion pressure from fewer approvals, slightly higher rates, and smaller approved loan sizes

Measurement Error Risk: Despite 50% volatility reduction, models still face sampling error from representing US economy with finite dataset

Q4 Lag Effect: Management expects Q3 model conservatism impact to continue affecting Q4 despite UMI improvement

⚠️

Credit Risk Exposure

Prime Segment Weakness: 720-750 FICO segment showing elevated default rates vs pre-COVID with higher UMI than subprime

U-Shaped Risk Curve: Sub-660 (UMI ~1.2-1.3) and 800+ performing well, but mid-700s segment showing elevated stress

Repayment Speed Ambiguity: Higher repayment speeds could indicate refinancing away from Upstart rather than improved borrower health

$1.2B Balance Sheet Risk: Elevated loan holdings create liquidity concerns with funding deal timing "not perfectly predictable"

🏦

Competitive Pressures

Competitor Volume Divergence: SoFi and LendingClub reported strong Q3 origination growth while Upstart's models tightened

Unknown Risk Models: Management admitted "we don't know what their models look like" when asked why competitors didn't tighten

Prime Segment Competition: Superprime (720+ FICO) showing "very competitive" environment with others posting healthy growth

Share Shift Risk: Analysts questioning if Upstart lost market share by "fighting with one hand tied behind your back" due to model conservatism

📜

Regulatory Uncertainty

Auto Fraud Scrutiny: High-profile bankruptcies in auto lending causing banks to do "a bit more diligence" and ask more questions

Extended Timelines: CFO noted auto fraud headlines "probably lengthened timelines" with diligence lists "doubled and tripled in size"

Deal Process Heavy Lifts: Multi-year funding agreements involve "legal processes, financing, bank relationships" that are "heavy lifts"

Black Box Perception: Industry continues to view AI models as opaque despite Upstart's transparency claims about fast-moving signals