Citi’s 15th Annual FinTech Conference
Logotype for Upstart Holdings Inc

Upstart (UPST) Citi’s 15th Annual FinTech Conference summary

Event summary combining transcript, slides, and related documents.

Logotype for Upstart Holdings Inc

Citi’s 15th Annual FinTech Conference summary

10 Nov, 2026

Operating performance and macro sensitivity

  • Q3 Contribution Profit rose 48% YoY for the second consecutive quarter, despite a nine-point average UMI headwind; fee revenue guidance and the implied Contribution Profit outlook were raised.

  • Full-year revenue and EBITDA guidance was lowered as worsening macro conditions pressured balance sheet income; management characterized execution as strong and macro as the key earnings volatility driver.

  • Fair value marks are more macro-sensitive than fee revenue because they reflect expected future loan cash flows; most impacts are captured quickly, though accounting complexity can create some lag.

  • Market rates, including Treasuries and SOFR, typically pass through to borrower pricing within about a month; rate increases in late September affected October loan pricing.

  • Management described macro changes as a headwind rather than a cap on growth; technology and marketing gains can bring borrowers back within approval limits, while sustained adverse conditions can offset those gains.

Underwriting, AI and competitive edge

  • Management said underwriting accuracy is 2.77x a traditional benchmark, with the model at 87% of the error range between random predictions and perfect cash-flow predictions.

  • Contribution Profit is the preferred measure of whether model advances create value; management said differentiated underwriting and sustained Contribution Profit growth signal competitive advantage.

  • Compute cost and latency remain major R&D challenges; model complexity increases resource demands, followed by optimization work before developers advance to the next model generation.

  • Proprietary lending algorithms require low-level optimization, and advances that reduce LLM token costs do not directly resolve these compute challenges.

  • Management sees AI agents potentially lowering the cost of comparing financial products toward zero, favoring the best product over customer lock-in and walled-garden strategies.

Product strategy and secured lending

  • Secured products, Cash Line and home lending are among the new product bets management is prioritizing; secured products turned Contribution Margin positive in 2026, one quarter ahead of the year-end goal.

  • The improvement in secured economics came from loan-sale execution, value captured on deals and cost management.

  • Auto-secured personal loans are closest to the existing personal-loan ecosystem; auto purchase distribution through dealerships is established, while HELOC is the least mature product and needs new home-adjacent partnerships and investment.

  • Auto refinance was discontinued because the market was not established enough to support customer education; strategy now favors products with existing demand and unmet technical service needs.

  • Multiple products diversify the portfolio versus personal loans alone; HELOC demand may benefit relatively when higher rates constrain home moves and refinancing, while higher rates and weaker borrowers remain headwinds across lending.

Partial view of Summaries dataset, powered by Quartr API
AI can get things wrong. Verify important information.
All investor relations material. One API.
Learn more

Latest events from Upstart