Upstart's AI Lending: Can Its Underwriting Model Stay Ahead?

Core Insights - Upstart Holdings emphasizes the use of real-time data and AI for credit underwriting, leveraging over 98 million repayment events to enhance model accuracy [1] - The company has observed early signs of rising risk in Q3, leading to tighter approval processes and a decrease in conversion rates from 23.9% to 20.6% [2] - Upstart is refining its calibration tools to reduce conversion volatility by approximately 50%, aiming for a more stable approval pipeline [3] Company Developments - Upstart is diversifying its offerings beyond unsecured personal loans, with growth in auto, home, and small-dollar products supported by advancements in property valuations and AI for document verification [4] - Despite management's confidence in new calibrations, investor concerns about approval volatility persist, compounded by sensitivity to macroeconomic trends like interest rates [5] Peer Comparisons - LendingClub Corporation reported a 37% year-over-year increase in loan originations to $2.6 billion and a 32% rise in revenues to $266.2 million, showcasing strong profitability with a 12.4% return on equity [6] - SoFi Technologies, Inc. achieved record net revenues of $961.6 million in Q3, a 38% increase year-over-year, with adjusted EBITDA rising 49% to $276.9 million and a 35% growth in its member base [7] Market Performance - Upstart's shares have declined by 22% over the past six months, underperforming the broader industry and the S&P 500 [8] - The company's current valuation indicates it is overvalued, with a Price/Sales ratio of 2.95X, aligning with the industry average [10] Earnings Estimates - The full-year 2025 Zacks Consensus Estimate for EPS has been revised upward, while the 2026 estimate has seen downward revisions, though both suggest significant year-over-year increases [11]