💸 “The Rich Fee” — Uber Eats’ spy pricing. Nvidia’s Coachella for Chips. The Spa/Pilates Economy. +Don’t Invest in Duke

What It Covers

Three business stories from March 18, 2025: Uber Eats caught using AI-driven personalized pricing that charges different customers different amounts for identical orders; Nvidia's annual GTC conference reveals 900x chip performance growth over two years; and U.S. retail crosses a historic threshold with service businesses now occupying over 50% of all retail square footage.

Key Questions Answered

• •AI Spy Pricing: Uber Eats charged five New York customers different prices for the same order simultaneously. The app's fine print admits prices are set using personalized data — contradicting official denials. Only New York law requires this disclosure, meaning the practice likely operates undetected elsewhere. Consumers should assume apps linked to premium credit cards like Amex Platinum may trigger higher prices.
• •Chip Performance Compounding: Nvidia's Blackwell chip delivers 68x more performance per watt than its predecessor Hopper. Its 2025 successor, Rubin, adds another 13x leap — meaning Rubin achieves roughly 900x Hopper's computing performance in just two years. Jensen Huang projects computing demand has grown one million times in two years, and expects Nvidia revenue to reach $1 trillion within two years.
• •March Madness Bracket Strategy: Apply value-investing logic to bracket picks by finding teams where win probability exceeds public pick percentage. Duke carries a 20% win probability but appears in 30% of Yahoo brackets — making it the most overvalued team. Michigan has a 19% win probability but only 15% of picks, representing better expected value similar to buying undervalued stocks.
• •Retail Services Milestone: For the first time in U.S. history, service businesses — spas, gyms, fitness studios, and med spas — lease more than 50% of all retail square footage. Med spa locations have grown 6x since 2010, reaching 10,000 U.S. locations, approaching McDonald's count of 13,000. Service tenants pay premium rents, making them preferred by landlords over traditional goods retailers.
• •Enshittification Economics: Personalized algorithmic pricing — where AI sets individual prices based on scraped personal data — differs fundamentally from dynamic pricing, which responds to supply and demand anonymously. When consumers perceive pricing as unfair, the result is boycotts, increased personal data protection behavior, and reduced consumer spending confidence, which economists warn could measurably contract overall GDP.