Ben Thompson from Stratechery on AI ads, the end of SaaS, and the future of media
TL;DR
Ben Thompson argues that AI companies should adopt Meta-style user profiling for advertising rather than contextual chat-based ads, which creates conflicts of interest and limits inventory; he also discusses Taiwan's urban convenience and how aggregators like Booking.com capture more value than suppliers.
🇹🇼 Taiwan's Urban Convenience 2 insights
Mixed-use zoning creates maximum convenience
Taipei's Japanese-influenced urban planning combines residential interiors with commercial exteriors and ground-floor shops, making it exceptionally convenient for residents despite 'dumpy' building exteriors.
Delivery economy transforming restaurants
Taiwan's massive Uber Eats market has caused many restaurants to close storefronts and operate exclusively as ghost kitchens, reducing the dine-out culture despite maintaining delivery quality.
🏗️ Aggregation Theory Dynamics 2 insights
Booking.com outperformed Google
A dollar invested in Booking.com 20 years ago returned more than Google, demonstrating how demand aggregators can capture more value than suppliers (hotel chains) despite being aggregated themselves by Google.
The double-layer aggregation risk
Booking.com aggregates hotels but is itself aggregated by Google—its biggest customer—creating a dependency where the aggregator remains vulnerable to upstream platform control.
🎯 AI Advertising Strategy 3 insights
Contextual chat ads are the wrong approach
OpenAI's current model of showing banner ads based on conversation context is the 'bare minimum' that limits inventory to specific chat topics and raises user suspicion about whether ads influence answers.
Meta-style profiling is superior
AI platforms should build comprehensive user interest profiles to serve discovery-based ads unrelated to specific queries, similar to Instagram, which shows users products they didn't know they wanted rather than harvesting existing intent like search.
User posture determines ad effectiveness
Instagram succeeds because users are in a relaxed, discovery-ready posture, while Twitter failed at advertising because users are in 'battle mode' seeking information, making them resistant to text-based ads.
💡 Monetization and Media Models 2 insights
Ads democratize AI access globally
Given the high cost of AI inference and the disparity between free and paid products, advertising is the only viable model to make AI accessible to six billion people lacking disposable income for subscriptions.
Content serves as social commonality
Thompson's Stratechery leveraged Twitter's early link-sharing culture to reach 200 countries, demonstrating that content works best as a 'something to talk about'—a social object that creates commonality between readers.
Bottom Line
AI platforms should abandon contextual conversation-based advertising in favor of Meta-style user profiling to serve discovery-based ads, avoiding conflicts of interest while funding universal access to expensive AI capabilities.
More from Stripe
View all
The 20-year journey to fully autonomous cars with Dmitri Dolgov of Waymo
Waymo Co-CEO Dmitri Dolgov details the 20-year technical evolution from Google's self-driving moonshot to 500,000 weekly autonomous rides, explaining why full autonomy requires augmenting end-to-end AI with structured intermediate representations and a 'three teachers' training framework rather than relying solely on scaled-up vision models.
Creating prediction markets (and suing the CFTC) with Tarek Mansour and Luana Lopes Lara
Kalshi founders Tarek Mansour and Luana Lopes Lara recount their four-year battle to launch the first CFTC-regulated prediction market in the US, culminating in a lawsuit against their own regulator to offer election contracts, and why their 'permission-first' approach ultimately enabled $10+ billion monthly volumes.
Bret Taylor of Sierra on AI agents, outcome-based pricing, and the OpenAI board
Bret Taylor explores how AI agents are shifting from polished but forgetful tools to messy, context-rich systems that leverage markdown memory and code repository structures, predicting software engineering will evolve from writing code to crafting 'harnesses' of documentation while enterprises move beyond APIs toward agent-accessible infrastructure.
Garrett Langley of Flock Safety on building technology to solve crime
Garrett Langley explains how Flock Safety grew from a neighborhood project solving car break-ins to a $500M ARR company serving 6,000+ cities by building solar-powered license plate cameras, AI search tools, and drones that help law enforcement clear over one million crimes annually through real-time data coordination.