Ben Thompson on Aggregation Theory, AI Agentic Commerce, and the Future of SaaS

In a wide-ranging conversation with Stripe founder John Collison, Stratechery's Ben Thompson discusses how aggregation theory continues to shape the internet economy, why AI is fundamentally changing SaaS business models, and what TikTok's ByteDance teaches us about platform dynamics.

The Evolution of Aggregation Theory

Thompson explains that aggregation theory, which describes how internet platforms capture user demand and intermediate supplier relationships, remains highly relevant but is evolving with AI. "The core insight was that in a world with zero distribution costs and zero marginal costs, the companies that control demand can intermediate supply relationships," Thompson says.

However, AI is creating new dynamics. Thompson notes that while traditional aggregators like Google and Facebook captured user attention through superior user experience, AI agents may bypass these platforms entirely. "If an AI agent can go directly to suppliers and fulfill user needs without going through an aggregator, that fundamentally changes the power dynamic," he explains.

AI Agentic Commerce

The conversation explores how AI agents are transforming commerce. Thompson points to ByteDance's success with TikTok Shop as a case study in how AI-driven recommendation systems can create entirely new commerce models. "What ByteDance has shown is that you can use AI to not just recommend content but to drive actual purchasing decisions," he says.

This shift toward "agentic commerce" represents a fundamental change from traditional e-commerce. Rather than users actively searching for products, AI agents proactively suggest and even purchase items on behalf of users. Thompson suggests this could dramatically reduce the power of traditional aggregators while creating new opportunities for companies that can effectively serve AI agents.

The SaaS Business Model Under Pressure

Thompson argues that AI is putting significant pressure on traditional SaaS business models. "The core SaaS model was selling software licenses with recurring revenue," he explains. "But if AI can automate many of the tasks that software was designed to assist with, the value proposition changes dramatically."

He points to companies like Microsoft, which is integrating AI deeply into its Office suite, potentially cannibalizing its own traditional software licensing revenue. "This is the classic innovator's dilemma," Thompson says. "The same AI that could extend their market could also undermine their existing business model."

TSMC and the Semiconductor Landscape

The conversation also touches on Taiwan Semiconductor Manufacturing Company (TSMC), which Thompson describes as "the most important company in the world that most people have never heard of." He explains how TSMC's foundry model has become essential to the AI revolution, manufacturing chips for companies like Nvidia, Apple, and AMD.

Thompson notes that geopolitical tensions around Taiwan add another layer of complexity to the semiconductor industry. "The concentration of advanced chip manufacturing in Taiwan creates both incredible economic opportunity and significant geopolitical risk," he says.

Running Stratechery in the Age of AI

Thompson reflects on how he's adapted Stratechery, his subscription-based tech analysis newsletter, in response to AI advancements. "The value proposition has shifted from simply providing analysis to providing context and synthesis that AI tools can't easily replicate," he explains.

He emphasizes the importance of developing a unique voice and perspective that goes beyond what AI can generate. "AI can summarize information, but it can't provide the kind of long-term thinking and pattern recognition that comes from years of studying an industry," Thompson argues.

The Future of Media and Content

The discussion addresses how AI is changing content creation and distribution. Thompson suggests that while AI will democratize content creation, it may also lead to increased concentration of audience attention. "The paradox is that as it becomes easier to create content, the value of trusted curators and brands may actually increase," he says.

He points to the success of platforms like Substack as evidence that audiences are willing to pay for quality content, even as free AI-generated content becomes more prevalent. "The key is providing something that AI can't easily replicate - whether that's unique analysis, entertainment value, or community," Thompson concludes.

The conversation provides valuable insights into how AI is reshaping fundamental business models and platform dynamics, while highlighting the continued relevance of aggregation theory in understanding the evolving internet economy.