Aggregation Theory

TL;DR: Pre-internet, you captured profits by controlling supply or distribution. Post-internet, you win by aggregating demand. The value chain flipped, and most strategy textbooks still haven’t fully updated for the new orientation. Ben Thompson’s framework is the one that did.

What it means

Thompson identifies the structural shift in how value is captured in the modern economy. The value chain has three parts: suppliers, distributors, consumers. Pre-internet, controlling distribution was the path to profits — a TV network, a record label, a newspaper. Post-internet, distribution is essentially free (the internet itself is the distribution), so power flows to whoever aggregates consumer demand. Google aggregates search intent. Facebook aggregates attention. Amazon aggregates purchase intent. Spotify aggregates listening. Netflix aggregates viewing.

The companies that figured this out early built businesses with margins that look impossible by old-economy standards — because they’re not in the old economy at all. They’re in a new game where the question isn’t “how do I get distribution?” (free) but “how do I become the place users go first?” (the entire game).

The argument

Aggregators win by owning demand. They commoditize suppliers by giving consumers what they want more efficiently than any single supplier can. This creates a self-reinforcing flywheel: more consumers attract more suppliers, more suppliers improve selection, better selection attracts more consumers (network-effects, moats). Scale economies kick in on top of that: the aggregator’s infrastructure gets cheaper per transaction with volume, which the suppliers cannot match individually.

The brutal asymmetry: the suppliers compete fiercely against each other while the aggregator captures the value. Restaurants on DoorDash compete to be at the top of search; DoorDash takes a cut from every order. Sellers on Amazon compete for the buy box; Amazon takes a cut from every sale. Musicians compete for Spotify playlist placement; Spotify takes the subscription revenue. The aggregator doesn’t have to be the best at anything except being the place users start.

This reframes distribution. If aggregation is the dominant strategy, then distribution isn’t about getting your product to users — it’s about becoming the surface through which users find all products. The aggregator doesn’t need to make the product; it needs to own the relationship with the consumer. Most strategy advice from the 1990s gets this exactly backwards.

Implications for startups. You either become an aggregator (extremely hard — requires massive scale and a one-time category-defining moment) or you differentiate so sharply that aggregators can’t commoditize you (counter-positioning, monopoly-vs-competition). This is where crossing-the-chasm-concept becomes critical: you need a beachhead small enough to dominate before the aggregators notice you exist. The middle ground — being a generic supplier on someone else’s platform — is the worst position in modern economics.

The ChatGPT version

ChatGPT is the most recent and clearest example of an aggregator forming in real time. Sam Altman saw this immediately and stated it bluntly: “a lot of people thought they wanted their apps to be inside ChatGPT but what they really wanted was ChatGPT in their apps” (openai-plans-altman). That sentence is Aggregation Theory in nine words. OpenAI is becoming the aggregation layer for AI capability, with model providers, plugins, and custom GPTs all positioned as suppliers competing for attention inside the surface that owns the demand. Ben Thompson made the argument structurally a decade ago and called the OpenAI version of it as it was happening (openai-windows-play).