The Distribution Moat in the AI Era

TL;DR

• AI is making software, writing and apps nearly free to produce. When supply gets cheap, value moves to whoever owns the relationship with the people choosing.
• That is how the internet's aggregators won, and the same logic holds for AI.
• The companies paying for the compute may keep surprisingly little of the gain.
• For a small builder, distribution and trust compound. Another feature doesn't.

Why it matters: The record labels survived the CD and nearly survived Napster. What they didn't survive was a hardware company that already had the buyers. The labels owned the songs; Apple owned the relationship. AI is running the same move, faster, and most people who make things haven't noticed that they're the labels.

The big picture

"First, the Internet has made distribution (of digital goods) free, neutralizing the advantage that pre-Internet distributors leveraged to integrate with suppliers. Secondly, the Internet has made transaction costs zero, making it viable for a distributor to integrate forward with end users/consumers at scale." — Ben Thompson, "Aggregation Theory"

Then: "suppliers can be commoditized leaving consumers/users as a first order priority," and the winners are those who provide "the best experience, which earns them the most consumers/users, which attracts the most suppliers." — Ben Thompson

Where the money goes next: "when modularity and commoditization cause attractive profits to disappear at one stage in the value chain, the opportunity to earn attractive profits with proprietary products will usually emerge at an adjacent stage." — Clayton Christensen, via Stratechery

AI fits the pattern:

"When it comes to AI, distribution and transaction costs are still free — the two preconditions for Aggregators — which means that the winners should be those with the most compelling products." — Ben Thompson, "Mythos, Muse, and the Opportunity Cost of Compute"

"…my bet is that owning demand will ultimately trump owning supply." — Ben Thompson

Between the lines

A version of the AI boom in which the technology wins and its builders don't:

"Rest of World equities do well, and AI does well, but the AI gains mainly benefit the end-user with only thin profit margins (or outright unprofitable) for the companies themselves, due to so much capex." — Lyn Alden

The backdrop is fiscal dominance: the point where "the sovereign debt load is big enough that raising interest rates increases the federal deficit by a larger dollar figure than it slows down bank lending" — Lyn Alden. In that regime, "almost every choice the Fed makes sucks" — Lyn Alden. Cheap capital keeps funding supply.

Put together: abundant capital builds abundant compute, compute makes everything else abundant, and the scarce thing left is the user's attention and trust.

Yes, but

The strongest counter-argument is that compute is scarce again: "The era of Aggregation Theory is behind us, and AI is again making technology expensive." — Doug O'Laughlin, via Stratechery

"It's opportunity costs, not marginal costs, that are the challenge facing hyperscalers." Companies "will not be able to assume they can serve everyone." — Ben Thompson

If one lab pulls clearly ahead on capability, owning compute could capture the surplus for a while. The demand side doesn't vanish in that world; it narrows to subjects and products people trust more than any model.

What I do about it

I build with AI every day, and the building keeps getting cheaper. What doesn't get cheaper is being the thing people choose to come back to. So that's where the time goes.

The bottom line: Own the demand, not just the supply. In AI, that means being the specific product, or person, people trust on a specific subject.