← Glossary

Trickle-Down Tokenomics

The implied economics of 'superintelligence for everyone' pitches: universal free-or-affordable AI is promised at the top, those who can pay for more tokens get better models, and the business model funding the free tier goes unstated. Coined by Dan on ADI Pod Episode 37, of Zuckerberg's superintelligence essay.

Context

Coined by Dan on Episode 37 during the discussion of Mark Zuckerberg’s 6,500-word superintelligence essay. The essay promises everyone “free or affordable” superintelligence — personal agents, a tutor and coach with a PhD in every subject — while never stating who pays for the compute. Data centers cost billions; Meta’s only proven business model is advertising; the essay simply doesn’t connect those facts. Dan’s answer for the funding gap: “ad blood money.” Shimin’s Orwell gloss for the access gap: all pigs are equal, but some pigs can pay for more tokens — the essay itself concedes that law enforcement, at least, “will have more advanced models.”

Trickle-down tokenomics names the whole shape: abundance is promised universally, the mechanism by which it reaches everyone is asserted rather than designed, and the historical track record of that mechanism is the one the name implies.

Why It Matters

The term is a test you can apply to any lab’s everyone-gets-AI manifesto — a genre growing fast enough that the hosts have started calling it “pulling a Dario.” Three questions fall out of it. Who pays for the compute? If the answer is unstated and the author’s business is advertising, you have your answer. Is access actually equal? Free-tier tokens and best-model tokens are different products; a promise of universal superintelligence that preserves a paid tier above it is a promise of universal pretty-good intelligence. Who is the messenger? On the episode, Shimin agreed with most of the essay’s actual positions — personal agents, open weights, concentration-of-power worries — and trusted the document anyway less for who signed it: the same essay from a neutral author would read as vision; from the owner of the ad machine it reads as positioning.

It also pairs with the essay’s supply-side boosterism: the claim that data centers create local economic value omits that a modern data center employs well under 100 operational people, and that once every county competes to host one, even the tax revenue gets negotiated away. The tokens trickle down; the jobs mostly don’t.

Related Episodes