“I often do this kind of Pepsi Coke test where I’ll put a couple of different models in front of people using some kind of a harness like open code and asking them to tell the difference. And everyone thinks they can tell the difference and the reality is no one can tell the difference… the thing that differentiates models outside of marketing is price… it’s now dominated by Chinese models.”
Clip via @paulkedrosky with @kantrowitz / Big Technology Podcast — AI Chips Are Losing Value Faster Than Almost Anything Else
The AI model layer is commoditizing. That’s not a prediction — it’s already the argument being made out loud.
Paul Kedrosky’s “Pepsi-Coke test” framing is sharp because it’s falsifiable. Put models head-to-head, strip the logos, and watch expert confidence evaporate. The argument: differentiation has collapsed into two variables — marketing and price.
And on price, the argument goes further: Chinese models are now setting the floor. That’s a structural shift in where the competitive pressure originates.
📐 The Structural Read — FourWeekMBA Analysis
On our Map of AI, the model layer sits in the middle of the stack — between chips and applications. When that layer commoditizes, value migrates up (to applications, UX, distribution) and down (to whoever controls inference cost). The companies winning in the middle are increasingly competing on price alone. That’s a brutal place to live.
“The thing that differentiates models outside of marketing is price.”
— As argued on Big Technology Podcast
🔧 Harness Theory — Framework Lens
If users genuinely can’t distinguish models, the winning move is not to build a better model — it’s to harness the cheapest one and invest the savings into product, UX, and distribution. This is Harness Theory in its purest form. The “builder” advantage at the model layer shrinks. The “harnessers” who route intelligently to the lowest-cost capable model gain structural margin.
⚠️ What This Argument Doesn’t Claim
This is one analyst’s observed argument from a podcast clip — not an industry consensus or a proven empirical finding. Blind taste tests have limits: task type, user sophistication, and evaluation criteria all shape outcomes. The argument is directionally interesting, not categorically settled. Treat it as a hypothesis worth stress-testing, not a verdict.
The Bottom Line
If Kedrosky’s taste-test argument holds, the model layer is already a commodity — and the real strategic question for every AI company is whether they’re building on top of that commodity or stuck competing inside it. Price pressure from Chinese models, as he argues, makes that question more urgent, not less.
Clip via @paulkedrosky with @kantrowitz / Big Technology Podcast — AI Chips Are Losing Value Faster Than Almost Anything Else
This is analytical commentary on a public podcast argument, not investment advice. Views attributed are the speaker’s as expressed in that episode.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
