At a glance
WHAT IT’S REALLY ABOUT
Benedict Evans: AI adoption surges as foundation models commoditize fast
- Agentic coding has emerged as the clearest near-term product-market fit, rapidly concentrating industry attention and driving demand/supply imbalances in compute and pricing.
- OpenAI’s strategy has visibly oscillated from many adjacent product bets (ads, commerce, browser, etc.) back toward a sharper focus on coding, reflecting uncertainty about how to reach mainstream daily use.
- Evans argues foundation models lack durable differentiation levers (e.g., network effects) and may resemble infrastructure layers like telcos/ISPs where value migrates “up the stack” to applications and workflows.
- AI adoption is accelerating faster than prior platform shifts, but key uncertainties remain: daily vs weekly consumer engagement, where pricing equilibria settle after massive capex, and which industries beyond software achieve breakthrough utility.
- The most consequential impacts may be industry-specific and organizational (law, consulting, finance, advertising/commerce) where AI changes tasks before it changes “jobs,” and where measuring ROI is difficult and often competed away as consumer surplus.
IDEAS WORTH REMEMBERING
5 ideasCoding is AI’s first undeniable wedge, not the final destination.
Evans frames agentic coding as the rare use case with customers “pulling it out of your hands,” but notes the open question is which other fields will hit a similar inflection and when.
The chatbot is a V1 UI; real products require workflow design, data, and guardrails.
Most valuable deployments won’t be “ask the model anything,” but purpose-built tools embedded in systems of record (e.g., Salesforce/Workday) or vertical apps that structure probabilistic outputs into reliable processes.
Foundation models may become commodity infrastructure with weak pricing power.
Without strong network effects or sustainable differentiation beyond spending, model providers risk ending up like telcos—massive capex and usage growth, but value and margins captured by layers above them.
Today’s token economics are a disequilibrium that will not persist.
Evans compares current AI pricing to early mobile data: mismatched bundles, surprise bills, and capacity strain that eventually normalize into a new equilibrium as capex scales and efficiency improves.
AI’s business impact often becomes consumer surplus that gets competed away.
Like Excel enabling far more analysis without proportional price increases, AI can boost productivity yet not translate cleanly into higher vendor pricing or enterprise willingness to pay—especially in competitive services markets.
WORDS WORTH SAVING
5 quotesThey built this amazing piece of global, incredibly sophisticated, very expensive global infrastructure with enormous growth in use all the time. And it changed all of our lives, and we all pay for it, and they didn't make any money from it because all the value moved up stack.
— Benedict Evans
I don't think foundation models are a product. I don't think a chatbot is a product. I think the value will be further up.
— Benedict Evans
One of the characteristics of tech is that the moment that you understand something and you know how it works and what's gonna happen is the moment you should move on to something else.
— Benedict Evans
Just because demand for tokens is infinite, that doesn't mean that you can't get to a different price equilibrium.
— Benedict Evans
The way that all of this is sort of fundamentally different from previous platform shifts, is that with, you know, 3G or the iPhone or b- b- with the web or whatever it was, you didn't know what was gonna happen next, but you knew the physical limits. Like, you know, nineteen ninety-five, you knew that telcos weren't gonna give everybody in the world broadband next week... And with generative AI, obviously we don't.
— Benedict Evans
High quality AI-generated summary created from speaker-labeled transcript.
