When Steve Jobs walked onstage in 2007 and held up the iPhone, nobody in that room could have pitched you Uber.

The people in that room weren't short on imagination. The thing that made Uber possible (a supercomputer in your pocket that always knew where you were) was just so new that no one had lived with it long enough to see what it unlocked. The phone had to sit in our hands for a few years first.

Then, in 2010, somebody asked a question that made no sense in 2007: what if a stranger's car could find me in ninety seconds?

I think about that a lot right now and one of the 2 people I follow in AI just wrote about it and I wanted to share what he articulated better than I ever could.

Ben Evans just put out his latest "AI Eats the World," the presentation he does twice a year. He organizes it in three parts: capital, deployment, and change. The first two are the easy part to see from here. The third is where Uber lives.

Capital is the number nobody can quite believe. The combined annual spend of Microsoft, Alphabet, Amazon and Meta is going from roughly $220 billion in 2024 to a guided $700 billion in 2026. For context, that's more than double what the entire global telecom industry spends, and it's closing in on global oil and gas. Data center construction in the US now outspends office construction for the first time in history. None of that will be new news for people who read my newsletter.

What is interesting is when he talks about where the value in the industry lies: it probably sits up the stack, in the applications and workflows built on top, and the models themselves drift toward being commodities. Powerful, essential, and increasingly interchangeable. If that's right, the company spending billions on the engine may not be the one that captures the margin. I happen to agree with this point of view.

Deployment is where he highlights a pattern that news headlines don't often highlight. Yes, we have the chatbots. Yes, agentic coding is real and pulling serious dollars. But walk into most large companies and AI is not running the core systems, and most people are not using it every day. Evans puts daily AI use at around 13 percent of US adults, even as weekly use climbs toward 35 percent. People are trying it, not living in it. The spend is enormous. The daily habit is still thin.

Then there's change, and this is the part most leaders skip right past.

Change is the Uber part. It's the question almost no one can answer on day one: what becomes possible that wasn't possible before? Evans uses the barcode, and it's the cleanest example I've seen. The barcode started as a way to know what was on the shelf. Useful, boring. But once you knew what you had, you didn't need to do the stock check anymore. So then what did you do with that time?

AI is sitting at the barcode's day-one moment right now. Most of us are using it to do the thing we already did, slightly faster. The faster website. The quicker first draft.

That's fine for now.

But what happens when a task that used to cost real money becomes nearly free, and you stop asking "how do I do this cheaper" and start asking "what could I do now that was never worth doing before." Customer interviews. Research. A coat recommendation based on a photo of your closet instead of "people who bought this also bought that." Different question entirely. (His examples, not mine)

So here's the question I'd put in front of any CEO reading Evans this week. For your business, is AI a new tool, a new line of revenue, or an existential threat? Most leaders answer "tool" by reflex, because that's the part you can see on the keynote slide.

But is there an Uber story hidden in your industry? How do we ask the right questions to find it first?

Reply and tell me which of the three it is for you. I read every one.
Alex