Lenny: My guest is Benedict Evans. Benedict was a longtime partner at A16Z as their in-house analyst and resident thinker. Before that, he was a longtime equity researcher. For the past six years, he’s been an independent analyst tracking the most important tech trends. Most recently, he’s spending all his time on how AI is changing our lives. In his words, AI is eating the world. In this conversation, we go deep on what we’re still not pricing in about AI’s impact, the rise of anti-AI sentiment, jobs, where in the value chain most of the value will accrue, and tons more.
As Big as the Internet — and Only as Big
Lenny: You just put out this deck called “AI is Eating the World.” What do you think people are still not fully pricing in when they think about the change coming to their lives and work?
Benedict: An interesting way of thinking about it — my most controversial opinion is that I think AI is as big a deal as the internet or mobile, and only as big a deal as the internet or mobile. Clearly there are people in tech who think this is more like the industrial revolution. And there are people underneath saying, “He thinks this is just as big as the internet — does he not understand how big this is?” And I’m like, smartphones were quite a big deal. The internet was quite a big deal.
But if you dig into the internet comparison, it’s like we’re in 1997. It’s very exciting. Most stuff kind of doesn’t work yet. Most of the stuff people are going to do hasn’t been built yet, and it’s not really clear how any of it’s going to work when it does.
The people who’ve already got it imagine that everybody in the world is already there. The truth is you’ve got this very wide distribution. People in tech who bought their cluster of Mac Minis and don’t use Google anymore. Then you look outside tech — most people using this are using it every week or two, maybe.
There’s a fractal point here. At the super high level, this is going to change absolutely everything. I don’t think it’s productive to say “Is it 20% bigger than the internet or 100%?” — those aren’t productive conversations. But it’s one of those fundamental changes and we don’t know how any of it is going to work.
The 1997 Timeline
Lenny: If we’re in this 1997 timeline for AI, do you have a sense of just the timeline to when things are radically changing?
Benedict: Unquestionably we’re already in that moment in software. There’s a conversation about what agentic and AI software development mean for the future of the software industry. There’s one extreme — nobody really believes you’ll just vibe-code your own Stripe. But clearly there are questions about what this means for the industry.
The other extreme is if you’re in a law firm: “This is all very interesting, but how exactly do we use this? How many associates are we going to hire next year? What does this mean for us?”
One of the analogies I used is: imagine you’re an accountant seeing the first spreadsheets in the late ’70s. Mind-blowing — you change the interest rate and all the other numbers change, a week of work in 30 seconds. But if you were a lawyer looking at that, you’d think, “That’s very clever and my accountant should see this, but that’s not what I do.”
Software developers are the accountants seeing VisiCalc. A lot of other people are still slightly puzzled. Even if you look at 13-to-18-year-olds, it’s still like 15-20% daily active users, another 20% weekly, and the other 60% not using it. So we’re in that 1997 moment.
Why AI Labs Are Investing in Professional Services
Lenny: Something you’ve been writing about is this unexpected investment in consulting services and forward-deployed engineers at AI labs. What’s going on there?
Benedict: If you have any experience of professional services — companies do not have lots of people sitting around waiting to do a big new project. You’re supposed to completely reimagine all the internal workflows of your company and work out which could be automated with AI. That’s a project that needs five or ten people to spend a month or two working it out. Then actually doing it is another project. Who’s going to do that? Because you don’t have a bunch of people sitting around not doing anything.
So you hire Bain or Accenture or whoever to help you work that out.
Lenny: You would think AI would mean we don’t need consultants anymore. Instead, the most cutting-edge AI labs are the ones most investing in these folks.
Benedict: One of the strands in my presentation is: what’s the hard part of the job? Is the hard part writing the code line by line? Is the hard part making the PowerPoint? Or is it something else?
Claude Code can write the code. But what code do you want? It can make the features, sure — but what features do you want? Who’s your customer? What’s the right product? How are you going to take it to market?
Why do you hire McKinsey? Are you hiring them to get a 75-slide deck? Narrowly, Claude will make a really crappy version of that. But that’s not what you paid them for. What you actually paid Bain to do is walk all over your enterprise and work out why you didn’t do that already, how the politics work, what you actually need to do, and go talk to your customers. The PowerPoint is just the task — that’s not what you hired them for.
The Job Apocalypse That Isn’t
Lenny: What’s your gist on the coming job apocalypse?
Benedict: Every time we have a new technology, it automates away a bunch of jobs and then that automation unlocks a bunch of new jobs. You can always see the job that’s going to go away and you don’t know the new job because it doesn’t exist yet.
We’ve had that process over and over again since 1800. Each time you get frictional pain and dislocation, a bunch of people lose their jobs, a bunch of towns get hollowed out — it all sucks. But when you come through on the other side, we’re all richer and we’re not worried about the crops failing anymore.
The question is: is there some a priori reason why this would be different?
One theory is it’s going to be way quicker. The adoption of AI is faster because you’re standing on the shoulders of giants — you don’t need to wait for everyone to buy expensive hardware. ChatGPT can get 900 million users because there are already 900 million people on the internet.
But the other answer is: you talk to these doomers on Twitter who act like every big company is going to buy ChatGPT tomorrow and fire all their staff in two weeks. These people are morons. A typical enterprise software sales cycle is like 18 months. People aren’t just going to tear out SAP and replace it with XYZ. Maybe in three to ten years the whole estate will look radically different — but it will take time sector by sector.
Even just looking at the most advanced AI companies — Anthropic, OpenAI — everyone’s increasing headcount. The companies you’d think would be least likely to add humans are adding many, many humans.
Tasks vs. Jobs: The Price Elasticity Question
Benedict: There’s a chart in my presentation of the number of people employed as accountants, which went up the whole way through the 20th century and has gone up again since the 21st. You have adding machines, punch cards, mainframes, databases, ERP, cloud, spreadsheets, PCs — and the number of accountants keeps going up. It must be more complicated than automation.
The Jevons paradox — price elasticity: if you make it cheaper to do something, do you do the same for less money, or do you do more for the same money, or do you do more for more money because you’ve got new ROI?
You could make the same point in software development. Before IDEs and libraries and operating systems, developers wrote all the code. Now if you write an iPhone app, 90% of the code is written for you by Apple. So we’ve got a tenth as many engineers now? Well, no.
One of the analogies that occurred to me is e-commerce. What Amazon does is get you the SKU. If you know what SKU you want, you go to Amazon and get it. If you don’t know what microphone to get, probably shouldn’t start on Amazon. Claude Code can write you the code — but what code do you want?
Will Foundation Models Have Pricing Power?
Benedict: There’s a quote from Sam Altman where he said they’re going to sell AI intelligence on a meter like water or electricity. And you look at this and think: my dear sweet child, you need me to explain the margin structure of the utility industry.
When you watch television, the TV company isn’t paying a percentage of your monthly bill to the electricity company. When you wash your clothes, Bosch isn’t paying a percentage of the washing machine price.
The models don’t seem to have network effects. There doesn’t seem to be a winner-takes-all effect where one runs away. You don’t have really radical differentiation in what the product is. Then why would you have pricing power?
If the chatbot isn’t the UX, and it needs to be apps, and the model companies can’t build all those apps — then the foundation models become undifferentiated commodity infrastructure. There’s a lot of science to it, but there’s a lot of science in mobile too. Nobel prizes in flat panel screens — still a low-margin commodity.
Global mobile industry has revenue of about a trillion dollars a year, spends $200 billion a year on capex. Mobile data consumption is an exponential curve, 1500-2000 times what it was in 2010. And the stocks have gone nowhere in 25 years — because it’s a low-margin commodity utility where all the cool stuff is made by somebody else further up the stack.
Lenny: So your sense is over time the foundational model companies will get their margins squeezed, and the bigger opportunity is in the application layer?
Benedict: This is a deterministic thesis. The models don’t seem to have network effects. You should have competition indefinitely. You don’t have really radical differentiation. Then why would you have pricing power? If you need thousands of applications built by different people, those can’t all be built by the model labs. So it should end up looking more like cloud than like Windows.
Now that may be completely wrong. Imagine having this conversation about the internet in 1997 — what would you have gotten right? You certainly would not have said a has-been PC company from Cupertino would win the whole thing.
Distribution as Moat
Lenny: There’s this thread about distribution becoming a bigger moat because as software is easier to build, the noise in the market goes up.
Benedict: Yeah. If the product is a commodity, then distribution is what matters. There’s an obvious comparison with web browsers — the browser product is just a really thin wrapper for a rendering engine. An input box and an output box. And what happened? Microsoft used distribution to break in. Then it turned out winning browsers didn’t matter because the value was further up the stack.
What’s happening now: Google is using distribution to drive Gemini. Meta sprayed AI on every surface and it wasn’t bad — it was fine. Distribution of an adequate product when the field is basically commodity — distribution and brand become a big deal.
The Anti-AI Sentiment
Lenny: I’m curious about the anti-AI sentiment that seems to be growing. AI is less popular than ICE. People trying to stop data centers. What do you think is going on?
Benedict: It’s a big fuzzy mess of different stuff. There’s tangible stuff — “my electricity bill went up” — which applies in a very small number of places. The water thing is weird because it’s completely fake. Data centers use water for cooling, but the Livermore Lab estimated US data center water consumption at about 0.017% of total. That’s a planning problem, not a data center problem.
Then there’s the jobs question, where the main answer from economists is: we really don’t know yet. There’s a bunch of charts that say yes and a bunch that say no.
Then you get niche things — people who draw book covers for young adult romance novels are upset. There’s a huge culture war over whether it’s okay to use AI. The “AI slop” question.
Some of this is like the backlash around social media but more compressed. Some was true, some sort of true, some wasn’t. You’ve got 20 different things: some really real, some really not real, a lot of fuzzy mess in the middle.
You Can’t Predict Which Things Get Exposed
Lenny: A few years ago the last profession you’d think would be automated is engineering and coding. Now it’s the most transformed role.
Benedict: I think trying to analyze every job and score it for “AI exposure” is ridiculous. There are two reasons. First, it’s the expert systems problem — you can’t describe a profession by breaking it down into which bits can be automated. You can’t look at a senior partner at a law firm and say “17% of their work could be automated.” This is horseshit.
Second: the taxi driver test. If we were in 1997, you’d say obviously taxi drivers can’t be automated by the internet — it’s got nothing to do with the internet. Maybe internet booking, but no, that’s not going to change anything. And of course, Uber completely changes the whole thing.
You can’t predict which things are going to be exposed. The stuff you don’t think is exposed — that’s how these things work. The big companies are things that didn’t look like they would work, didn’t look like they were exposed.
It’ll Probably Be Okay
Lenny: Knowing all this — things are going to change a lot, but it’ll probably be okay broadly — what would you recommend people do to be more successful in this future?
Benedict: The only answer I think one can have: don’t stick your head in the sand and say “I hate all of this stuff.” That gives you a great feeling of moral superiority and you can go on Bluesky and shout at everybody about how evil AI is. Great, I’m happy for you. But that’s not going to help.
What helps is diving into this completely, submerging yourself in it, and coming out understanding what you can do with it, how this changes things, how you can be a great hire.
That may not be particularly comforting, but I don’t think there’s an alternative. You have to dive into this and absorb it and internalize it and think about what it means — just as you and I did with mobile and with the internet.
Lenny: Benedict, this was amazing. I learned a ton. I feel better after this conversation.
Benedict: As I always say, my parents had good SEO. Google “Benedict Evans.” Sign up for my newsletter. And if you want me to come present to your board in the Caribbean, let me know.