Lenny's Podcast

A rational conversation on where AI is actually going

Benedict Evans
May 31, 2026 5 min read 1h 19m episode
TL;DR

Benedict Evans—independent analyst and former a16z partner—makes the case that AI is as big as the internet, and only as big. We're in 1997: exciting, mostly not working yet, killer apps not built. Foundation model companies are heading toward commodity infrastructure. The real value will accrue elsewhere—almost certainly to whoever owns distribution.

Key Moments
Benedict Evans
"My most controversial opinion is that I think that AI is as big a deal as the internet or mobile, and only as big a deal as the internet or mobile... we're in 1997. It's very exciting. Most stuff kind of doesn't work yet. Most of the stuff that people are going to do hasn't been built yet."
The opening thesis — AI is transformational but not uniquely unprecedented, and the hype is running ahead of where we actually are.
▶ 0:00
Benedict Evans
"Sam Altman said we're going to be selling 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 marginal structure of the utility industry to you. Because guess what? When you watch television, the TV company isn't paying a percentage of your monthly bill to the electricity company."
Why "selling intelligence like electricity" is a naive business model — utilities are indispensable and low-margin, not platform winners.
▶ 32:38
Benedict Evans
"The models don't seem to have network effects. So there doesn't seem to be a winner takes all effect where one of these will run away ahead of the other. So you should have competition indefinitely... then why would you have pricing power?"
The structural argument for why foundation model companies likely become commodity utilities — no network effects means no moat.
▶ 38:28
Benedict Evans
"You talk to these doomers on Twitter and they would act like every big company is going to buy ChatGPT tomorrow, and then in two weeks time, they'll fire all their stuff. And these people are morons... Enterprise software sales cycle is like 18 months if you're lucky... I know people aren't going to tear out SAP and replace it with X, Y, Z. Maybe in five, in like three, five, 10 years."
Why jobpocalypse predictions ignore how enterprise software adoption actually works — it's slow, political, and measured in years, not weeks.
▶ 21:00
Benedict Evans
"What's the hard part of the job? Is the hard part of the job writing the code line by line?... Or is the hard part of the job something else? Is it the task or the job?... What you actually pay Bain to do is go and walk all over your company and work out, yes, but why is it that you didn't do that? And how do the politics of this work?... The PowerPoint is just like the task. But that's not what you hired them for."
The task vs. job framework — the right lens for evaluating AI's actual impact. Automating the task sometimes kills the business model; often the task was never the point.
▶ 12:59
About Benedict Evans
›
Benedict Evans is an independent technology analyst and former partner at Andreessen Horowitz, where he served as their in-house "thinker" — tracking the most important technology trends for founders, investors, and operators. Before a16z, he spent years as an equity research analyst. For the past six years he's been publishing a widely-read weekly newsletter and bi-annual research presentations on where tech is heading. His most controversial opinion: AI is as big a deal as the internet or mobile — and only as big.
Takeaways
1
Models commoditize; distribution wins. Foundation models lack network effects and face relentless competition. The companies that will capture value are the ones that own the distribution layer — not the underlying AI. Meta spraying "adequate" AI on every surface may be enough to win.
2
Task vs. job is the right AI question. Don't ask "what percentage of this role can AI do?" Ask whether the task is actually the job. McKinsey's job isn't writing PowerPoints. A lawyer isn't paid to search case law. Confusing task with job leads to both overblown fear and missed opportunity.
3
The jobpocalypse is wrong for the right reasons. Not because AI won't change work — it will — but because enterprise adoption cycles are 18 months minimum, not overnight. The change will be slower, stranger, and more uneven than the doomers predict. Prepare, but don't catastrophize.