May 31, 2026 · 1h 19m · lennys-podcast
The most rational take on AI you’ll hear this year
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this in-depth conversation, technology analyst Benedict Evans joins Lenny Rachitsky to deliver a rational, historically grounded perspective on artificial intelligence, enterprise value capture, and workforce evolution. Evans demystifies alarmist labor and AGI narratives, explaining why distribution, vertical workflows, and human adaptability will define the next decade of technology.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Lenny holds 15.5% of the talking time here. How this is scored →
Lenny as informed peer 2.8 Guest teaching 5.7 Guest disagreement 3.6 Lenny pushing back 1.2
speaking balance: gold is Lenny, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 1:02:28 Eviscerating O*NET job automation scoring models
Evans forcefully dismisses quantitative job exposure matrices as a deluded expert systems fallacy, arguing that attempting to calculate percentage exposure for professions like senior law partners is completely absurd.
Hardest push from Lenny ▶ 26:03 Challenging historical parallels with AGI cognitionLenny challenges Evans's historical cyclicality argument by pressing that potential AGI and superintelligence represent an unprecedented replacement of human cognition rather than standard mechanization.
Biggest teaching moment ▶ 31:25 Telecom utility economics versus foundation model hypeEvans delivers a detailed economic critique of Sam Altman's metered intelligence thesis, demonstrating through mobile carrier capex history how undifferentiated infrastructure providers see margins collapse while application layers capture value.
Lenny holds their own ▶ 29:46 Introducing Andreessen's trillion-dollar TAM thesisLenny demonstrates his own market analysis depth by citing Marc Andreessen's thesis on expanding TAMs and accelerated enterprise revenue velocity to test Evans's assumptions about corporate value capture.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Lenny as informed peer | Guest teaching | Guest disagreement | Lenny pushing back | Why |
|---|---|---|---|---|---|---|
| Episode Preview and AI Perspective Highlights | 0 | 0 | 1 | 0 | Lenny delivers the monologue intro, previews the episode themes, and plays introductory audio clips framing Benedict Evans's rational perspective on AI hype. | |
| The 1997 Internet Analogy and Current State of AI | 2 | 5 | 3 | 1 | Lenny asks what markets are failing to price in regarding AI disruption. Evans offers a contrarian framing that AI is only as big as mobile or the internet, analogizing current adoption to 1997 where most infrastructure and applications remain unbuilt. | |
| Sponsor Break: WorkOS Enterprise Developer Platform | 2 | 6 | 2 | 1 | Following a sponsor read, Lenny asks how far society is from widespread transformation. Evans uses the history of VisiCalc and spreadsheets to illustrate the jagged frontier of technological adoption across different sectors. | |
| Why AI Labs Invest in Professional Services | 3 | 5 | 2 | 1 | Lenny observes the counterintuitive trend of frontier AI labs buying consulting services and deploying forward engineers. Evans explains enterprise operational inertia and why organizations require third-party consultants to re-engineer internal workflows. | |
| Deconstructing Tasks Versus Jobs and Jevons Paradox | 3 | 7 | 4 | 1 | Evans dissects the difference between automating discrete tasks versus entire jobs using Jevons paradox and historical accounting data. He mocks online AI influencers claiming they can replicate McKinsey slide decks with generic LLM prompts. | |
| Historical Labor Shifts and Enterprise Sales Cycles | 4 | 7 | 6 | 2 | Lenny cites tech companies increasing headcount and questions the narrative of catastrophic labor collapse. Evans dismisses Twitter doomers as morons and rejects Dario Amodei's labor market theories, pointing out that enterprise sales cycles take years to restructure industries. | |
| Historical Precedents: Calculators, Barcodes, and Research | 2 | 6 | 2 | 1 | Lenny appreciates the historical reassurance of Evans's perspective. Evans references vintage 1950s IBM calculator advertisements and retail barcode history to demonstrate how technology perpetually expands market capacity. | |
| The Ambiguity of AGI and Superintelligence Definitions | 3 | 6 | 5 | 2 | Lenny pushes Evans on whether AGI and superintelligence represent fundamentally unprecedented disruptions compared to historical shifts. Evans argues that definitions of AGI are moving goalposts and vibes forecasting, comparing semantic debates to crypto versus blockchain arguments. | |
| Market TAM Expansion and the Commodity Dilemma | 4 | 7 | 5 | 2 | Lenny brings up Marc Andreessen's thesis on expanding enterprise TAMs. Evans counters Sam Altman's metered utility vision by citing telecom history, showing how mobile carriers invested billions in capex only for profits to accrue to application layers. | |
| Sponsor Break: Vanta Compliance Automation Platform | 3 | 6 | 4 | 1 | Lenny summarizes Evans's thesis that foundational model labs face margin compression while application layers capture value. Evans elaborates with a dot-com bubble anecdote illustrating why undifferentiated commodity suppliers lack sustainable pricing power. | |
| Distribution as the Primary Moat in AI | 4 | 6 | 3 | 1 | Lenny suggests distribution is becoming the ultimate moat as software becomes frictionless to build. Evans agrees and compares the dynamic to the 1990s browser wars, noting Google and Apple will leverage massive default surface distribution. | |
| Deconstructing Anti-AI Sentiment and Resource Myths | 3 | 7 | 5 | 1 | Lenny asks about growing societal anti-AI sentiment and protests. Evans breaks down the claims, citing Lawrence Livermore laboratory data showing data centers use only 0.017 percent of US water and explaining how cultural panics mirror historical social media backlashes. | |
| Parenting, Societal Risks, and Technological Disruptions | 3 | 6 | 3 | 1 | Lenny asks how Evans is preparing his children for an AI-altered future. Evans discusses historical database panics and the UK Post Office Horizon scandal to emphasize that societal harm stems from systemic implementation rather than the novelty of technology itself. | |
| Career Evolution, Skills Synthesis, and Paradigm Shifts | 3 | 6 | 2 | 1 | Evans outlines a framework for career resilience based on intersectional skills. He uses the streaming music industry's U-shaped recovery curve to show how paradigm shifts create entirely new business models rather than simply replicating old formats. | |
| Fallacies of Granular Job Automation Scoring Models | 3 | 8 | 7 | 1 | Lenny notes how unexpectedly coding was automated. Evans attacks governmental granular job-scoring models like O*NET, branding them deluded expert system fallacies, and contrasts Uber's taxi displacement with Airbnb's limited impact on business hotels. | |
| Navigating Radical Uncertainty and Practical Career Strategies | 3 | 6 | 5 | 2 | Lenny asks for pragmatic advice for professionals facing radical uncertainty. Evans cautions against moralizing resistance on platforms like Bluesky, urging professionals to actively submerge themselves in AI tools to understand their practical capabilities. | |
| AI Corner: Personal Workflows and LLM Limits | 3 | 4 | 2 | 1 | Lenny opens the AI Corner segment. Evans explains why LLMs struggle with his core work of synthesizing precise data, sharing modest personal use cases like room redecorating and voice memo dictation. | |
| Lightning Round: Books, Cinema, and Vintage Phones | 2 | 4 | 3 | 1 | Lenny leads the lightning round covering book recommendations, classic cinema, and Evans's collection of 20 to 30 vintage pre-smartphone mobile devices. |
Statements from this episode (23)
Evans: AI impact equals internet and mobile, but not industrial revolution
“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.”
Evans: Picking AI winners today is like predicting Excite versus Yahoo
“And so then you can kind of get into calling those races where, again, it's like being in 1997 and saying, well, is it going to be Excite or Yahoo? And the answer was no, generally.”
Evans: Claude Code is to software development what VisiCalc was to accounting
“Software developers are the accountancy of VisiCalc. Like, oh my God, this changes everything. Like before VisiCalc and after VisiCalc. Before, before Claude Code and after Claude Code.”
Evans: Survey shows only 15-20% of teens are daily active AI users
“Even if you look at like, 13 to 18 year olds or something, it's still like kind of 15, 20% of people are daily active users, and another 20% are weekly active users. And then the other 60% of those people in that demographic, how long you say they are not usin…”
Evans: Enterprises lack spare staff to re-architect workflows for AI
“And so, you're supposed to, like, Completely re- imagine all of the internal workflows of your company and work out which of them could be automated really quickly with AI. That's a project. That's a project that needs like five or 10 people to sit down and sp…”
Evans: Management consultancies are hired for organizational navigation, not slide decks
“What you actually pay Bain to do is to go and walk all over your, and to try your company and work out. Yes, but why is it that you didn't do that? And how did the politics of this work? And what do you actually need to do? And let's go and talk to your custom…”
Evans: Accounting employment has consistently grown despite multiple waves of automation
“There's two charts in the presentation of the number of people employed as accountants, which went up right the way through the 20th century, and has gone up again since the beginning of the 21st century. So you have adding machines, and punch cards, and mainf…”
Evans: Running Anthropic does not make Dario Amodei a labor expert
“I don't think the fact that you run AI lab suddenly gives you, or rather, and if you're going to use argument from authority, then it should be relevant to the field. So, like, I'm interested in Dario's opinions on where models are going to go in the next six …”
Evans: Enterprise software sales take 18 months, exceeding VC funding cycles
“Enterprise software sales cycle is like, 18 months if you're lucky. You know, this is always a problem. The enterprise sales cycle is shorter than the venture backed startup funding cycle. Longer, longer, rather longer. Like it takes you longer to get an enter…”
Evans: Enterprise AI software replacement will take three to ten years
“So 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, yes, that whole estate will look radically different. And all those jobs will have changed, but it will take, you know, two, three, four, f…”
Evans: 1950s IBM ad pitched refrigerator-sized calculator as 150 engineers
“The ad is, the slogan of the title of the ad is, it's an IBM ad. It says an IBM electronic calculator. This is before it was called a computer. It's an electronic calculator. It's the size of a fridge. It's like having a 150 extra engineers.”
Evans: AI forecasting is purely speculative because we lack core theories
“We have no theory of what human intelligence is. We have no theory of why these models work so well. We have no theory of how much better they will get. So we're all just kind of vibes forecasting as to what will happen.”
Evans: AI will transform the world even if models stop improving today
“Even if like the model stopped, you're getting better tomorrow. If this is it, and we hit a brick wall tomorrow, this is an incredibly useful technology that's going to change the world and get rolled out over the next 10 years.”
Evans: AI Model Providers Will Lack Pricing Power and Become Commodities
“And so it does sort of seem to me that like, if the chatbot isn't the UX and it needs to be apps and the model companies aren't going to build that. And the models themselves are basically commodities, as, at least as you can see them as users, then why would …”
Evans: Major tech incumbents are unlikely to be fundamentally disrupted by AI
“You can look at, you know, Google, Apple, Facebook, Amazon, and say, it's hard to see a problem for them, really, with all of this. You can certainly see questions for all of them and one of them may drop the ball, but it's worth, you know, kind of remembering…”
Evans: Distribution and brand matter most when AI models are commodities
“So distribution of an adequate product when the field is basically commodity distribution on brand become a big deal.”
Evans: Apple Intelligence was the most compelling AI assistant vision yet
“If you go back and watch the WWDC from 20, 24, the whole second half of it is Apple intelligence. That was like the most compelling vision of a personal AI assistant. I've still, still the most compelling vision I've seen.”
Evans: Apple has roughly one billion devices capable of edge AI
“And in that situation, of course Apple's got like a billion devices that can run this on edge.”
Evans: Livermore study shows US data centers consume 0.017% of water
“Livermore lab. Did a study at the end of 2024 where they estimated US data center water consumption, and it came out at about 0.017% of US water consumption.”
Evans: AI will amplify society's worst instincts beyond deepfakes
“We connected everybody and unfortunately that meant we connected all the bad people and all of our own worst instincts and every problem in society. And so that will happen again with AI. You know, the, we can deep fake news are like the obvious thing we can s…”
Evans: Scoring precise percentages of job automation from AI is deluded
“I was looking at this as whole, there's a sort of US government called own data set called own natural or something like that, which tries to kind of analyze every single job and then people try and kind of score it and they try and say, well, you know, this p…”