Dec 6, 2024 · 1h 9m · more-or-less
#76: In a World with 99.99% Accurate AI… (feat. Aaron Levie) · More or Less Podcast
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
Box CEO Aaron Levie joins the More or Less podcast to debate the economic consequences of near-perfect AI, Big Tech antitrust enforcement, and the political and deregulatory shifts emerging under the incoming Trump administration. Through vigorous discussion, the panel contrasts high-conviction AI optimism with warnings of software industry deflation and regulatory paralysis.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 50.9% of the talking time here. How this is scored →
The hosts as informed peer 5.2 Guest teaching 3.5 Guest disagreement 2.9 The hosts pushing back 4.1
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 36:08 Levie defends software expansion against the zero-sum premise
Levie firmly rejects Sam's deflationary argument, characterizing the host's perspective as a flawed zero-sum mindset that incorrectly assumes the enterprise software market has already reached peak digitization.
Hardest push from the hosts ▶ 47:40 Sam refutes the AI ad expansion hypothesisSam systematically dismantles Levie's optimistic monetization premise by pointing out that concise AI answers collapse ad inventory and destroy browsing surfaces rather than expanding monetizable attention.
Biggest teaching moment ▶ 9:35 Levie on negative network utility in Web3 tokensLevie articulates a precise structural critique of token incentives, explaining why financialization turns Metcalfe's law upside down by degrading the utility for late-arriving participants into an MLM dynamic.
The host holds their own ▶ 55:05 Sam attacks enterprise software pricing resilienceSam utilizes his deep venture background to explain how lower barriers to code creation enable enterprises to rebuild bespoke internal tools cheaply, stripping incumbents of pricing power and destroying software gross margins.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| More or Less Podcast Title Sequence | 3 | 1 | 2 | 3 | The hosts engage in light-hearted banter with Aaron Levie about past history in Web 2.0 and Twitter shitposting. Sam playfully prods Aaron about his fluctuating enthusiasm for crypto and politics before establishing the episode roadmap. | |
| Critiquing Web3 and Evaluating Bitcoin as Digital Gold | 4 | 5 | 3 | 4 | Levie explains why he soured on Web3, detailing that blockchains are fundamentally inefficient databases for non-financial applications and criticizing tokenomics for creating negative utility curves. Sam Lessin probes Levie's brief period of euphoria, but Levie systematically articulates his critique of financialization. | |
| Crypto Deregulation and the Trump Administration Transition | 4 | 3 | 2 | 2 | Jessica asks Levie how the incoming Trump administration and SEC appointments impact his perspective on crypto. Levie acknowledges the bullish regulatory tailwinds while standing by his technical distinction between financial and software architecture use cases. | |
| Silicon Valley Politics, Deregulation, and National Competitiveness | 5 | 4 | 3 | 3 | Levie discusses why he engaged with Democratic policy circles despite holding growth-oriented, deregulation views aligned with Bessent and tech leadership. The hosts prompt him to explain Silicon Valley's political realignment and the competitiveness of the American tech ecosystem. | |
| The Democratic Party's Strategy and Post-Election Identity Crisis | 5 | 3 | 3 | 3 | Sam and Dave compare the Democratic party to a founderless startup while Aaron diagnoses the party's ideological factionalism and unwillingness to make hard choices. The dynamic is collaborative and conversational, with Sam quoting Nate Silver via Nellie Bowles. | |
| Big Tech Antitrust, Data Gravity, and Platform Lock-In | 5 | 5 | 3 | 3 | Levie provides a balanced view of antitrust, explaining the friction caused by platform data gravity and licensing lock-in versus the need for healthy M&A. Dave Morin asks for clarification on data gravity, which Levie defines from an enterprise software standpoint. | |
| The M&A Debate: Venture Capital, Big Tech, and Innovation | 6 | 4 | 4 | 5 | Sam Lessin pushes back against conventional venture capital complaints regarding blocked acquisitions, arguing that mid-tier M&A rarely produces lasting value. Levie counters by outlining how blocking Big Tech M&A disincentivizes early-stage funding and distorts market competition. | |
| The 99.99% Accuracy Hypothesis and Enterprise Software | 5 | 5 | 4 | 6 | Levie posits that reaching near-perfect accuracy in reasoning models will unlock hundreds of billions in enterprise software value. Sam challenges the foundational premise, noting that high accuracy is an enormous unproven assumption and asking whether the shift will be deflationary instead. | |
| The Personal App Revolution and Box's No-Code Strategy | 6 | 3 | 2 | 3 | Dave Morin outlines his thesis on the personal app revolution where non-engineers build customized tools. Levie affirms the concept and explains Box's newly launched no-code app builder designed to empower users without needing probabilistic user interfaces. | |
| Challenging the AI Bull Case: S-Curves and Accuracy Asymptotes | 7 | 4 | 4 | 7 | Sam attacks the AI bull case directly, asking what happens if accuracy gains asymptote early on an S-curve like autonomous driving did. Levie defends his thesis by arguing that even current capability levels justify existing venture capital investments across targeted verticals like coding and search. | |
| AI Search Monetization, Inventory Collapse, and Consumer Commerce | 7 | 4 | 5 | 7 | Sam articulates how direct AI answers collapse advertising inventory compared to traditional search, predicting deflationary economics for digital commerce. Levie contends that reduced transaction friction invariably expands overall consumer purchasing volume, using food delivery and convenience as examples. | |
| Examining Platform Incentives, Training Roadblocks, and AI Faith | 6 | 3 | 3 | 6 | Jessica raises reporting from The Information on frontier labs hitting training roadblocks and challenges the panel on why they express unshakeable confidence in AI models. Sam chimes in that AI enthusiasm behaves like a religion, while Aaron and Dave acknowledge the architectural friction within platform incumbents. | |
| The Software Deflation Argument and Digital Market Economics | 7 | 4 | 5 | 8 | Sam forcefully argues that cheaper software creation destroys enterprise pricing power and software margins rather than expanding spend. Dave and Aaron push back with the long-tail bespoke hypothesis, arguing customers will pay high margins for hyper-specific custom workflows that were previously uneconomical. | |
| Analyzing Google's AI Dilemma and Strategic Maneuvers | 6 | 5 | 2 | 3 | Jessica asks about Google's strategic dilemma in AI, noting their technical talent versus innovator's dilemma risks. Levie delivers an insightful assessment of Sundar Pichai's tightrope between maintaining core search ad revenue and shipping competitive Gemini native products. | |
| Vintage Internet Brands and Media Asset Acquisitions | 5 | 4 | 2 | 3 | The conversation shifts to vintage internet assets, media turnarounds like Jeff Lawson purchasing The Onion, and Elon Musk's Department of Government Efficiency. Levie argues that slashing bureaucratic red tape is far more consequential for GDP than simply trimming headcounts. | |
| Pop Culture Corner: The Protein Boom and S-Curves | 4 | 1 | 1 | 2 | Jessica introduces her pop culture corner segment discussing celebrity protein snack brands and Japanese convenience store trends. The discussion is casual and comedic, concluding with references to whether consumer dietary trends follow an S-curve. | |
| Wrapping Up the Interview with Aaron Levie | 3 | 1 | 1 | 2 | The hosts wind down the podcast with appreciation for Aaron's appearance, holiday party chatter, and playful insider banter about hiring math tests versus anti-AI application filters. |
Statements from this episode (31)
Lessin: AI hype cooled in 2024 as costs and reality set in
“By my read, you know, it's been an interesting year, and I would say the year where maybe like, The reality of AI sunk in a little bit. The reality of the cost of AI sunk in a little bit. I don't know. I actually think we came down a little bit from the hype t…”
Levie: Blockchains are too costly and inefficient to host internet infrastructure
“The thing that lost me was when everybody was arguing that, that what we should do is decentralize all of the infrastructure of the internet. That's just literally, it just costs more to then run every internet service. And it's actually not a very good databa…”
Levie: Token-incentivized social networks fail because later users receive less utility
“I just don't think you can build a social network where financial You know, where monetization or incentives is at the core of the network. I think people have to join these things because they want the utility of whatever the network is. And that is how you s…”
Levie: A digital gold asset cannot support dozens of competing cryptocurrencies
“But like, but I am sort of a, probably a toxic maxi in the sense of like, I don't think you can have 10 of those. I think you kind of have to have everybody decide this is the one and you can't sort of fragment off into 50 others.”
Levie: Trump administration is unquestionable boon for crypto builders
“Oh, I mean, the Trump administration is very bullish for anybody doing crypto. There's no question.”
Levie: US leads world in AI, biotech, transportation, and space
“Like we are leading in AI. We're leading in biotech. We're leading in transportation in, in most categories. We have the chance to lead in, in energy in most areas. We're leading in space. So you kind of look at all of these areas where we're clearly in front …”
Levie: Unrealized capital gains tax was a disaster for capital cycle
“You had this unrealized capital gains thing. That was a total disaster that was sort of always creeping out there. That was going to obviously be bad for just the capital cycle.”
Levie: Trump and Elon Musk will reduce regulation and spur growth
“But like when you then compound, you know, Elon in there, it's like night and day in terms of, you know, one side is going to just reduce the regulation. It's going to be driving, it's going to be a growth oriented, you know, more dynamic, you know, kind of dr…”
Levie: Republican economic philosophy aligns more with my instincts than Democrats'
“And I think it's a, it's clearly a distinct philosophy from what the Democrats pitch. And it actually aligns, frankly, more to my instincts, but there's other things on the Democrat side that kind of social topics and others that, that, you know, sort of have …”
Levie: Democratic Party lacks identity and splinters on every topic
“I think that, you know, unfortunately, the very downer version of this would be, I think the Democrat Party doesn't know who would, what it wants to be, right? So, so forgetting even who leads it, it's, you know, there's just so many factions within the Democr…”
Levie: Restricting tech M&A halts the full cycle of innovation
“I probably favor more allowance for M&A to happen. I think there's been some kind of, you know, dubious cases around, you know, why can't iRobot get, get bought by Amazon? Why can't Facebook buy Meta, buy, buy the Giphy or whatever, the 50 person VR company? L…”
Levie: Internet data moats differ from historical monopoly dynamics
“The data moats that, that exists now in the internet are just totally different than almost any other form of antitrust, you know, kind of monopolistic dynamics that I think we've ever seen in history. Like, you know, clearly, obviously if you're building what…”
Levie: FTC scrutiny is hyper-targeted at just four or five Big Tech firms
“I think if you look at the actual track record it, you know, this is probably, you know, unfortunately, a negative on, on the FTC environment, but it does seem hyper targeted at four or five companies. You know, Splunk was acquired for thirty billion dollars b…”
Lessin: Most Big Tech acquisitions fail and end up shut down
“A lot of the big companies that did a lot of these acquisitions, they kind of all ended up shutting most of them down. Most of them don't work. Sure. You get it every once in a while you get like an Instagram, but like most of them are kind of nothings.”
Levie predicts 10 times more software will be created
“There'll be way more software, 10 times the amount of software that gets created.”
Levie: Extrapolating existing AI capabilities would pay for all VC funding
“If I'm wrong and it taps out sooner than I think you would just have exactly what AI is good at today, plus, you know, 20% or whatever, and then fully extrapolate it out. So, so that, that to me is, that, that pays all the, for all the VC funding.”
Lessin: Some leading AI companies are losing $5 billion annually
“Some of these things are losing five billion dollars a year.”
Morin: Voice AI will reliably book restaurant reservations within five years
“Five years from now, you will, Sam.”
Sam Lessin: AI search collapses ad inventory by answering queries directly
“AI today, even if you're a really big bull and you think that Gemini and open AI is like better at answers is not actually going to expand inventory, right? If anything, it collapses inventory because you're doing fewer searches, right? It just gives you the a…”
Levie: Increased AI device activity will counteract ad inventory collapse
“And that's why I think it will counteract whatever your sort of 10 blue link shift is.”
Sam Lessin: Most Google searches concentrate on a tiny set of things
“And also the reality is almost all Google searches are for like six things, right? Like the internet is actually much smaller. People's demands are much smaller than people think they are. The internet is very large, but the actual used internet is actually qu…”
Lessin: OpenAI CFO Sarah Friar floated an advertising business model
“I mean, you even saw this week, we saw Sarah Fryer, the CFO of OpenAI be like, maybe we'll do ads. And then there was a big reaction to that, right?”
Lessin: OpenAI and Anthropic have hit roadblocks in training models
“We've also done a series of stories recently, the information that have shown Roadbox, OpenAI and Anthropoc have hit in training. Not that they can't get through them, they work and, you know, but there are things that aren't working that we're working, right?”
Sam Lessin: AI will compress SaaS margins by making software easy to clone
“There's pricing pressure. All your margins go away. You say, look, I can rebuild box.net with three engineers in two months for my need now. I don't need you guys. You can't charge me premium pricing. And the reality is that's actually the easier application o…”
Dave Morin: AI will enable tens of millions to build high-margin custom software
“I think there's a large explosion of tens of millions of more people building more specific, more personalized software to various use cases that you can charge a higher margin for because it's custom to like, I don't know, to use Aaron's example, like the, yo…”
Sam Lessin: AI will completely wreck the porn market
“AI is going to completely wreck the porn market. Full stop.”
Levie: Google is the most threatened Big Tech company by AI
“Google is, is in, you know, singularly the trickiest spot, I think of the big tech players. Cause you know, this is not particularly disruptive to Meta. It's basically all upside. It's all kind of like an adjacent side business to AWS and Amazon. It's all net …”
Jessica Lessin: Yahoo will eventually go public
“You guys, Yahoo's gonna go public at some point. It's gonna be crazy.”
Levie: Department of Government Efficiency will be 2025's biggest story
“I mean, Doge will be the, I think will be the story of the year.”