
Summary
Written by a model from the transcript; not yet reviewed.
This episode explores the bull case for Anthropic's entry into the public markets through an interview with Box CEO Aaron Levie. The discussion addresses high valuations, token consumption, and the shift from basic prompts to agentic workflows that consume significantly more compute. The guest argues that AI is still in the early stages of enterprise adoption and that token usage will expand across various industries like manufacturing, cybersecurity, and life sciences. The conversation also tackles common counterarguments from critics, including infrastructure debt, growth slowdowns, and diminishing returns, providing a nuanced perspective on whether the explosive revenue growth can be sustained.
- Enterprise adoption of AI is still in its early phases, meaning token consumption and revenue growth are likely to continue expanding across the economy. 3:34
- Agentic workflows decouple knowledge work from human time limits, allowing a single person to deploy automated swarms that consume massive amounts of tokens. 15:59
- Frontier models are increasingly used as orchestrators that direct tool use and decision-making, while more efficient models handle high-volume data processing. 29:36
Chapters
- 0:00Podcast Intro and Host Greeting
The episode opens with the host introducing the Big Technology Podcast and teasing an optimistic discussion about Anthropic’s upcoming IPO.
- 0:40Defining AI Monetization Models
The host explains how AI companies charge for services, contrasting usage credits, outcomes, seats, and hybrid models, setting the stage for the episode’s focus on Anthropic.
- 1:38Aaron Levy’s Optimistic View on Anthropic
Box CEO Aaron Levy joins to argue why Anthropic will thrive in the public markets, contrasting his optimism with Gary Marcus’s skepticism.
- 2:40Token Consumption and Frontier AI Growth
Levy discusses how token usage is expanding beyond coding into manufacturing, cybersecurity, and other domains, driving Anthropic’s revenue growth.
- 4:53Anthropic’s Revenue Explosion and TAM Potential
Levy highlights Anthropic’s 12x revenue jump from Q3 2025 to Q3 2026 and the massive total addressable markets across industries.
- 6:00Agentic Workflows and Token Efficiency
The conversation turns to how frontier models orchestrate tasks while cheaper open‑weight models handle bulk token consumption, boosting overall spend on Anthropic’s platform.
- 15:27Future of AI Adoption and Enterprise Impact
The hosts discuss how AI will permeate every industry, from manufacturing to cybersecurity, and the implications for enterprise productivity and AI investment strategies.
- 30:38Fable 5.2 and Model Performance
Levy praises Anthropic’s latest frontier model, Fable 5.2, noting its superior reasoning, coding, and design capabilities compared to rivals.
- 38:52Rebutting Critics: Debt, Token Maxing, and Diminishing Returns
Levy addresses Gary Marcus’s concerns about valuation, debt, token growth slowdown, and diminishing returns, arguing they are manageable in the long run.
- 47:03AI Adoption in Industries and Duration Mismatch
Speakers debate how AI could capture a significant portion of industry work and the potential mismatch between investment timelines and industry change.
- 47:34OpenAI Funding Round Lessons
Discussion of OpenAI’s recent funding round, highlighting investor concerns and the impact on valuation expectations.
- 49:19Scenario Planning for AI Capital Allocation
Speakers outline the importance of scenario planning for AI investments, weighing compute limits against underutilized capacity.
Highlights
- 4:40
Unprecedented 12x Revenue Growthdata
Anthropic's revenue surged from 1.4 billion dollars in Q3 2025 to 17 billion dollars in Q3 2026, marking a dramatic 12x increase at tens of billions of dollars in scale.
- 7:41
Diffusion into Customer Walletsdata
The number of customers spending 100,000 dollars or more per year grew from 950 in late 2025 to 11,900, proving that broad-based utility and adoption are actively occurring.
- 15:59
Decoupling Time from Valueinsight
Knowledge work is no longer rate-limited by typing speed or mouse movement, as agents allow a single user to deploy the equivalent of thousands of workers using a set compute budget.
- 23:55
Inference Becomes Profitabledata
While training remains capital-intensive, Anthropic's unit economics shifted so that they now make 1.72 dollars for every dollar spent on compute, demonstrating successful monetization.
- 29:36
Frontier Plus Efficient Routingstrategy
Complex tasks will use frontier models for high-level reasoning and orchestration, while cheaper open-weights models process large volumes of logs and documents, causing spending to rise across both ecosystems.
- 32:50
Modern frontier models can synthesize documentation from dozens of companies across different industries to generate software designs and enterprise experiences that human product managers could not accomplish alone.
Quotes
“I think what I can make the case for is why I believe that you're still going to have, you're going to have a very strong market for frontier AI.”
Unnamed speaker · 2:51 “All of that leads me to be very bullish on the market broadly.”
Unnamed speaker · 3:23 “The entire stack, you know, makes money on that.”
Unnamed speaker · 15:47 “I think the round is still open?”
Unnamed speaker · 47:53 “I don't know what percentage I'd peg it at, but like 100% that could happen.”
Unnamed speaker · 48:15 “I think there's, I think some of, some of when I hear those scenarios, they, they, they sometimes come from the perspective that underestimates the utility that these models are generating.”
Unnamed speaker · 50:12 “And so now imagine that for any, any workflow in your enterprise that deals with content, that's, that's effectively what, what we built out.”
Unnamed speaker · 55:02
Sponsored segments
- Progressive 0:21–0:38
- Chargebee 0:59–1:16
- LinkedIn Pulse 35:20–36:26
- Intelligence AI 36:26–36:58
- DXC Technology 37:06–37:56
- Dell Technologies 38:06–38:36
- Progressive 56:44–57:00
About this episode
Aaron Levie is the CEO of Box. Aaron joins Big Technology to discuss the bull case for Anthropic’s IPO and the future of enterprise AI. Tune in to hear why he believes AI agents will drive massive demand, how adoption could expand beyond coding, and why cheaper models could benefit the frontier labs. We also cover AI’s reliability challenges, the risk that revenue won’t arrive before the bills come due, and how Box is putting agents to work. Hit play for a clear-eyed discussion of AI’s business potential and the assumptions behind the optimism. --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. Want a discount for Big Technology on Substack + Discord? Here’s 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b Learn more about your ad choices. Visit megaphone.fm/adchoices
Transcript
10,902 words · 6 speakers
Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.
Topics
- Anthropic
- AI
- tokens
- Box
- manufacturing
- inference
- Fable 5.2
- OpenAI
- RL
- Gemini
Page updated · transcript and summary are machine-generated