May 22, 2024 · 35m · big-technology
Box CEO Aaron Levie — What Cheaper, Faster, and Smarter AI Gets Us
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 a live recording of the Big Technology Podcast, host Alex Kantrowitz interviews Box CEO Aaron Levie about the accelerating pace of generative AI, declining compute costs, the transition to autonomous agents, and major competitive platform dynamics across open-source and proprietary ecosystems.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 29.5% of the talking time here. How this is scored →
Alex as informed peer 3.6 Guest teaching 4.2 Guest disagreement 2.2 Alex pushing back 2.3
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 3:03 Levie dismisses the AI slowdown narrative
Levie bluntly rejects Kantrowitz's premise that recent AI releases are underwhelming, comparing critics craving constant breakthroughs to addicts.
Hardest push from Alex ▶ 0:50 Kantrowitz challenges the AI hype cycleKantrowitz sets an adversarial opening frame, arguing that industry releases from Google and OpenAI fall short of promised godlike AI and sentience.
Biggest teaching moment ▶ 12:36 Levie educates on unit economics and the cloud precedentLevie reframes Kantrowitz's doubts about LLM profitability by walking through historical TAM expansion in cloud computing driven by developer accessibility.
Alex holds their own ▶ 11:44 Kantrowitz probes model pricing vs. training CapExKantrowitz drills into the sustainability of 50% API price drops against multi-billion dollar capital expenditure and OpenAI's delayed funding rounds.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| Live Show Welcome and Opening Remarks | 4 | 6 | 5 | 5 | Kantrowitz provocatively challenges current AI hype by claiming recent releases lack groundbreaking progress, calling out the absence of GPT-5 or sentience. Levie flatly rejects the premise, comparing tech commentators seeking immediate breakthroughs to addicts and schooling Kantrowitz on the 500x context window expansion in 18 months. | |
| The Four Vectors of AI Improvement and Startup Strategy | 5 | 7 | 2 | 2 | Kantrowitz accurately identifies model pricing drops as a core operational bottleneck for founders. Levie expands this into a masterclass on the multi-vector framework for model evolution, detailing trade-offs startups face when designing architectures ahead of capability curves. | |
| Unit Economics and Cloud Parallels in AI Business Models | 6 | 7 | 3 | 4 | Kantrowitz pushes Levie on the financial viability and gross margin sustainability of aggressive price cutting by foundation model labs. Levie responds with an extensive historical analogy comparing AI token pricing elasticity to the early economics of on-demand cloud computing. | |
| The Evolution from Chatbots to Autonomous Reasoning Agents | 5 | 5 | 2 | 3 | Kantrowitz asks whether newly demonstrated reasoning features deliver on rumored Q-star capabilities. Levie clarifies the technical reality of agentic workflows, explaining that models operate as reasoning brains coordinating external tools rather than monolithic solvers. | |
| AI Personalities and the 'Her' Persona Debate | 4 | 4 | 3 | 3 | Kantrowitz playfully corners Levie on OpenAI's controversial 'Her'-style voice assistant personality before pivoting to AI tutoring. Levie deflects the personality drama and focuses on the high-conviction educational and societal benefits of democratized personalized tutoring. | |
| Reading Levie's Tweets: Scaling Businesses via Compute vs. Headcount | 3 | 4 | 2 | 2 | Kantrowitz introduces his tweet-reading segment, citing Levie's post on substitution effects in business. Levie breaks down how enterprise scaling is shifting from human headcount addition to throwing raw compute at operational tasks. | |
| Reading Levie's Tweets: Meta's Open Source Play and 'Zuck Unleashed' | 3 | 6 | 2 | 2 | Kantrowitz prompts Levie on his tweet celebrating Meta's open-source strategy under Mark Zuckerberg. Levie delivers a comprehensive strategic breakdown of Zuckerberg's platform incentives to commoditize foundational AI against closed ecosystem competitors. | |
| Reading Levie's Tweets: Real-World Impressions of Tesla FSD | 3 | 3 | 2 | 2 | Kantrowitz quizzes Levie on his hands-on experience with Tesla FSD and visual memes from social media. Levie provides lighthearted commentary on autonomous driving progress and political ironies regarding state regulations. | |
| Mid-Roll Break and Live Sponsor Spotlight: EZ Newswire | 1 | 0 | 1 | 0 | A live-audience sponsor read for EZ Newswire where Kantrowitz delivers ad copy with playful interruptions from Levie about the uniqueness of live ad reads. | |
| Show Resumption and Transition to Audience Q&A | 2 | 0 | 0 | 0 | Kantrowitz handles live housekeeping and transitions the session to audience Q&A, introducing recurring contributor Ranjan Roy. |
Statements from this episode (11)
Levie: AI context windows improved 500x in 18 months
“And at the start of ChatGPT in let's call it November of twenty-twenty-two, the context window was somewhere on the order of about 4000 tokens. Just yesterday Sundar announced two million tokens on the latest Gemini model. So when you think about it there's li…”
Levie: OpenAI is intellectually honest and its demos work in practice
“But I'd say ChatGPT and OpenAI generally have been incredibly intellectually honest kind of You know, stewards in this space. And so when you see a demo from them, it tends to actually, you know, work like that in practice.”
Levie: OpenAI cut high-quality token costs tenfold in 18 months
“And the fact that we've now done that, or not we, I mean, OpenAI has basically done that, you know, I don't know, four or five times in the past 18 months. So we're already maybe a 10th of the total amount of cost of a kind of fairly high quality token just in…”
Levie: AI startups should build for models existing in three years
“Which is, if you're watching this curve, you probably should be building software for what is going to exist in three years from now, as opposed to today, because you don't want to be building software that sort of assumes that the tokens are expensive, and th…”
Levie: 10x cheaper AI models will drive 100x more usage
“If you could make, again, kind of wave magic wand and you say like we have GPT five or GPT six, and it costs like a 10th
Of what today GPT-IV costs I would argue that you'll probably get a hundred X more usage of AI, not, not just 10 X, you know, more usage of…”
Levie: Top five LLM providers will thrive, but 15th horizontal player will fail
“I think the business model is still very good for the top, let's say five or so model providers where, where you will run into a question is, you know, could you be the 15th LLM, you know you know, training company.
That, that seems tough, especially if your j…”
Levie: Cloud hyperscalers do not need high margins on AI itself
“Ultimately, like if you're a hyperscaler in the cloud, what you really want is all of our workloads.
You just want us to build our full application on your tech stack.
So
You're not really trying to make that much margin on the AI itself.
You actually want, li…”
Aaron Levie: LLMs serve as the common reasoning engine across agent architectures
“If you go and ask like 10 AI startups that are doing anything with agents, you'll probably get, you know, more or less 10 different architectures of kind of how the agent actually functions. But what is at least similar to all of them is the LLM or the model i…”
Levie: Companies will scale by adding compute instead of human headcount
“And the equation goes from, okay, I got to call the HR team. I've got to make sure we have budget. We have to go hire a lot of people to now it's like, well, do you want like a hundred leads or a thousand leads or 10,000 leads? Not, and that's not going to be …”
Levie: Meta's open-source AI push will be a long-term strategic success
“So I think it's both going to be a commercial success. And I think it's like structurally, strategically something that, that is is going to look, you know, like a very good decision in the long run.”
Levie: More compute will inevitably fix current imperfections in AI technology
“Maybe it doesn't work perfectly right now, but it's like, there's nothing that is, is stopping it from working perfectly in a world of more compute, And just more breakthroughs on, on, on the on the models themselves, and that is kind of where we're at right n…”