May 20, 2025 · 33m · allin
Sergey Brin, Google Co-Founder | All-In Live from Miami
Sergey Brin · 16m spoken Jason Calacanis · 6m spoken Chamath Palihapitiya · 3m spoken David Friedberg · 2m spoken David Sacks · 2s spoken
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
At All-In Live in Miami, Google co-founder Sergey Brin joins hosts Jason Calacanis, Chamath Palihapitiya, and David Friedberg to discuss his return to Google, technical breakthroughs in reasoning and AI infrastructure, the evolution of voice interfaces, and the future of AGI.
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 42.5% of the talking time here. How this is scored →
The hosts as informed peer 3.3 Guest teaching 2.8 Guest disagreement 1.4 The hosts pushing back 1.7
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 22:02 Dismissing Larry Page's evolutionary claim
Sergey directly rejects the philosophical premise raised by Friedberg that humans are an evolutionary stepping stone to AI, attributing the quote to nerds drinking too much wine.
Hardest push from the hosts ▶ 21:38 Pressing on humans as evolutionary stepping stonesFriedberg refuses to move past Larry Page's assertion about silicon intelligence surpassing humans, pressing Sergey for a direct comment on human evolutionary obsolescence.
Biggest teaching moment ▶ 13:40 Reframing the humanoid robotics hypeSergey educates the hosts on why humanoid forms are over-hyped, explaining that modern AI simulation enables software to navigate non-human physical configurations efficiently.
The host holds their own ▶ 25:24 Detailing chip transpilers and hardware stacksChamath demonstrates deep technical domain knowledge by questioning Sergey on transpiler layers, memory constraints, and the abstraction between NVIDIA GPUs and Google TPUs.
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 |
|---|---|---|---|---|---|---|
| Sergey Brin's Return to Google and AI Excitement | 2 | 2 | 1 | 1 | Jason and Sergey discuss Sergey's return to Google and compare the pace of AI to the early web era. The mood is conversational and friendly with no pushback. | |
| Hands-On Coding and Technical Freedom | 3 | 2 | 1 | 1 | Chamath asks about technical focus areas within the AI stack. Sergey details his shift from pre-training to post-training and thinking models. | |
| Gemini Deep Research and Unconventional Prompting | 4 | 3 | 1 | 1 | Jason explains his multi-query Gemini usage for F1 statistics. Sergey humorously shares that AI models perform better when threatened with physical violence. | |
| Parenting, Education, and College in the AI Era | 3 | 3 | 1 | 1 | Chamath asks how AI impacts parenting and education. Sergey reflects on AI advancing faster than high school curriculum, leading hosts to discuss college value. | |
| Robotics Progress and Humanoid Skepticism | 4 | 4 | 3 | 2 | Friedberg and Chamath question hardware progress and humanoid form factors. Sergey challenges humanoid hype, explaining that simulation lets AI master any shape. | |
| AI Developer Productivity and Internal Bureaucracy | 3 | 3 | 1 | 3 | Sergey admits Gemini was bizarrely placed on an internal banned coding list. Chamath playfully pushes back, reminding Sergey he is the co-founder and ultimate boss. | |
| Specialized Models vs. Open Source Breakthroughs | 4 | 4 | 1 | 2 | Chamath asks about model specialization and open-source progress. Sergey explains how ML models converged into transformers and praises DeepSeek. | |
| Human-Computer Interaction, Google Glass, and the Singularity | 4 | 3 | 3 | 3 | Friedberg presses on Larry Page's quote about humans being evolutionary stepping stones. Sergey laughingly dismisses it as drunk nerd talk and shares an AI management story. | |
| Infinite Context Windows and AI Hardware Infrastructure | 5 | 4 | 2 | 3 | Chamath asks about hardware transpilers and infinite context rumors. Sergey explains why hardware architectures like TPUs vs GPUs still require direct optimization. | |
| Voice Interfaces, Response Times, and Model Stacking | 4 | 2 | 1 | 1 | Jason and Chamath describe interjective voice workflows and model stacking. Sergey agrees that model latency and inference speed are advancing rapidly. | |
| Sergey Brin on Personal AI Workflows and Office Culture | 3 | 3 | 2 | 2 | Sergey discusses office culture limits on voice chat. Jason suggests ad-supported free tiers, but Sergey explains Google's generational model pricing strategy. | |
| Announcement for the Upcoming All In Summit in Los Angeles | 0 | 0 | 0 | 0 | Housekeeping and event announcement for the All In Summit in Los Angeles by Jason. No guest participation. |
Statements from this episode (20)
Sergey Brin: AI is the most exciting technology of my lifetime
“Being a computer scientist, it is, you know, the most exciting thing, you know, of my life, just technologically.”
Calacanis: Google has 5 or 6 products with over 2B users each
“Now you have over two billion people using, I think, six products or five products have over two billion.”
Brin: AI advances so fast that monthly progress is noticeable
“These AI systems actually change quite a lot, you know, the, like, if you went away somewhere for a month and you came back, you'd be like, whoa, what happened?”
Brin says returning code contributions were just minor access-related changelists
“The code I submitted wasn't very exciting. I think I needed to, like, add myself to get access to some things and, you know, a minor CL here or there.”
Brin returned to Google with no real executive responsibilities
“Feels privileged to be able to. Kind of go back to the company, not have. Any real executive responsibilities, but be able to actually go deep into every little pocket.”
Brin: Post-training and thinking models are a huge, uncapped step forward
“And more recently, the post-training, especially as the thinking models have come around. And that's been, you know, another huge step up in general in AI. So you know, we don't really know what the ceiling is.”
Sergey Brin: AI models can win math and coding contests against top humans
“Like, they, you know, can win, like, math contests and coding contests, things like that against, you know, some top humans.”
Brin: Google sold robotics startups because AI software lagged behind hardware
“I mean, I think we've you know, acquired and later sold like five or so robotics companies and you know, Boston being one of them. I guess if I look back on it, we built the hardware. We also had this more recently. We built out everyday robotics internally, a…”
Brin is a skeptic of humanoid robots despite industry hype
“I'm probably the one weirdo who doesn't, who's not a big fan of humanoids, but maybe I'm jaded because we've, you know, we at least acquired at least two humanoid robotic startups and later sold them. But the reason is, I mean, the reason people want to do hum…”
Brin: Google internal policy briefly banned engineers from using Gemini
“Okay, I, like recently, I just had a big tiff inside the company, because we have this list of what you're allowed to use to code, and what you're not allowed to use to code, and Gemini was on the null list.”
Brin: Google is internally testing third-party AI coding tools like Cursor
“No, we're trying to, you know, roll out every possible kind of AI and trying external ones, you know, be whatever the cursors of the world, all, all of those to just see what really makes people more productive.”
Brin: Specialized AI learnings inevitably get merged into general foundation models
“We are generally able to, after we do that, take those learnings and basically put that capability into a general model. So there's not that much benefit. You know, you could, you can get away with a somewhat smaller specialized model, a little bit faster, a l…”
Brin admits DeepSeek's release closed the gap with proprietary AI models
“Deep Seek released a really surprisingly powerful model when it was January or so. So that, that definitely closed the gap to proprietary models.”
Brin admits the technology was simply not ready for Google Glass
“There are a bunch of things I wish I had done differently, but honestly, it was just like the technology wasn't ready for Google Glass.”
Brin: I rely on AI for math and coding over my skills
“With my skill of math and coding, I feel like I'm better off just turning to the AI now.”
Brin claims management is the easiest business function to automate with AI
“Management is like the easiest thing to do with AI.”
Brin: Google mostly uses custom TPUs over GPUs for Gemini
“Well, we mostly, for Gemini, we mostly use our own TPUs.”
Brin: Open-plan offices make AI voice mode hard to use at work
“During work, I can't really use voice mode too much. I usually use it on the drive.”