Jan 31, 2025 · 1h 49m · allin
DeepSeek Panic, US vs China, OpenAI $40B?, and Doge Delivers with Travis Kalanick and David Sacks
Travis Kalanick · 29m spoken Chamath Palihapitiya · 21m spoken Jason Calacanis · 18m spoken David Friedberg · 15m spoken David Sacks · 13m 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
In this episode of the All-In Podcast, the hosts and guest Travis Kalanick analyze Chinese AI startup DeepSeek's market disruption, OpenAI's funding dynamics, federal government efficiency efforts under DOGE, and the future of food robotics and autonomous transport.
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 69% of the talking time here. How this is scored →
The hosts as informed peer 5.2 Guest teaching 4.8 Guest disagreement 2.2 The hosts pushing back 2.6
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 39:40 JCal attacks OpenAI hypocrisy
Jason Calacanis forcefully rejects Sam Altman's complaints about China distilling O1 by highlighting that OpenAI originally scraped public internet data without permission.
Hardest push from the hosts ▶ 33:20 JCal challenges Microsoft's loyalty to OpenAIJason Calacanis refutes the notion that Microsoft is a faithful partner by pointing out they host DeepSeek R1 to subvert OpenAI at lower costs.
Biggest teaching moment ▶ 1:28:30 Travis educates hosts on power grid constraints for AVsTravis Kalanick surprises the hosts with back-of-the-envelope calculations proving full EV ride-hailing in California would require doubling the state's entire electrical grid capacity.
The host holds their own ▶ 27:20 Chamath demonstrates technical mastery over GRPO and PTX assemblyChamath Palihapitiya displays superior technical knowledge by detailing how DeepSeek bypassed Nvidia CUDA lock-in using PTX bare-metal assembly and custom GRPO reinforcement learning.
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 |
|---|---|---|---|---|---|---|
| Travis Kalanick on Cloud Kitchens & Food Automation | 3 | 5 | 1 | 1 | Travis Kalanick outlines Cloud Kitchens automation vision and robotic bowl builders while Chamath asks inquisitive questions about personalized dietary tracking. | |
| David Friedberg's Eatsa Retrospective & Historical Automats | 6 | 3 | 1 | 1 | David Friedberg details his historical experience building Eatsa bowl automation, while Travis adds context on QSR brick-and-mortar layout constraints. | |
| David Sacks on Life in D.C. as AI Tsar | 2 | 1 | 1 | 1 | David Sacks shares lighthearted personal anecdotes about his new role in the White House complex and visiting the Navy Mess. | |
| DeepSeek-R1 Announcement and Market Fallout | 3 | 6 | 2 | 2 | Sacks debunks exaggerated media claims around DeepSeek's six million dollar cost by distinguishing final training run costs from overall compute cluster investments. | |
| DeepSeek's Innovations, Algorithmic Constraints, and CUDA Bypassing | 7 | 4 | 2 | 3 | Chamath explains DeepSeek's algorithmic innovations like GRPO and low-level PTX assembly programming that bypassed Nvidia CUDA. | |
| Debating Distillation, Model Hosting, and OpenAI's Reactions | 5 | 3 | 4 | 5 | The group debates distillation claims while Jason Calacanis aggressively highlights OpenAI's hypocrisy regarding open-source data scraping. | |
| The Open-Source Impact on AI Business Models and Economic Value | 6 | 4 | 3 | 4 | Hosts and guests debate AI value capture across layers, citing Jevons paradox and whether frontier models retain competitive moats. | |
| Chinese Business Dynamics, Hardware Sanctions, and Silicon Innovation | 6 | 7 | 2 | 2 | Travis shares accounts of hyper-competitive Chinese business culture and food delivery logistics, while Chamath provides Singapore GPU export data. | |
| OpenAI's $40B Capital Raise and AI Investment Dynamics | 5 | 8 | 3 | 3 | Travis provides insider perspective on Masa Son's aggressive investment tactics and warns against the risks of overcapitalization in tech startups. | |
| DOGE Strategy, Federal Spending Cuts, and Public Sentiment | 6 | 4 | 2 | 3 | David Friedberg outlines Ray Dalio's deficit framework and treasury yield risks while the host team evaluates DOGE federal efficiency measures. | |
| Autonomous Vehicle Commercialization, Power Grid Constraints, and Urban Real Estate | 6 | 8 | 3 | 3 | Travis reveals a major hidden bottleneck in autonomous vehicle expansion by showing California's power grid would need to double to support full ride-sharing electrification. | |
| Federal Reserve Rate Hold, Treasury Yield Spiral, and Spending Reductions | 7 | 4 | 2 | 3 | Chamath analyzes short-term Treasury paper refinancing risks and reads technical safety feedback from a commercial pilot regarding DCA airspace. |
Statements from this episode (31)
Travis Kalanick predicts people will only cook as a hobby
“Look people will cook in the future as a hobby.”
CloudKitchens will roll out automated bowl builders to five customers in April
“We're rolling out with five customers in April that are using the machine”
Travis Kalanick reveals CloudKitchens operates tens of thousands of delivery kitchens globally
“We, of course we have the real estate, so we have kitchens, you know, tens of thousands of kitchens around the world.”
David Friedberg says Eatsa had a term sheet with Chipotle in 2016
“We actually had a term sheet with Chipotle, This was nine years ago to actually put this into Chipotle stores.”
David Friedberg's Eatsa generated $3 million annually from an 800-square-foot store
“We built these Itza restaurants that were smaller footprint. We had an 800 square foot restaurant that was doing three million a year in revenue, and it had a handful of people working in it, but we were putting about 800 people an hour during the lunch rush t…”
David Sacks reveals his Eisenhower Building office and West Wing access
“I have an office in the old Executive Office building, otherwise known as the Eisenhower building, and then I have a pass where I can just walk over to the West Wing if I want to walk over to it.”
Jason Calacanis claims OpenAI got caught red-handed stealing everyone's content
“And then there's a bunch of debate on whether they stole, which is kind of rich coming from open AI, which got caught red-handed stealing everybody else's content.”
David Sacks estimates China's AI model gap narrowed to 3-6 months
“If you had asked most people in the industry a few weeks ago, how far behind is China on AI models, they would say six to 12 months. And now I think they might say something more like three to six months, right?”
David Sacks claims US frontier AI training runs cost tens of millions
“What the founder of Anthropic said, and what I think Brad has said, being an investor in OpenAI and having talked to them, is that the final training run cost was more in the tens of millions of dollars about nine or 10 months ago.”
Chamath Palihapitiya warns OpenAI's consumer strategy guarantees a collision with Meta
“Which puts them on a collision course with Meta. It's the only company that could really impact that because the only company right now that has Billions of eyeballs of DAUs per day.”
Chamath Palihapitiya says DeepSeek's chain-of-thought proves its model was distilled
“So the reason why this is happening is like, you're seeing this chain of thought, you're seeing the several layers, and then it's catching it after the fact. So we know that this is distilled from some other model.”
David Sacks notes DeepSeek V3 self-identified as ChatGPT-4
“A month ago, we had a press cycle in Silicon Valley when DeepSeq's V-three model came out, that DeepSeq V-three was self-identifying as ChatGPT. When you would ask it, who are you? Like, what model are you? Five out of eight times, V-three would tell you that …”
Jason Calacanis predicts all large language models will become open-source commodities
“Gavin Baker came on this podcast and said it's the fastest deprecating asset in the world was a large language model. He's been proven right. They're not worth anything. They're all going to be open source. They're all going to be commoditized, and that's for …”
Travis Kalanick expects cheaper AI to paradoxically increase total industry revenue
“When AI gets cheap, you know what's going to happen, guys? There's going to be a lot more AI, right? I don't think, I think the price elasticity on this one is actually positive. So as the price goes down, the revenue usage, everything's going to go up.”
David Sacks argues DeepSeek-R1 only matches a four-month-old OpenAI model
“The R-one model is, is basically comparable to O-one, which OpenAI released four months ago and was training on internally, call it nine or 10 months ago. So OpenAI is on O-three now. Its frontier is ahead of where R-one is.”
Travis Kalanick says mastering copying eventually forces a flip to genuine innovation
“When you get really, really good at copying and that time gets tighter and tighter and tighter and tighter and tighter, you eventually run out of things to copy. And then it flips to creativity and innovation.”
Travis Kalanick claims US food delivery apps are copying old Chinese features
“A lot of the innovation you will see coming out on Uber Eats or DoorDash Like the stuff that's coming out now is stuff that existed three years ago, four years ago in China, maybe longer.”
Chamath Palihapitiya expects China to use AI to design simpler chip workarounds
“And if China is forced to engineer its way around it, yeah, Freberg, the answer to your question is they'll use these models to design chips that can be manufactured in simple ways, and they'll make simple stuff.”
Travis Kalanick reveals he rejected Masa Son's money for funding competitors
“I did not take money from Masa. So he begged me to take money for years, and we did not take it because he is a, he's a, what's the word I'm looking for? He's a promiscuous investor. So once he invests in you should probably count on him and using your informa…”
Travis Kalanick says Saudi Arabia invested $3.5B before the Vision Fund existed
“We did stuff with the Saudis, Before even Vision Fund existed, they stroked a three and a half billion dollar check when that was like the biggest thing that ever happened.”
David Friedberg argues specialized AI model networks will beat monolithic models
“It doesn't make sense to have one large do everything model. This mixture of experts architecture, ultimately you can kind of think about taking a large model, making two copies of it, and then trying to shrink each model down to whatever the necessary, so tha…”
David Friedberg predicts a wave of lawsuits will target DOGE spending cuts
“Over the next couple of months, a lot of lawsuits will fly.”
Chamath Palihapitiya predicts DOGE spending cuts could exceed $2 trillion
“I suspect to be honest, it could be more than two trillion dollars when it's all said and done.”
Travis Kalanick calculates a $500B surplus using 2019 spending and 2024 revenues
“2009, if we took 2019 spend, right, the year before COVID, and put it up against 20, 24 revenues, five hundred billion dollar surplus.”
Travis Kalanick says all-EV ride-sharing would require doubling California's energy capacity
“If all of the miles in California went EV ride sharing, you would need to double the energy capacity of California.”
Travis Kalanick warns expanding California's grid by 10% will take years
“Even getting 20% more, 10% more is going to be a gargantuan five to 10 year exercise.”
Travis Kalanick identifies utility upgrades as the primary commercial construction bottleneck
“In the U.S. And really around the world, utility upgrades is the long pole in the tent in, in construction development.”
David Friedberg warns nearly 30% of US federal debt matures in 2025
“Nearly 30% of the debt is gonna get refinanced this year.”
David Friedberg notes the latest US Treasury auction barely reached 2x coverage
“The last auction barely had two X coverage.”