Thumbnail for Grok 4 Wows, The Bitter Lesson, Elon’s Third Party, AI Browsers, SCOTUS backs POTUS on RIFs

Grok 4 Wows, The Bitter Lesson, Elon’s Third Party, AI Browsers, SCOTUS backs POTUS on RIFs


Episode Details
Channel

All-In Podcast

Published

7/11/2025

Episode Summary

In a detailed discussion featuring guests Travis Kalanick and Keith Rabois, the panel explores pivotal trends in technology and politics. Travis Kalanick, co-founder of Uber, elaborates on his current venture, Cloud Kitchens, which is building infrastructure for the future of food delivery. He outlines a vision for `autonomous burritos` driven by `end-to-end automation`, exemplified by his `bowl builder` robot that slashes labor costs. He also hints at potential involvement with autonomous driving company Pony AI to create new market options. The conversation shifts to the AI arms race, dominated by Elon Musk's company, `xAI`, and its new model, `Grok 4`. Trained on the `Colossus` supercomputer, Grok 4 has reportedly benchmarked higher than rivals from `OpenAI` and Google's `Gemini`. Chamath Palihapitiya frames this success using `The Bitter Lesson`, an essay by Rich Sutton arguing that scalable computation wins over systems dependent on `human-labeled data`. This principle is connected to the strategies of Tesla's `Tesla FSD` and the move towards training `LLMs` on `synthetic data`. This fuels a debate on the `short halflife on human label data` as an investment thesis. The panel also analyzes the emergence of the `agentic browser`, discussing `Perplexity`'s strategy and suggesting its best opportunity may be to challenge `Bloomberg`. On the political front, the hosts, including Jason Calacanis, discuss Elon Musk's proposed `The American Party`. They analyze its potential strategy of winning a few key seats to gain leverage, the role of the Senate `filibuster`, and how recent `Federal Elections Commission` rule changes empowering `super PACs` could enable such a movement. This is followed by an analysis of a landmark `Supreme Court` ruling on `federal workforce reduction`, which affirmed executive power in a case brought by the `AFGE` union. The episode, where hosts David Sacks and David Friedberg were absent, concludes with Travis Kalanick revealing his acquisition of the backgammon engine `XG`, which he plans to enhance with AI.

Key Topics & People

Co-host of the All-In Podcast who interviewed Senator John Fetterman on various political and economic topics.

OpenAI
OpenAI
Organization

Leading AI research lab and creator of ChatGPT, mentioned regarding open-source models and AI scaling.

Tesla
Organization

Electric vehicle and AI company building self-driving tech and humanoid robotics, praised by Kalanick as the 'Google' of physical AI.

Uber
Uber
Organization

Ride-sharing company founded by Travis Kalanick, referenced for its disruption of transportation and strategic use of capital.

Cloud Kitchens
Organization

The initial food-focused business unit of Atoms, operating globally under various localized names.

Elon Musk
Elon Musk
Person

CEO of Tesla and Boring Company, noted for moving his businesses out of California to Texas due to a better operating environment.

Podcast host interviewing Travis Kalanick and Michael Dell live in Austin.

Founder of Uber and Atoms, officially exiting stealth mode to discuss his physical AI and automation startup.

A host of the All-In Podcast who provides analysis on the SaaS market, arguing that AI is creating a new value layer on top of existing SaaS, rather than making it obsolete.

CEO of Nvidia, heavily investing in the AI scaling and foundational model companies.

Co-host of the All-In Podcast participating in the capital markets discussion.

Gemini
Technology

Google's AI model suite, permitted for lawful use cases within classified military networks.

Bloomberg
Organization

A major financial software and media company where CZ worked in New York for four years, managing a team of developers working on trading systems.

xAI
Organization

An artificial intelligence company founded by Elon Musk. It merged with SpaceX to create a combined entity valued at $1.25 trillion, aiming to leverage space infrastructure for AI development.

Pony AI
Organization

A company involved in the self-driving technology sector, mentioned as a participant in the global race for autonomous mobility.

LLMs
Technology

Large Language Models, the technology underlying many recent AI advancements like ChatGPT. The foundational paper for the Transformer model, which enables LLMs, is mentioned as being only from 2017, highlighting the rapid pace of change.

Supreme Court
PoliticalEntity

The highest federal court in the United States, which ruled 5-4 in favor of the federal government's ability to remove razor wire installed by Texas at the border.

Llama
Llama
Technology

An open-source large language model developed by Meta, seen as a key force in decaying the market value of proprietary closed models.

A venture capitalist who publicly debated David Sacks on X (formerly Twitter) regarding the merits and risks of the TikTok ban legislation.

Perplexity
Organization

An AI-powered search engine company whose model of providing direct answers with citations has been adopted by Google, potentially disrupting its business.

Colossus
Colossus
Technology

The name of xAI's AI training supercomputer, located in Memphis, designed to be one of the largest and most powerful in the world.

bowl builder
Technology

A robotic machine developed by Cloud Kitchens that automates the assembly of bowl-based meals, handling ingredient dispensing, mixing, and packaging for delivery.

Grok 4
Technology

An advanced large language model from xAI that has shown significant performance improvements, positioning it as a top competitor to OpenAI's models.

The Bitter Lesson
ScientificConcept

A principle in AI research, articulated by Rich Sutton, which holds that general-purpose methods leveraging massive computational power will ultimately outperform more specialized, human-curated approaches.

Political action committees that can raise unlimited sums of money from corporations, unions, and individuals but are not permitted to contribute to or coordinate directly with parties or candidates. Their influence is debated on the podcast.

DoorDash
DoorDash
Organization

An American company that operates a food ordering and delivery platform, mentioned as a key part of the logistics ecosystem for automated kitchens.

XG
Technology

eXtreme Gammon, a top-tier backgammon software engine purchased by Travis Kalanick, who plans to apply modern machine learning techniques to advance the game's analysis and training capabilities.

The U.S. regulatory agency for campaign finance. Its 2023 rule change allowing Super PACs to fund ground operations is identified as a significant development enabling new political strategies.

A procedural rule in the U.S. Senate requiring a supermajority of 60 votes to end debate on most bills. It is discussed as a major obstacle to legislative action that is on 'borrowed time'.

AFGE
Organization

A large union representing federal government employees that sued the Trump administration to block an executive order aimed at downsizing the federal workforce, leading to a Supreme Court case.

A major political and legal issue concerning the President's authority to downsize the federal government. A recent Supreme Court ruling supported the executive's power to plan for such changes.

The American Party
PoliticalEntity

A new political party proposed by Elon Musk, intended to operate outside the traditional two-party system with a platform focused on fiscal responsibility, technological excellence, and pronatalism.

agentic browser
Technology

A new product category where a web browser is enhanced with AI agents capable of autonomously performing complex tasks for the user across multiple web services, such as booking flights or managing emails.

An investment thesis discussed by the panel, suggesting that the value of companies providing human data labeling services will decline rapidly as AI models become capable of effective self-labeling and learning.

synthetic data
Technology

Artificially generated data used to train AI models, seen as the next frontier for AI development once the existing corpus of human knowledge has been exhausted. It allows for continuous self-improvement.

Tesla FSD
Technology

Tesla's advanced driver-assistance system that aims for full self-driving capability, exemplifying 'The Bitter Lesson' by using a camera-based approach to collect vast amounts of data for a general learning system.

An approach to AI development that relies heavily on human input, such as manually labeling data or coding specific rules. 'The Bitter Lesson' argues this approach is ultimately inferior to scalable computation.

A pioneering computer scientist in reinforcement learning and the author of the influential 2019 essay 'The Bitter Lesson'.

A key business strategy principle discussed in the context of food robotics, which posits that for automation to be truly effective, it must cover the entire process from ingredient input to final packaging without creating new manual bottlenecks.

A concept championed by Travis Kalanick representing the vision of end-to-end automation in the food industry, combining robotic kitchens with autonomous delivery vehicles to dramatically lower costs.