$39B founder says his company could 100x in 5 years

Episode Summary

Executive Summary: Brett Adcock argues that AI and robotics are at an inflection point far bigger than the internet, with Figure focused on humanoid robots for real-world work and Hark building AI systems that use computers, remember context, and eventually function like a persistent Jarvis-style assistant. He frames the next five years as binary: massive scale or failure, and says product execution, talent, and autonomous capability are the core constraints.

Main Topics: Figure’s humanoid robotics thesis (Priority: 5/5): Adcock says Figure is building intelligent humanoid robots designed to do economically useful work in homes and work sites, with the long-term goal of general-purpose labor. Hark and AI-to-human symbiosis (Priority: 5/5): He describes Hark as an AI lab creating a digital companion that can use computers, remember everything, and interact like a superhuman assistant across devices and contexts. Computer use, multimodality, and memory as the next AI frontier (Priority: 5/5): A major theme is that useful AI must go beyond chat: it needs vision, speech, persistent memory, and the ability to operate software like a human through browsers and interfaces. Device and interface reset beyond phones and glasses (Priority: 4/5): Adcock believes current phones and computers are outdated for AI and that the industry needs radically new AI-native hardware, not just wearables or incremental form-factor changes. Talent scarcity and Meta’s recruiting strategy (Priority: 4/5): He says only a small number of people truly know how to build frontier AI models and argues Meta is buying talent effectively, though he prefers mission-driven teams over mercenaries. Founder mindset, binary outcomes, and obsession (Priority: 4/5): Adcock repeatedly emphasizes hard things, conviction, long hours, and all-in commitment, saying his companies are high-risk, high-upside bets that can become enormous or fail. Personal trade-offs and operating discipline (Priority: 3/5): He explains that he sacrificed most social life, focuses almost entirely on work and family, uses tools like Hark to manage his workflow, and relies on health monitoring to stay functional.

Key Arguments: AI progress is accelerating faster than the internet ever did, and deep learning is now working well enough to justify radical bets on both robotics and software agents. A useful AI system must be able to use computers autonomously because most real workflows still live in browsers, not APIs. Figure’s robots are already doing real customer work, and the bottleneck is increasingly intelligence and generalization, not manufacturing. Hark is designed to be a persistent personal AI companion with memory, vision, voice, and system access, not just a chatbot. The next major consumer device will likely be a new AI-native computing platform that replaces phones and computers rather than a minor wearable add-on. There are very few people who truly know how to build frontier AI; compensation is skyrocketing, and mission alignment matters more than money for long-term execution. Hard problems can be easier to pursue strategically because they face less competition, attract top talent, and can yield much larger outcomes. His companies are at a decisive stage where execution over the next one to two years will determine whether they scale dramatically or stall. He prioritizes building real products over hype and dismisses many current AI startups and devices as low-quality slop. His personal operating system is extreme focus: work, family, and almost nothing else.

Data Points: Fortune-estimated net worth: $19 billion - Mentioned at the start as the public valuation estimate for Brett Adcock's wealth. Veteri sale price: Over $100 million - He referenced his first company being sold for over 100 million dollars. Archer valuation: $40-$50 million range - The host suggested this was Adcock's approximate current valuation range three years ago. Hark team size: 80-90 people - He said Hark has roughly this many team members. Figure team size: Around 1,000 employees - He said Figure has about 1,000 people in the organization. Robotic sorting speed: 2.9 seconds per package - He cited a logistics task where Figure robots matched or exceeded a customer requirement of 3 seconds per package. Customer task duration requirement: 5 hours a day, 5 days a week - The package-sorting use case was framed as a production workload for a customer. Talent pool estimate for frontier AI: 20-30 people in California - Adcock estimated only a tiny number of people truly know how to build very good AI models. AI compensation packages: $15-$20 million stock offers; $36 million Meta offer - He described offers used to recruit senior AI talent, including a Meta package that beat Hark's equity value. Junior AI compensation: $750,000 to $2 million per year - He said younger AI researchers are seeing annual compensation in this range due to market bidding. School shootings increase: 10x over the past 10 years - He cited this statistic when discussing Cover's mission. Vision for AI-device market: 1 billion units per year - He said the core device category must be something that can scale to mega-device levels like phones or computers. Predicted full human speech Turing test: 2027 - He predicted AI voice systems will be indistinguishable from humans by then.

Pivotal Quotes: "I don't care about that at all. I give like zero shits about that." — Brett Adcock: His reaction to being told his estimated net worth had risen to $19 billion. "There is no flat line here. It's either like it goes down or it goes up, right? Either like it's binary." — Brett Adcock: He explained the risk profile of Figure and his other ventures as all-or-nothing. "The end state is BCI in the brain." — Brett Adcock: He described the long-term trajectory for AI-native interfaces beyond phones, glasses, and wearables.

Implications: The conversation signals a near-term race to build autonomous AI agents, humanoid labor, and post-phone hardware. For founders and investors, the winning teams may be small, obsessive, and execution-heavy, while many current AI products could quickly become obsolete.