Poddavsnitt

Greg Brockman, co-founder and president of OpenAI, discusses the release of GPT-6 and recent breakthroughs in mathematical problem-solving. He argues that AI is now capable of generating genuinely new knowledge and predicts a shift where cybersecurity spending will vastly exceed software spending due to changing technical physics.

Misha Laska, co-founder and CEO of Reflection AI, discusses the company's mission to build frontier open-weight intelligence. He details the operational challenges of scaling a research lab to 300 people and the strategic shift from focusing solely on reinforcement learning to building end-to-end open models to ensure geopolitical and technical independence.

Walter Goodwin, founder and CEO of Fractile, discusses the company's mission to build high-speed inference chips for large-scale models. He explains the "full-stack" approach to silicon design, the current landscape of AI ASICs, and the structural differences between internal hyperscaler efforts and independent chip companies.

Michael Lee, co-founder and CEO of Sequence Holdings, discusses the firm's strategy of acquiring incumbent businesses to refound them as AI-driven market leaders. He details the recent $7.7 billion take-private acquisition of Baldwin, supported by the Dell family office, and explains the rationale behind using a permanent holding company structure to align capital with long-term technological transformation.

Stefano Ermon, Stanford professor and co-founder of Inception, discusses his role in the development of diffusion models and his transition from academia to industry. He explains the technical advantages of diffusion-based language models over autoregressive models, specifically regarding inference speed and parallelization.

Brian Armstrong, CEO of Coinbase, discusses the evolution of the "everything exchange," the growth of stablecoin payments, and the emergence of "agentic finance" (AI-fi). He explains how Coinbase is enabling AI agents to have their own financial accounts and the necessity of crypto rails for micro-transactions in the agentic economy.

The speakers discuss how artificial intelligence is becoming a utility across industries, with a focus on its impact on semiconductor chip design at Arm. They highlight that verification and validation consume the bulk of design time and that AI can dramatically accelerate these steps, potentially shrinking design cycles from 24‑36 months to as short as six months. The conversation also covers supply‑chain constraints, capital intensity, and the long‑term outlook for AI‑driven chip development.