physical AI

Jensen Huang describes how NVIDIA survived an early technology failure by confronting reality, learning quickly, and changing direction. He argues that NVIDIA’s enduring insight was to accelerate important algorithmic domains, rather than focus only on building chips. He explains how first-principles thinking and systems design guide the company’s work on AI, agents, robotics, and autonomous vehicles. His advice to founders is to stay curious, learn what matters, and build resilience by taking hard problems one day at a time.

Eric Landau describes how his work in particle physics and quantitative trading shaped his view that AI progress comes from scaling data and compute. He explains how Encord’s product and the market gradually converged, and why the company now focuses on data infrastructure for physical AI across robotics, autonomous vehicles, logistics, and manufacturing. The talk covers collecting, curating, annotating, and evaluating multimodal data at scale, along with Encord’s growth, customers, and competition. Landau closes with advice for founders: talk to customers and peers, learn from mistakes, and accept the emotional ups and downs of building a company.