Deep Future Jensen's Law
Apr 21, 2025
Explore the intriguing concept of Computational Maximalism and its role in AI advancements. Discover why Moore's Law is considered obsolete and the limitations of lithography. Dive into the rise of extreme ultraviolet technology and how innovative transistor architectures, like FinFET, push performance boundaries. Learn about shared memory and data-center-scale computing that enhances parallel processing for AI. Finally, understand Jensen's Law, highlighting how NVIDIA's GPUs enable massive AI systems and the leap towards zetaflop-class computing.
AI Snips
Chapters
Books
Transcript
Episode notes
Setting The Scene With Personal Framing
- Pablos recounts industry roles like 'hacker magician' and 'man of the future' to set tone for the episode.
- He frames the conversation around Moore's Law's 60th anniversary to introduce chip history.
Why Moore's Law Was Rooted In Lithography
- Moore's Law described as transistor density doubling roughly every two years, driven by lithography advances.
- Pablos emphasizes that the process relied on progressively finer ultraviolet light to etch transistors onto silicon.
EUV Is The Most Complex Semiconductor Tech
- Extreme ultraviolet (13.5 nm) lithography became the necessary breakthrough after deep UV reached limits.
- The EUV systems are extraordinarily complex and costly, made primarily by ASML with massive global supply chains.
