Stray Reflections Not an AI bubble
21 snips
Jan 16, 2026 Silicon's pivotal role in computing history takes center stage as the discussion delves into the limits of Dennard scaling. The podcast explores how performance now comes at a higher power cost, revealing why AR/VR once struggled with heating issues. The complexities of AI advancements come to light, emphasizing the need for power and infrastructure. The rise of capital as a limiting factor shifts the landscape, highlighting the importance of computing literacy for businesses. Finally, the conversation examines how AI's physical demands reshape technology's relationship with materials and energy.
AI Snips
Chapters
Transcript
End Of Dennard Scaling Reshaped Computing
- Dennard scaling ended around 2005, so transistor miniaturization no longer gave free speed, cost, and efficiency gains.
- The power wall forced computing progress to trade off performance for much higher energy and cooling costs.
AI Progress Is Capital And Energy Intensive
- AI advances now cost more energy and infrastructure with each capability leap.
- Training large models requires enormous compute and capital to build, cool and maintain systems.
Make Compute Literacy A Core Skill
- Learn how silicon works and factor power, latency, and bandwidth into business decisions.
- Treat compute literacy as core business knowledge, not just software fluency.