scripod.com
Chips, Memory, and Power | Pat Gelsinger

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Chips, Memory, and Power | Pat Gelsinger

The a16z Show

4 HOURS AGO
The a16z Show

The a16z Show

4 HOURS AGO
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

a16z's Raghu Raghuram and Guido Appenzeller sit down with Playground Global General Partner and former Intel CEO Pat Gelsinger to discuss the next wave of semiconductor innovation and the physical constraints shaping the AI buildout. Drawing on his experi...

Highlights

Pat Gelsinger joins Raghu Raghuram and Guido Appenzeller to examine the physical limits of AI infrastructure, from chip design and memory to networking, power, and agent-oriented computing.
00:00
Energy capacity equals economic capacity
08:58
AI design is only the beginning
16:09
Extreme specialization may not survive changing workloads
24:47
AI is a memory workload
36:31
Optics becomes necessary at scale
46:02
Energy is an upper bound on AI growth
51:13
Humans are more patient than agents

Chapters

From Intel’s early tools to the physical limits of AI
00:00
Automation accelerates design, but bottlenecks move downstream
07:17
Memory innovation and the economics of heterogeneous AI
22:27
Optics, cluster architecture, and the energy ceiling
38:40
Changing computing physics and virtualizing agents
48:06

Transcript

Pat Gelsinger: In an AI digital age, energy capacity equals economic capacity. Why build a new data center and buy a million GPUs if I can't power that? You're going to see more and more defaults happening on many of those data center projects because the ...