Demystifying VLSI Technology: Exploring Its Future Possibilities

·1 hr 12 min·2 clips

Carver Mead argues analog neural-network hardware could reduce inference energy by a factor of ten.

1. Computer Science Channel presents “Demystifying VLSI Technology: Exploring Its Future Possibilities” through a UC San Diego roundtable on the VLSI revolution and its next 20-plus years. 2. Moderator Andrew Kong leads Carver Mead, Qualcomm executive John Smee, and technology investor Sanjay Jha in a discussion tied to Mead's Kyoto Prize recognition. 3. The central thesis is that semiconductor innovation must move beyond automatic process-node scaling toward energy-efficient architectures, physical experimentation, and system-level design. 4. Smee describes automotive chips designed for software changes over 10, 15, or 20 years and contrasts them with ambient IoT devices using extremely low power and energy harvesting. 5. Jha highlights CMOS RF, optical computing, gallium-nitride micro-LED displays, silicon-carbide motor control, and QuantumScape's solid-state battery work as future semiconductor applications. 6. Mead recalls 1970s Caltech research on efficient silicon use and says process-node competition caused designers to neglect architectural efficiency and alternative computing structures. 7. Mead discusses accidental single-electron behavior in Fairchild bipolar transistors and Ivan Sutherland's proposed superconducting flux-quantum logic operating near 4 Kelvin. 8. Mead says analog neural computation modeled on dendritic trees could use roughly one-tenth the energy of digital inference, while Sanjay Jha criticizes von Neumann fetch-and-store overhead. 9. The panel argues that MOSIS-style prototyping could shorten learning cycles from roughly six months for a 1-nanometer or 3-nanometer wafer to about two weeks. 10. Best for semiconductor students, chip designers, and researchers in emerging computing architectures.

As heard by us

A UC San Diego roundtable on why VLSI's system-level challenges may matter to AI's future.

VLSI occupies the quiet, load-bearing side of technology, and this UC San Diego roundtable looks at how that work may connect to future AI systems.

Read the full review in PlayNext →

Why you'd press play

You get a front-row seat to the semiconductor dungeon where AI's future starts sweating over power, density, and who gets paid for the hard work.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Computer Science Channel