How does advanced high-volume productive capacity add up over time to national power? It’s a subject not compassed by today’s electrical engineering departments, where students learn to theorize and design chips, memories, and other spectacularly advanced devices to be built on Si MOS, SiC, GAN, or with MEMS – but, crucially, not how to build companies to manufacture them.
One could, with only modest exaggeration, maintain that the Stanford University Department of Electrical Engineering did not merely participate in the creation of the modern semiconductor industry so much as preside over its birth.
How did Stanford and the Valley create the chip industry?
And why did we lose the bulk of it?
How does control over advanced manufacturing — and above all the silicon metal-oxide semiconductor — translate to national power?
Ten years before founding Hanover in 2018 and becoming a full-time venture capitalist, I moved to Silicon Valley to work in the semiconductor industry at Cypress, where we worked on power-efficient mixed-signal processors. One of the first such chips conducted the hazy, high-speed analog calculations used to measure whether a finger was touching the click wheel of a popular shiny white MP3 player.
Since then, I’ve been fortunate to count Ali Keshavarzi, noted R&D leader at TSMC, Intel, and Cypress — and an adjunct professor of EE at Stanford — as a venture partner at Hanover, where @allisonswope, me, and the rest of the team have been investing in critical hardware and software companies, including in core areas like AI codegen, sim and physics, energy conversion hardware, and battery manufacturing. And I have sat in the boardroom during these difficult scale-ups. Recognizing the very large gap between policymakers in D.C. and commercial leaders in the Valley, it became clear that an offering inside Stanford’s EE department was critical to provide soon-to-be chip designers with a way to understand the geopolitical battle they are soon to join. And real Silicon Valley chip veterans need a place to tell their war stories.
So we — and, truth be told, mainly Ali — created EE 292P. We decided to call it ‘ABNI’ — Atoms, Bits, and the National Interest.
We are teaching EE 292P this Winter quarter at Stanford. Here’s the listing.
The goal? To gather friends from the semiconductor industry, Stanford’s EE department (as well as key people from GSB, my own alma mater the Law School, and economics) to trace the build-up of American might from Shockley’s lab to the Traitorous Eight to the great inversion point somewhere around the 1990s.
EE students don’t typically get to think about this stuff. It’s time we brought some techno/economics to this all-important field around which so much of the world now revolves.
At Hanover we’ve been focused on building companies leveraging CMOS, SiC, GaN, and other processes–as well as world-changing applications built on them, like Zeromatter. But for an investor making only a few decadal bets — and as someone whose inclination is to be heavily involved in the company — I want a real at-bat in each investment. I can accept technical risk. What I cannot accept is an incorrect techno/economic judgment that moots my investment because of something that could have been understood by having good knowledge of prices and supply and roadmap and capacity and cost and quality. But figures in these categories are among the world’s most secret information.
If it’s batteries, do you invest in powder, anode, cell, pack, system? If it’s chemicals, do you invest in IP/discovery, compounds, CRO, CDMO? If it’s logic chips, do you design? Sell IP blocks? Chips or boards? How integrated?
These are techno/commercial questions that stem from a broader reality about innovation waves, production capacity, and national power. The U.S. continues to make the most important advancements. Since the 1990s, those advances mostly come back to us in the form of consumer surplus. The U.S. enjoys a “cutting edge” economy. We really have designed our economy to sit on a knife’s edge. As one storied semiconductor CEO said to me recently, “we are too high-tech in America.”
Whether that’s something to remedy is a question we’ll talk about. And how to fix it is another one. EE students play a critical role, but even at Stanford they are not often equipped or inclined to push for policy solutions or commercial forms that will work. And sometimes policy results are counterintuitive or unpredictable. (Our RF design students will feel right at home.) Their tech does not exist in a vacuum. EEs make it, but they get impacted by it too.
We’ll hear from leaders from TSMC, Applied Materials, ASML, Cypress. Purdue, Berkeley, Georgia Tech, Stanford’s departments of EE, CS, business, law, and economics. SambaNova and Google. DARPA and Intel. The Fed. We will hear from “real” VCs — the ones who competently evaluate complex technologies ab initio. And more.
We’ll also talk about the shadowy part of this world: prices, cost, hot lots, disti, and defects. Competition. What, for example, do the internals of SMIC, Huali, ASMC, CanSemi, and HyGon look like? What about Primarius, Hyperform, and Empyrean vs. Synopsis and Cadence?
This is an experimental course, to be sure. But I am convinced that today’s Stanford EE students can rapidly become part of a U.S.-led revitalization in the most important areas of high-volume advanced manufacturing.
Ali deserves immense credit for designing an information-dense cross-disciplinary course, and it’s been an honor to work with him. We will post some resources from the course as we go, so please keep checking in. Let’s build !
Joe Malchow
Portola Valley, Calif.