Page 9 - EE Times Europe Magazine – November 2023

EE|Times EUROPE 9

CTO INTERVIEWS
Intel’s Greg Lavender: ‘We’re Going to

Democratize AI’

By Pat Brans

ntel technology chief and senior vice president Greg Lavender has two broad the web browser and then the cloud. I would
areas of responsibility. As CTO, he drives the company’s future technical say my DNA is in networking. But I’ve always
innovation and research program. And as general manager of its software maintained an intellectual interest in AI, even
when I was a professor.
Iand advanced technology group, he is responsible for defining a singular But what’s really changed? I think what’s
artificial-intelligence software stack to support Intel’s range of business and been obvious is all the investment over the
last 40 years in compute, networking, storage
hardware offerings. and so on. Scale has enabled a new kind of
Lavender began his career as a professor of computer science at the AI to happen now because you need massive
University of Texas at Austin. During the 14 years he held that position, he was amounts of compute, massive amounts of
bandwidth, massive amounts of storage and
also involved in starting several companies, which meant moving back and good-quality data to do all this training and
forth between abstract and concrete worlds. then inferencing. I think that’s the natural
His current gig is not the first time Intel has played a big role in his life. culmination of the last 40 years of investment.
That started in the 1970s, when he was still in his early teens. Now, after more EETE: How would you describe the
than 35 years of working on software and hardware product engineering and change in the industry over the last 10
or 15 years with respect to new types of
advanced research and development, Lavender considers it his mission to “put application areas for integrated circuits?
the mojo back into Intel.” Lavender: I read that the demand in the
semiconductor industry will grow to a trillion
EE TIMES EUROPE: You’ve had a very futuristic weapons concepts] were on every- dollars over the next decade. What’s driving
interesting career so far. When you body’s minds around that time. that? The opinion of some authors is that
began in the 1980s, did you have any I took some courses on natural-language there are four big forces in play.
idea where things would be today? processing. At the time, I was intellectually The first is in automotive. The transfor-
Greg Lavender: It goes back further than curious about AI, but I couldn’t see a career mation of the entire industry to EVs is one
when I began working. My father was a com- path in it. So I chose computer network- part of that. Then there are smart vehicles,
puter scientist starting in the ’50s. He taught ing—and that was wise, because over the last sophisticated infotainment systems in the
me to program in FORTRAN when I was 40 years of my career, it has been essential. vehicles and autonomous driving. You will
14 years old. And I grew up with a Heathkit Everything since then was built on the foun- need a network of sensors and sophisticated
computer that I built and that ran an dation of computer networking—starting with graphics. You will need cloud computing to
8085 processor. So I’ve been programming the internet itself, then e-commerce, through help analyze all the data for the maps and
Intel processors since high school, starting
with the 8085 and then continuing with all
the successive generations.
But I would say the big inflection point for
me was my first job out of school. I was an
inexpensive engineer who knew the
x86 assembler from having programmed
the 8085 and 86. I got a job working for the
defense contractor TRW and got stuck on an
advanced research project. That turned out to
be the classified ARPANET.
So my first job out of school in 1983 was
implementing network protocols like TCP/IP,
X.25, FTP and TELNET. It wasn’t considered
real computing. It was considered a research
project, and I was part of a small team.
Needless to say, that experience had a great
influence on me.
I went to grad school, and I debated about
whether to do AI or to do computer net-
working. If you remember, [Japan’s] Fifth
Generation AI project and Star Wars [Ronald
Reagan’s proposed missile defense initiative,
which depended on a wide array of The industry “wants an alternative to Nvidia,” says Lavender.

www.eetimes.eu | NOVEMBER 2023