AMD CEO Lisa Su on 10 Years as CEO and AI's Integrated Future

Lisa Su
2 followers
I speak with Lisa Su, fresh off the stage at South By Southwest 2024, about AI PCs and the further growth potential in the AI market.
#lisasu; #influencer; #technology
Lisa Su, thank you so much for having us here on Overtime here at the AMD campus in Austin.
We're just talking about Oracle's report.
Of course, we need to be careful that they always give their guidance on the call.
But so much of what they're reporting is coming out of data center,
which has been just a huge growth area for you.
Today at South by Southwest, though, you were talking also about the AIPC,
about the client side.
So hoping to start there.
Yeah, absolutely.
So John, thank you for being in Austin.
We're very happy to host you on our campus today.
Look, it's been a great day.
I really enjoyed being at South by Southwest.
It seems like everyone is in Austin this week, all the movers and shakers.
And yeah, what we talked about today at the keynote was really a bunch of different things.
I mean, first of all, AI is really still at the very beginnings of what we can do in technology.
We can talk a lot about the data center.
I usually start there.
But I did spend some time talking about AIPCs.
And I'm very excited about this continuum of AI products as you go from cloud to edge to PC
and really trying to get the user experience such that everybody has AI somewhere in their lives.
Help me with my skepticism here, though, on the AIPC itself,
because it seems like this latest boom in interest in AI and some of the economic activity
came out of OpenAI's chat GPT release.
Investors really started to realize, OK, this is going to be big.
See it happening in the data center.
But when we start talking about inferencing and the client side, I don't see my kids saying,
oh, I need a new AIPC to run this new AI version of Fortnite, right?
They're still asking for the graphics cards, AMD, where AMD plays,
to do the same kind of things before.
So do we need to see new software that's putting demands
on the hardware in a new way before the AIPC gets momentum, if it does?
Yeah, well, let me take a step back, John, and kind of tell you,
describe a little bit of how I see the market evolving.
So, true, like, you know, chat GPT and all of these large language models
showed us that you can get access to a tremendous amount of capability
with all of the training and inferencing that you do in the cloud.
But it turns out that, you know, people really have a lot of personal data.
And the way you want, you know, you use your PC, it's actually a personal productivity tool.
So what I see is that we're at the beginning of the era
where we can make much, much more capable personal assistants in the PC form factor.
And things like, you know, I can ask my PC,
when was the last time I saw John Fort?
What did we talk about?
Remind me what he was interested in.
Those types of things, which today I'd have to spend, you know,
probably an hour looking through, like, all kinds of things.
That's one example.
Actually, on the South by Southwest stage today,
we showed a really good example of, you know, doing, you know,
text to image where you could do it locally.
So you don't have to connect to the cloud.
You can do it right in your own, you know, PC.
And again, these are things that as the technology gets better,
I am absolutely sure that everyone's going to want an AI PC.
Also on stage, you're joined by Weta, the really high end special effects creators.
Does the AI PC, to your point about text to image,
really catch on with professionals first and their iteration ideas,
and then that filter through the way it has
with so many other graphics powered capabilities?
I do think that there is a thing about expert users will start first.
So we actually had David Connolly on stage from WetaFX.
Actually, big congratulations to them.
They won an Oscar last night for War Is Over.
That was entirely rendered on AMD technology.
So we were very excited about that.
But what you see is, you know, again, for whether you're a Hollywood, you know,
movie studio or you're a content creator,
or you're just a, you know, sort of a regular person who has lots and lots of data,
AI can help you sift through that, can help you use it in a more productive way,
can take things that used to take, you know, days or hours,
and you can reduce it to, you know, minutes in terms of finding things,
whether it's text to image or now you saw, you know, text to video.
It's really an opportunity to just become much, much more productive.
And, you know, the technology that we're putting in today's PCs,
you know, I'd like to say that we've had CPUs, GPUs.
Now we have, you know, AI accelerators that are there
just to make it much more powerful in very small form factors.
So now we've talked a bit about the PC.
Let's broaden out because a lot of the analyst world
has been abuzz for the past several weeks
about this $400 billion total addressable market
for AI-related technology that you laid out.
Again, my job to approach these things skeptically,
we just went through this pandemic period
where we saw this huge surge in demand for PCs,
for e-commerce, for goods, and a lot of companies said,
oh, this is going to continue up and to the right.
And it didn't.
So how can we be sure this is an exception?
So, John, the way I think about AI,
and, you know, you've heard this from others,
is, you know, AI is actually, you know,
the most important technology that we've seen
over the last 50 years.
You know, it's arguably the most important.
And when you think about what it can do for enterprises,
for, you know, all of us, for, you know, research,
for development, all of that capability,
you need more compute.
Now, most people would say they've been really surprised
at just how AI has taken off over the last 12 or 14 months.
It is true.
We've taken a technology that has been around for a while,
but was frankly very difficult to use,
and we've turned it into something
where everybody on the planet can use AI technology,
you know, because it's language-based,
and you don't have to be a programmer,
you don't have to be an expert,
you can just ask the AI what to do.
Now, the truth is, AI is still in its infancy,
so we're still at a place
where we know that it will get better.
Like, we know that you can make these models better.
We know that you can make this inference faster.
We know that we can make it more accessible.
So I do see a place
where we're going to see tremendous growth
over the next, you know, three, four, five years.
We are super excited about the partnerships
that we have in AI, in the cloud,
and, you know, really trying to deliver
all of that infrastructure,
as well as in the edge and in the client form factors.
Speaking of partnerships, one of those is Microsoft,
which has been using AMD chips in the cloud.
I like to look at the calendar.
We marked 10 years since Satya Nadella
was named CEO of Microsoft back in February.
It's going to be 10 years
since you were named CEO of AMD in October.
And, you know, I was also looking at the stock chart.
We were around three bucks a share on AMD
when you came in, just shy of 200 today.
It's pretty good.
Also to note about those two 10 years,
data center has been key.
You know, Satya Nadella's Azure and work there
that transitioned from server and tools
into that, into cloud.
And then your work, I think when you came in,
90% of AMD revenues were in PC.
In fiscal 23, less than half, okay?
So why was data center for both of these companies
in retrospect so important over the last decade?
Well, I will say, you know, it's been a great 10 years.
It feels like it's not been that long.
But one of the things that's most important
with every company is to decide what you're best at.
And you're absolutely right.
AMD has always been a very good company.
But frankly, we needed to decide what we were best at.
And, you know, back then,
the bet was high-performance computing.
When it may not have been the sexiest part of the business,
I fundamentally believe that high-performance computing
was going to be the most important technology trend
for us going forward for the next five to 10 years.
I think we've seen that.
You know, we've seen data center
become much, much more important.
You've seen cloud become more important.
It's really difficult to build these chips.
You know, we talk about, you know,
new Zen 4 and Zen 5 cores.
We talk about chiplet technologies.
We talk about, you know, our MI300,
our latest AI chip has 150 billion transistors.
So these are hard technologies.
But once you put them together,
you realize how much they really drive,
you know, sort of computing overall.
So yes, we love PCs.
You know, I can talk to you all day about that.
But we also see a world where high-performance technology
really is the most important ingredient
for solving some of the world's most important problems,
and frankly, working with
some of the most important partners out there.
And so I'm very, you know, pleased with,
you know, the progress that we've made.
And data center is a place where,
you know, you need the biggest iron,
and it's hard to build, you know, these systems.
So since you frame it that way, I have to ask you,
what is your least popular current important bet?
Since everybody else is caught up
on high-performance computing.
Well, the beauty of our portfolio now is,
it's a very broad portfolio.
You know, we are in the largest data centers.
We are in the client devices.
We also have a very broad embedded portfolio.
So, you know, I'm super happy
with the acquisition of Xilinx.
We just passed the two-year anniversary
where we brought in an incredible embedded portfolio,
6,000 new customers.
We're in the Mars Rover.
I didn't say it's my least favorite.
We're in the Mars Rover.
We're in lots of cars.
That's not your least favorite.
I'm saying your most underestimated piece
of your portfolio strategically.
I think the most underestimated piece, John,
is how all of these pieces come together.
So it is not about just what you do in the data center,
or just what you do at the edge,
or what you do in the PC.
It's the fact that we are one of the very few companies
that have all of the pieces.
So we have all of the compute engines
that you need to optimize for your application.
And I don't know that people really understand that.
They don't understand necessarily yet
how cloud influences client.
That's going to become a lot clearer
over the next several years.
I'm very excited by that, by the way,
because when you think about where technology is going,
as much as the largest data centers in the world
are filled with these high-performance computing chips,
you realize that there are so many other places.
And if we can get these systems
really integrated synergistically,
and that's why when you ask me,
you're a skeptic about AI PCs,
I would love to talk to you a year from now
and see how you feel about it.
Because I feel really that this is probably
one of the most important inflection points
for the PC form factor.
And this is how people are going to experience AI
at a personal level.
You talk about vertical integration,
makes me think of Supermicro,
which is a company that you've known well for a long time,
but they were like a specialty muscle car maker.
And now, in part because of AI,
they've gone mainstream.
And a lot of people are wondering,
is this high-performance computing wave
changing the overall demand,
mass market demand for hardware,
where performance and that kind of integration
that they're doing,
and it sounds like you're talking about for AMD
in a different plane,
is just going to be mainstream?
Well, we love the partnerships that we have.
Supermicro is a great partner.
Dell is a great partner.
Lenovo, HP Enterprise.
We work with all of these guys.
But I think what's different with computing today
is there is a need for lots of different types of solutions.
So as you think about the hardware solutions,
it's not just one size fits all.
And the fact is,
people are willing to invest ahead of the curve in AI.
When you looked four or five years ago,
yes, infrastructure was important,
but now AI is essential.
Every boardroom is talking about how you use AI.
Every nation state is talking about how you use AI.
And so people are willing to invest ahead of the curve.
And our job is, frankly,
to make sure that there's more supply out there,
the opportunity to work closely with our partners
to give you lots of different solutions
depending on what you're trying to do.
And yeah, I think it's a good time to be in computing.
Well, you're having a good time here in Austin.
So thanks for having me here on your campus here
to talk about it, Lisa.
I appreciate it.
Thanks so much, John.
