00:00:02
There's a company nearly every chipmaker relies on that doesn't actually make anything tangible. Yet, its Blockbuster
00:00:08
IPO in September valued it above 54 billion. It's the largest public offering in over
00:00:12
2 years. The valuation of the company has tripled. If you have a smartphone in your pocket
00:00:17
or in front of you, you have an ARM circuit somewhere inside of it. We are the CPU, the heart of everything.
00:00:23
They're the winner of the CPU side. the foundation models, the software, it's moving far faster than the hardware. So,
00:00:30
what we're seeing is people investing faster and faster into new hardware, which ends up being a good thing for us.
00:00:37
Ladies and gentlemen, please welcome ARM CEO Renee Hos. [Applause] [Music] Thank you so much.
00:00:51
How are you? Welcome. Welcome, David. Hey, good to see you. Hi. Hello, Renee. What are you banging these days?
00:00:57
3 milligrams of AL pouches or you're up to nine. I know you're competing with Nvidia, so
00:01:02
you probably want to go with the nine, right? I will go with the nine with Jensen. You
00:01:06
have to go big. You have to go big with Jensen. What's that like to compete against Nvidia?
00:01:12
Well, I will say uh Nvidia is a customer of ours. So, I'm not going to say Jensen
00:01:16
is my competitor uh today, but you know, I worked for Nvidia for for many many years as as you know. uh and he's
00:01:25
fantastic right and uh learned so much working there working for him working with him and then Nvidia you know almost
00:01:32
acquired ARM in 2020 uh so I almost you know had a chance to work with him again
00:01:37
what did you learn from Jensen you know one of the things about Jensen that is amazing I think it's also true
00:01:43
for people like uh Michael Dell uh Masa you have these entrepreneurs who started
00:01:49
their companies uh 30 years ago 40 years ago go and and they're still running it.
00:01:55
So, you have this amazing set of characteristics of vision, speed, fearlessness, taking risk, and a an
00:02:05
ability to pivot uh very very fast. And and I saw that a lot at NVIDIA. You know, when I was there, we were only
00:02:12
about $4 billion in sales. And uh and at that time, we were looking at lots of different ways to grow business models
00:02:18
and such. And I just remember being, you know, one story we were at a a strategic
00:02:23
offsite and it was supposed to be a review of road maps where we were looking at each one of the general
00:02:28
managers going through what they uh projected in their business and what was intended to be a roadmap review turned
00:02:35
into we're changing the strategy. We're abolishing this product line. We're going to move 2,000 engineers off of
00:02:41
project X onto project Y. And by the way, we were only about 6,000 people at the time.
00:02:46
What was project X? What was project Y? So we were involved uh at that time in trying to do uh mobile chipsets
00:02:52
connecting to an Intel processor, right? And back in the day uh for those who remember PC architecture doing these
00:02:59
chipsets competing with Intel was was really hard business and Intel was making it very very hard to compete uh
00:03:05
relative to the integration that they did. And in fact that was the genesis of starting to pivot to ARM in a very big
00:03:12
way inside Nvidia because at that time Jensen looked at what was going on with SOC's and ARMbased architecture and
00:03:18
moved everybody onto the program. Let's maybe take a step back and level set for the audience. So just to give
00:03:23
some background um Masayoshi Sun and Soft Bank took ARM private took private yeah for $32 billion.
00:03:29
$32 billion and then tried to sell it famously. Yes. Couldn't find a bidder. Could not find a bidder.
00:03:36
Hung on to it. took it public. It's now a $150 billion market cap company. That's right.
00:03:42
And you were telling us backstage he famously, you know, refuses to sell a share.
00:03:47
So, it's like a slow kind of process of just building the the shareholder base, but you've done phenomenally well as a
00:03:53
business. Just set the landscape for people that want to understand Nvidia, the most
00:03:58
valuable company in the world, but it's a it's a window to understanding AI. Mhm.
00:04:03
Why does what do they make that's so powerful? And why aren't there other competitive solutions at at that level
00:04:10
of scale yet? And how do you think that changes over the next 5 10 years? Oh boy. Uh lot lot a lot there to uh to
00:04:17
describe. So the the way to think about Nvidia uh and to some extent I I don't even though I'm the CEO of ARM I don't
00:04:25
want to tie it necessarily back to ARM but in our world what really drives demand is compute workloads. you know at
00:04:32
the end of the day is compute workloads and when a new workload is uh essentially either identified and or
00:04:39
invented then it comes down to what is the best architecture processor-wise to address that workload. So let's look at
00:04:46
AI. You know the lightning bolt moment of of Alex Net uh and the work actually that the Demison team were working on.
00:04:53
AI particularly training uh is a very very complex parallel problem that is well suited for a GPU and in fact the
00:05:02
very first work done by the engineers on AlexNet was not with Blackwell. It was not with a an AI processor but it was
00:05:09
with a gaming GPU a gaming card. So Nvidia was in a a very very good place to seize that moment relative to the
00:05:18
deepmind moment slalexnet slash the transformer/training and fast forward training these complex
00:05:25
AI models as Dennis was just talking about this is a huge huge amount of work now what role does ARM play there every
00:05:32
one of these workloads requires a CPU to not only run the computer but help the accelerator run and that's where Nvidia
00:05:39
is a customer today their most advanced chip called Grace Blackwell is 72 ARM CPUs with a Blackwell architecture and
00:05:47
that's that's where Nvidia plays today. So back to uh where does Nvidia fit? There is competition. You know Demis
00:05:54
talked about uh prior with Google they do their own chip called TPUs. Uh obviously Nvidia is the leader with
00:06:01
general purpose but right now we're in this interesting world where people are looking at is it a general purpose chip
00:06:07
is it a custom chip etc etc. It's a fascinating time to be in this industry for sure.
00:06:13
Where do you think companies like Tesla, you know, Tesla recently merged two pads and now they're working on AI5
00:06:20
and AI6? Um, and some of the more emerging companies like Cerebras and there's a whole slew of companies now,
00:06:27
Grock and others that have raised enormous amounts of money. Um, do you believe that the the role of ARM should
00:06:33
be to be the lack of a better phrase, the arms dealer to all of those folks that need that capability or at some
00:06:40
point do you think that you know you see enough of it where you're like gosh I could just do this better?
00:06:46
Maybe a little bit of both. Uh, I mean today the role we play is we are now increasingly that microprocessor that
00:06:53
connects to these accelerators whether it's something that's done by Cerebras or it's something that's done by Nvidia.
00:07:00
uh something done by uh by Google, they're connected. Uh could we do something ourselves custom? It's
00:07:06
possible. Could we also supply the intellectual property to somebody building a custom chip? We're doing that
00:07:11
today. So to some extent um we're in a very unique place that not only can we provide the solution whether it's
00:07:19
standard or custom but as AI moves from gigawatt data centers to running in these headsets or running in a wearable
00:07:26
or running in something that needs to be energy efficient you still need to run the compute workload but now you need to
00:07:32
run the run the AI workload and that is a place that I think only ARM is uniquely positioned to address.
00:07:38
So you're going to make chips and compete with Nvidia. Uh, I'm not going to say that today, but could we do that?
00:07:44
I hinted in the last conference call that we're looking at going a little bit further than we do today.
00:07:50
Could we see in the, you know, next few years, could we see a divergence in the market between training and and
00:07:55
inference? Because what I've noticed is that you've got XAI and OpenAI and, you know, Google's already doing it with
00:08:01
TPUs. They're they're building their own chips for inference, which might be, I don't know, 99% of the workloads. They
00:08:09
seem to acknowledge that Nvidia is the best at training and they don't seem they haven't at least announced an
00:08:14
effort to challenge Nvidia for for training. So is there is there a possibility that you know the the market
00:08:21
could sort of bifurcate into training chips and inference chips and inference gets much more competitive? Yes. I and I
00:08:28
also think you have a third bucket where training distills down to simpler training chips that you don't need to
00:08:34
run a trillion parameter model. You could have a giant model that now treats and teaches smaller models, mixture
00:08:42
experts, 20 billion parameters that can be a mix of inference and training doing
00:08:47
reinforcement learning where the chip is now helping uh learn trained areas. It's
00:08:53
almost like the professor teaching a student who can also be a student teacher, right? Who can do a little bit
00:08:58
of both. Uh and then there's inference that over time will be very dedicated and particularly as you get to uh end
00:09:04
points that you can't have a GPU that you know runs at at a kilowatt of power you just it's impossible
00:09:09
right so if you have robots in the field we have 500 million robots what is the chip market going to look like for
00:09:17
robotics how what makes it different than what we have today on the embedded side versus the data center side for AI
00:09:23
in yeah physical AI is going to be a gigantic market I mean today quite candidly bigger than data centers.
00:09:30
Uh yeah, I think so. Uh and because I think they're going to today they largely use repurposed automotive chips,
00:09:35
right? Things that have functional safety uh compliance around ADAS, but they're not specific for actuators or
00:09:44
specific for smaller parts of the joint. So physical AI, particularly AI that can
00:09:50
learn, uh is I think going to be a giant market because the robots themselves will have tens of chips, hundreds of
00:09:56
chips. So yeah, from a unit standpoint, it could be huge. Uh the numbers are going to be well beyond what we what we
00:10:01
see today. You started the business or ARM started really making reference designs and then
00:10:07
working with partners. Does that give you a different perspective on things like export controls and export
00:10:13
restrictions and the role that China plays in this ecosystem than say a different kind of vendor who would
00:10:20
actually be you know originating trying to tape out themselves and trying to sell through
00:10:24
to some extent? uh although we don't build anything right our business model is we do the design someone else has the
00:10:30
chip built mostly at TSMC some at Samsung even Intel uh but because we are early in the value chain relative to the
00:10:39
software ecosystem in other words we probably see what people are doing earlier than anybody else because
00:10:46
ultimately we're the link between the hardware and the software so on export control yes to some extent we have a
00:10:52
very big lens into it now today The China ecosystem actually follows the global ecosystem uh which which is good
00:10:59
uh from the standpoint that every mobile phone in China it doesn't run Google Android but it runs a version of Android
00:11:06
and it leverages this the app ecosystem that comes off of Android. Same thing with autonomous vehicles. They leverage
00:11:12
the the ADAS stack that was created by uh by ARM and then Qualcomm and Nvidia. So right now the China ecosystem on
00:11:20
software looks a lot like uh like the west which for us is obviously great. Uh and we have a very you know market
00:11:27
opinion in terms of where we want things to go. It's great if the global ecosystem remains open.
00:11:31
What's your take on um President Trump taking 9 10% of Intel and and how did that company miss this entire revolution
00:11:42
so badly? So, you know, semiconductors, which I've spent my entire career at. I I started
00:11:49
TI in 1984, and I've just been semiconductors my my whole career. There are long product cycles. It takes a long
00:11:56
time to develop chips. It takes a long time to invest in fabs. It takes a long time to define architectures and
00:12:02
ecosystems. If you miss a few, uh time is very very uh you will be punished for that. And I
00:12:09
think Intel has unfortunately been punished on a few areas. They were punished on on mobile obviously they
00:12:15
missed that completely. They were also p punished in terms of manufacturing of uh
00:12:21
of going to EUV uh on uh EUV is a an advanced uh methodology for building the smallest chips on the planet. They
00:12:29
decided not to invest in that probably a decade ago at the rate that TSMC did and
00:12:34
they fell behind. Once you fall behind in chips, it's very, very difficult to catch up because the cycle gets on top
00:12:42
of you. TSMC now has the best fabs in the world. The leading edge companies, Apple, Nvidia, AMD, they all build a
00:12:49
TSMC. TSMC gets better at what they're building. An Intel, a Samsung, they don't get the
00:12:55
opportunities. It just compounds. And and and that flywheel once it compounds and it compounds, it compounds, it's
00:13:01
very hard to catch up. series of position. So if you think about maybe then Intel
00:13:05
having lost its footing, you did mention EUV and the leaders there like companies
00:13:10
like ASML and then even one step back companies like Carl Zeiss that make these lenses. Those are critical
00:13:17
infrastructure that the west needs. Is there a role for the government to be spending more capital to incubate those
00:13:25
kinds of things so that we have a little bit more diversity in the supply chain?
00:13:29
So that you know if you contrast and compare there's the Intel investment but then there's these other things that are
00:13:35
still maybe we should also be doing. Oh 100%. I mean if you look at um one of one of the most critical components in
00:13:41
building chips are these rare earth compounds and there's a belief that oh China has cornered the market because
00:13:46
they have all the access to these rare earth minerals. The access for the minerals are global. There's no issue in
00:13:53
getting access to materials. Yeah, the issue is in the refinement and actually building the factories that can
00:13:59
refine the materials. Again, that's a decades level of investment. And I'll tell you one thing
00:14:05
that I when I I lived in China for a number of years and one of the things that I was very impressed with when I
00:14:10
lived there and still am is the uh industrial policy that sits inside the central government that will last uh
00:14:19
respectfully an election cycle and it will essentially be something that they require a lot of the folks who are in
00:14:27
the Ministry of Technology to be engineers to be thinking about a policy on on building. So to your question,
00:14:33
should the US do it? Absolutely. Okay. So Rene, let me put you on the spot. Look, between the Korea trade
00:14:38
deal, the Japanese trade deal, the European trade deal, you know, we have close to now two trillion of investment
00:14:43
capital that these countries will make into the United States. How do we go about creating
00:14:50
an ASML type company or capability or you know these lenses like how do we do that? What universities do we go to or
00:14:59
what labs do we go to? What do we do? I I I think there probably needs to be more of some of the US companies working
00:15:06
together. And I'll say this because ARM is not a US company, but we I would do the same if I would working together uh
00:15:12
pool pulling capital for some of these initiatives to essentially get some type of grounding. You need universities uh
00:15:18
but you need corporations to get behind this as well as well as uh financing private equity all kinds of different
00:15:25
capital because this is a this is a huge capital investment that also requires investment from companies and and and
00:15:32
private equity but at the same time needs to last for years. Just talking about the fabs TSMC's built
00:15:39
this facility in Arizona. There was reports about the inability to get labor to train labor to get a workforce that I
00:15:46
don't know what the right term to use is culturally the workforce would operate the same way as they do back in Taiwan
00:15:52
and they were really challenged and they had to bring folks over to Arizona to work the facility. These were news
00:15:57
reports so I I we don't know this firsthand. Do you think we have the capacity to do fabs in in the United
00:16:03
States on uh on shore here? And what's it going to take if you were in the administration? Let's say you were the
00:16:09
AISAR for example. What would you advise the president to do to ensure that that
00:16:13
happens successfully? Yeah, I don't want I don't want to take anything away from David. He's doing an
00:16:17
amazing job as the AISAR. You've hit a very key tenant though relative to uh worldclass manufacturing inside the
00:16:24
United States and what is required to uh to make that happen. We had it decades ago, believe it or not. There was there
00:16:32
was a time where the leading contract manufacturers uh in the world were US-based companies
00:16:39
uh and uh and we knew how to do that. And if you go back 30 years ago when Apple and Compact used to build their
00:16:47
own PCs and they had their own factories, believe it or not, then all of that went to companies like
00:16:53
Flextronics and SCI etc etc. So we had that uh ultimately for cost reasons that began to move all the way to uh to the
00:17:02
Far East into Foxcon in China etc etc. There's a great book uh Apple in China that documents a lot of this to your
00:17:10
point in terms of you know could we get that back in some ways there's no reason
00:17:14
why we why we couldn't but it is a mindset TSMC is a 24/7 operation where if a line goes down or a customer's got
00:17:22
a problem not only are the technicians need to be ready to go the engineers to be need to be ready to go and that is
00:17:29
something that uh I think we've lost the muscle memory inside the United States quite frankly on how to go do that. I
00:17:35
mean, we may have had it a generation or so ago. I don't know that we have it now. And we certainly haven't trained a
00:17:41
generation of folks to look at manufacturing jobs as being something that is as lucrative and prestigious.
00:17:46
They're sort of thinking, "Oh, it's a blue collar job. I don't want to go into that way." It's not viewed that way uh
00:17:50
in Taiwan, right? And in Taiwan, if you say you're working for TSMC or studying to go off and do that, it's a highly
00:17:56
prestigious kind of thing. So, it's it's not just the AISAR's uh problem. I think
00:18:02
it's uh it's deeper than that in terms of us getting so you've diagnosed the problem. Do you
00:18:05
have a solution or recommendation? Is there a short form that you could highlight?
00:18:08
I I you know I think we've seen a huge amount of work already done by universities. I was at Carnegie Melon uh
00:18:14
a couple weeks ago. They now have micro electronics classes for chip design. That was gone a number of years ago.
00:18:20
There weren't even people designing chips. So I think getting manufacturing operations excellence uh into the
00:18:25
universities making that a field of of discipline uh that the universities get behind to build up that capacity in the
00:18:33
US. I think that's required. Let me go back to export controls which Jamath mentioned. And I'm not sure people here
00:18:38
know exactly how these things work, but basically if a product like a advanced semiconductor is put on the export
00:18:44
control list, it means that the company that's selling it or the buyer, they have to apply for a license from the
00:18:50
commerce department to get their purchase order fulfilled. And the the commerce department will then, you know,
00:18:57
process that license request and it goes through some inter agency committee and
00:19:01
five different departments will basically have to sign off on it. And best case scenario, it takes months, but
00:19:06
there are license applications that literally have been in the hopper for 2 years, by which time the chip is
00:19:11
obsolete. And believe it or not, there are a lot of people on groups in Washington right now who are calling for
00:19:17
literally every sale of a advanced semiconductor worldwide to be a licensed sale uh because they think that GPUs are
00:19:25
like plutonium or something and they're inherently scary. I mean, this is seriously the the the discourse that's
00:19:30
going on right now. And in fact, there was um there's a major uh rule that was put forward called the Biden diffusion
00:19:36
rule in the last 5 days of the Biden administration that basically did require every sale of a GPU worldwide to
00:19:42
be licensed subject to some carveouts. Uh we we rescended that, but there is a neverending clamor and pressure in
00:19:49
Washington to bring back these sorts of of rules. And the irony is that the people who are advocating for these
00:19:55
things called themselves China hawks. But it seems to me that the whole basis of the semiconductor industry, the
00:20:01
reason why it's moved so fast, why you get new chips every year is it's really been left alone by the government for
00:20:07
the most part and hasn't it hasn't been a highly regulated industry. And I'm curious, what do you think will happen
00:20:13
to the industry and the pace of innovation if the government now makes it heavily regulated in the way that I'm
00:20:19
describing? You you brought up a great point and I think I think we may even have a couple of those in the queue that
00:20:24
hasn't been approved for for a couple of years. You're right. Semiconductors have
00:20:29
not been regulated traditionally. And because of that, if you look at the real heart of what drives uh semiconductor
00:20:36
growth, compute, whether it's Intel, whether it's ARM, whether it's Nvidia, that's the West. And why is that the
00:20:43
West? Because that requires both innovation at the chip level and a global software ecosystem. And the world
00:20:48
works really well when it's flat and there isn't constraints relative to who you sell to or how ecosystems get built.
00:20:57
If you shut off supply of a computing architecture into other parts of the world, what what will happen? Certain
00:21:04
parts of the world that have the capabilities either in terms of people, technology,
00:21:09
uh innovation, they will find a way and they will find a way around around the problem. And once that happens, you've
00:21:16
now created two parallel universes. And then the US and the West would be at risk of that other ecosystem being an
00:21:24
ecosystem of choice. So if you can navigate for those licenses being expedited, uh the the world works really
00:21:31
well in semis when it's flat and a global ecosystem. Uh may the best company win. Renee, the company started
00:21:38
in Cambridge and uh originally all the employees were there, but now it's sort of, you know, I
00:21:44
think 50% of the employees are in the UK. Um, tell us about building a company there and just multiculturally and where
00:21:53
you're going based on sort of, you know, where technology is going. company was started in uh in the UK in Cambridge uh
00:22:00
in a barn uh part of a joint venture for uh the Apple Newton uh building a processor combination of a joint venture
00:22:07
of Apple and BLSI technology. They needed a lowcost chip that could run off a battery. They contracted a company to
00:22:14
build the chip. The chip wasn't so good, but a bunch of guys said, "You know what? The design's pretty good and why
00:22:20
don't we try to build a business from it?" And that that's how ARM uh ARM was born. I'm the fourth CEO. Um, I'm the
00:22:26
first one that is not from the UK. Uh, and I'm what I've been trying to do in the in the three and a half years that I
00:22:32
took over is to keep the great scientists and technology innovation that we have in Cambridge, but inject a
00:22:39
bit of a a Silicon Valley uh aggressiveness and and twist to uh to moving faster and going quicker. Uh now
00:22:46
as you said half the employees are in the UK but we've got folks globally 2,000 people in Bangalore uh probably
00:22:52
over over a thousand in the United States different parts of Europe. So it's a highly global company and we go
00:22:58
where the talent is and we look for great engineers. Are you able to find great STEM talent
00:23:02
still here or do you need now more investment in core double E and chip design? We need far more investment. Uh our
00:23:09
business is not one yet where I can say I'm hiring less people because of AI. I'm certainly hiring less finance people
00:23:16
and legal people. Sorry Jason and Spencer if you're in the audience. But for engineers, uh, AI for development,
00:23:22
AI for creation, AI for science, that's still a hard problem to solve. Uh, which
00:23:29
is why we need more engineers to develop chips, which is great. I think back to is there more demand for compute? Is
00:23:34
this AI wave that we're seeing going to continue in the world of generating AI for science and creation? I think
00:23:42
there's a ways to go. Leveling up for a second and looking at our relationship with China and to get a
00:23:47
little geopolitical here, how do you view China versus America? Is this going to be a winner take all with AI? Or can
00:23:56
these two, you know, powers get along? Are we competitors? Are we collaborators? Are we destined to fight
00:24:05
uh and go to war in Taiwan like we talked about last year on this stage? What's your take on it? And is there a
00:24:10
path to us having a great collaboration with China? I'm going to be an optimist here, Jason,
00:24:15
and say I think yes. Uh I think uh that that China views some of the things around AI in terms of whether there are
00:24:24
these are things like guard rails or policies or things to keep things in such a way that we've got the right
00:24:29
level of safety checks. I think their their their minds are in the right space. And I say this just based upon
00:24:34
conversations I've had with folks over there. I wouldn't necessarily compare it to the nuclear arms race, but in some
00:24:41
ways it's not dissimilar in the sense that you need the the the countries that have the capabilities to be willing to
00:24:47
sit at the table to have the conversations and China in my experience has shown that so far.
00:24:52
Ladies and gentlemen, Renee Hos. Thank you. [Applause] Thanks, Rene. [Music] Thank you so much.