WEBVTT - Arm CEO Rene Haas Talks Company Outlook

0:00:02.520 --> 0:00:07.040
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news.

0:00:08.080 --> 0:00:10.799
<v Speaker 2>Look, they two beat the street on earnings, on earnings

0:00:10.800 --> 0:00:13.360
<v Speaker 2>for share basis. You look at revenue dialing up almost

0:00:13.400 --> 0:00:16.720
<v Speaker 2>twenty percent. But again it's a world a conservative forecast.

0:00:16.760 --> 0:00:19.360
<v Speaker 2>And what this means not only for smartphones, but what

0:00:19.360 --> 0:00:21.560
<v Speaker 2>it means for AI demand as well. We've got to

0:00:21.600 --> 0:00:24.880
<v Speaker 2>get to it with arms CEO Rene Haas, who joins

0:00:24.960 --> 0:00:27.160
<v Speaker 2>us now, and I put it to you, you've been

0:00:27.560 --> 0:00:30.360
<v Speaker 2>out everywhere at the moment. There's deals being done, of course,

0:00:30.360 --> 0:00:32.360
<v Speaker 2>a soft bank and open AI that will dig into.

0:00:32.800 --> 0:00:35.440
<v Speaker 2>But what about this forecast and what it signals for

0:00:35.479 --> 0:00:37.640
<v Speaker 2>your business? Is it just you being conservative as a

0:00:37.680 --> 0:00:39.640
<v Speaker 2>newly public company.

0:00:40.080 --> 0:00:43.760
<v Speaker 1>Thanks Kerlen, and appreciate you having me on this morning. Yeah,

0:00:43.760 --> 0:00:46.960
<v Speaker 1>I don't know if I agree with the dialogue muted forecast.

0:00:47.560 --> 0:00:52.600
<v Speaker 1>We beat our projections for Q three and we actually

0:00:52.720 --> 0:00:56.800
<v Speaker 1>raise guidance four Q four and for the year, and

0:00:56.840 --> 0:00:59.959
<v Speaker 1>we're looking at about four billion dollars of annual revenue,

0:01:00.320 --> 0:01:02.560
<v Speaker 1>which is about twenty five percent.

0:01:02.280 --> 0:01:03.480
<v Speaker 3>Increase over the last year.

0:01:03.880 --> 0:01:06.720
<v Speaker 1>So we're super happy with the trajectory of the business

0:01:06.760 --> 0:01:08.040
<v Speaker 1>and super happy with the performance.

0:01:08.800 --> 0:01:13.600
<v Speaker 2>Maybe just investors are too exuberant they're too exuberant around

0:01:13.640 --> 0:01:15.880
<v Speaker 2>the fact that you've already said your compute platform is

0:01:15.920 --> 0:01:18.360
<v Speaker 2>going to help open ai and SoftBank make good on

0:01:18.400 --> 0:01:22.280
<v Speaker 2>their Advanced Enterprise AI, the optimism around your AI future

0:01:22.319 --> 0:01:25.080
<v Speaker 2>as well as how integral you are smartphones. So is

0:01:25.120 --> 0:01:28.679
<v Speaker 2>it that are people too excited about the AI opportunity here?

0:01:30.040 --> 0:01:32.679
<v Speaker 1>Yeah, I'm going to try to decouple what the market

0:01:32.760 --> 0:01:34.640
<v Speaker 1>is saying about our results and just talk about the

0:01:34.680 --> 0:01:38.480
<v Speaker 1>results in general, of which we are again incredibly pleased by.

0:01:38.520 --> 0:01:41.520
<v Speaker 1>We are seeing extremely strong demand for our products across

0:01:41.560 --> 0:01:44.800
<v Speaker 1>the board. We had record royalties. We were up twenty

0:01:44.840 --> 0:01:49.040
<v Speaker 1>three percent year on year. Our version nine compute subsystems,

0:01:49.080 --> 0:01:52.240
<v Speaker 1>which drive higher world TIID rates. We're just growing faster

0:01:52.320 --> 0:01:55.560
<v Speaker 1>than the market, which is what we told investors when

0:01:55.600 --> 0:01:57.400
<v Speaker 1>we went public back in Analyst Day a year and a.

0:01:57.400 --> 0:01:58.600
<v Speaker 3>Half ago plus.

0:01:59.120 --> 0:02:02.720
<v Speaker 1>So yeah, we're not seeing any slowdown in demand quite frankly, Carol.

0:02:02.800 --> 0:02:05.840
<v Speaker 1>In fact, we're seeing an acceleration in it, particularly driven

0:02:05.840 --> 0:02:07.560
<v Speaker 1>by all things around artificial intelligence.

0:02:08.880 --> 0:02:12.160
<v Speaker 4>Rene. Speaking of future growth, one of the things that

0:02:12.320 --> 0:02:15.600
<v Speaker 4>ARM is also banking on is its high profile partnership

0:02:16.040 --> 0:02:19.239
<v Speaker 4>in Stargate, and you're a big player in that could

0:02:19.240 --> 0:02:22.239
<v Speaker 4>you kind of tell us what investors can expect this

0:02:22.480 --> 0:02:24.440
<v Speaker 4>year to gain from that venture.

0:02:25.800 --> 0:02:28.119
<v Speaker 1>I can say, broadly speaking, the way you think about

0:02:28.160 --> 0:02:31.839
<v Speaker 1>Stargate is it's an infrastructure project unlike any other five

0:02:31.919 --> 0:02:35.680
<v Speaker 1>hundred billion dollars. One hundred billion dollars will spend initially,

0:02:36.160 --> 0:02:38.600
<v Speaker 1>and these are to build out the world's largest AI

0:02:38.720 --> 0:02:41.959
<v Speaker 1>data centers for training and inference. We are the key

0:02:42.000 --> 0:02:46.120
<v Speaker 1>technology partner around CPU. We're the only CPU technology partner

0:02:46.160 --> 0:02:49.760
<v Speaker 1>that was named, which is terrific obviously for US. One

0:02:49.800 --> 0:02:53.320
<v Speaker 1>can imagine going forward a lot of opportunity for technology

0:02:53.360 --> 0:02:56.600
<v Speaker 1>innovation there. But in a near term we'll be linked

0:02:56.639 --> 0:02:59.480
<v Speaker 1>with the Grace Blackwell ship using the Gray CPU which

0:02:59.480 --> 0:03:01.000
<v Speaker 1>is armed bay and the.

0:03:00.960 --> 0:03:02.919
<v Speaker 3>Black Lill GPU which is video based.

0:03:03.400 --> 0:03:06.240
<v Speaker 1>But these large frontier models are going to require Jackie

0:03:06.280 --> 0:03:09.880
<v Speaker 1>a huge amount of power for training and inference, and

0:03:10.200 --> 0:03:12.120
<v Speaker 1>we're just so excited about being involved in this. It's

0:03:12.160 --> 0:03:14.160
<v Speaker 1>going to be fantastic. Well.

0:03:14.200 --> 0:03:16.800
<v Speaker 4>At the same time, this project also kind of puts

0:03:16.880 --> 0:03:20.440
<v Speaker 4>you in the middle of this US China tech rivalry.

0:03:20.680 --> 0:03:23.280
<v Speaker 4>You're at the front lines of it, being the top

0:03:23.320 --> 0:03:26.520
<v Speaker 4>supplier to American AI companies, And I'm curious if you

0:03:26.560 --> 0:03:30.200
<v Speaker 4>think that perhaps tying your fortunes so closely to open

0:03:30.240 --> 0:03:31.160
<v Speaker 4>Ai is a risk.

0:03:32.680 --> 0:03:34.360
<v Speaker 1>I'm not sure I think about it quite that way.

0:03:34.600 --> 0:03:38.480
<v Speaker 1>Open Ai is a fantastic partner. Arguably they're the leader

0:03:38.720 --> 0:03:41.520
<v Speaker 1>at the moment in terms of adoption of chat, GPT

0:03:41.720 --> 0:03:45.400
<v Speaker 1>and their APIs and their enterprise suite. We work with everyone,

0:03:45.480 --> 0:03:47.560
<v Speaker 1>though we work very closely with Google, we work closely

0:03:47.600 --> 0:03:51.160
<v Speaker 1>with Meta Aws, you know all the partners. But we're

0:03:51.240 --> 0:03:53.000
<v Speaker 1>very thrilled to be involved with open Ai. They have

0:03:53.080 --> 0:03:54.200
<v Speaker 1>phenomenal technology.

0:03:54.960 --> 0:03:58.240
<v Speaker 2>Rene. We're going to lean into that China Us story.

0:03:58.440 --> 0:04:02.600
<v Speaker 2>Deep Seek extraordinary the push and pull has in the narratives.

0:04:02.880 --> 0:04:05.760
<v Speaker 2>Can you explain as to whether you think we are

0:04:05.800 --> 0:04:08.560
<v Speaker 2>going to need as much compute as we all foresaw before.

0:04:09.840 --> 0:04:12.760
<v Speaker 1>Oh, I think unquestionably, Caroline, And in fact, I think

0:04:13.120 --> 0:04:15.160
<v Speaker 1>even more than we probably have been thought about.

0:04:15.520 --> 0:04:16.680
<v Speaker 3>So why is that?

0:04:17.240 --> 0:04:19.280
<v Speaker 1>Well, if you look at what deep Seek has done,

0:04:19.279 --> 0:04:21.400
<v Speaker 1>which is an amazing, by the way piece of work

0:04:22.080 --> 0:04:25.600
<v Speaker 1>they've done what's called distillation, and distillation is building on

0:04:25.680 --> 0:04:29.040
<v Speaker 1>top of an existing frontier model which takes lots of

0:04:29.080 --> 0:04:32.240
<v Speaker 1>power and lots of compute to generate. If you think

0:04:32.279 --> 0:04:36.719
<v Speaker 1>about GPT four, which were was hundreds of megawatts GPT

0:04:36.839 --> 0:04:37.680
<v Speaker 1>five sixty seven.

0:04:37.760 --> 0:04:39.360
<v Speaker 3>These are in the gigawatts.

0:04:39.760 --> 0:04:42.640
<v Speaker 1>So even though you may have models that can distill

0:04:42.800 --> 0:04:46.039
<v Speaker 1>or build a con the frontier models, the frontier models

0:04:46.040 --> 0:04:48.520
<v Speaker 1>still need a lot of power. And then with that

0:04:48.520 --> 0:04:51.240
<v Speaker 1>that drives a huge demand for inference, which is really

0:04:51.279 --> 0:04:54.680
<v Speaker 1>where the larger use loads are. And more efficient inference

0:04:55.000 --> 0:04:59.839
<v Speaker 1>means you can drive AI to much smaller devices, to smartphones, earbuds,

0:05:00.720 --> 0:05:04.279
<v Speaker 1>devices that don't have a lot of capability for power.

0:05:04.640 --> 0:05:07.080
<v Speaker 1>It's great for ARMED because that's where we are today,

0:05:07.080 --> 0:05:09.080
<v Speaker 1>it's where we exist, and now you can run those

0:05:09.080 --> 0:05:12.400
<v Speaker 1>AI workloads on the CPU. So I think actually what

0:05:12.440 --> 0:05:14.880
<v Speaker 1>deep seat proves is that we're just going to see

0:05:14.880 --> 0:05:17.240
<v Speaker 1>an even more accelerant towards towards compute.

0:05:18.400 --> 0:05:21.160
<v Speaker 4>Well on that front, you know you have a significant

0:05:21.160 --> 0:05:24.839
<v Speaker 4>business in China. Are you concerned at all about perhaps

0:05:25.200 --> 0:05:29.800
<v Speaker 4>how the slowing smartphone business might affect Arm's future growth.

0:05:31.320 --> 0:05:32.400
<v Speaker 3>Yeah, it's a great question.

0:05:32.520 --> 0:05:34.880
<v Speaker 1>And you know, again what we've seen in smartphones with

0:05:34.960 --> 0:05:39.000
<v Speaker 1>your single digits type of growth, we're in the double

0:05:39.040 --> 0:05:42.120
<v Speaker 1>digits growth in terms of royalties, in some cases twenty

0:05:42.160 --> 0:05:45.800
<v Speaker 1>percent growth. And why is that we're putting more technology

0:05:45.800 --> 0:05:48.920
<v Speaker 1>into these phones, our version nine, our compute subsystems drive

0:05:49.000 --> 0:05:53.040
<v Speaker 1>higher royalty rates. So we're just growing faster than the market,

0:05:53.160 --> 0:05:55.640
<v Speaker 1>and particularly in the premium segment which requires more and

0:05:55.680 --> 0:05:58.719
<v Speaker 1>more compute. So while the units have been slowed in

0:05:58.760 --> 0:06:01.000
<v Speaker 1>some areas, it actually has an impacted as very much.

0:06:01.080 --> 0:06:02.359
<v Speaker 3>Just look at our record royalties.

0:06:03.200 --> 0:06:05.080
<v Speaker 1>To have this kind of number in a market where

0:06:05.080 --> 0:06:06.960
<v Speaker 1>people think things are slowing down, just.

0:06:06.920 --> 0:06:10.560
<v Speaker 2>Phenomenal record royalties. Upside in licensing. I would just want

0:06:10.600 --> 0:06:12.680
<v Speaker 2>to get to the litigation if I'm met here RENAA,

0:06:12.680 --> 0:06:14.920
<v Speaker 2>and briefly, perhaps it hasn't gone in the way that

0:06:14.960 --> 0:06:16.760
<v Speaker 2>you thought with Qualcom. What does it mean for the

0:06:16.760 --> 0:06:17.520
<v Speaker 2>future earnings?

0:06:18.920 --> 0:06:22.680
<v Speaker 1>Really, there's no impact on future earnings, Caroline. The trial

0:06:22.760 --> 0:06:24.960
<v Speaker 1>was a mistrial on one of the verdicts, so the

0:06:25.000 --> 0:06:27.599
<v Speaker 1>dispute is still open. But to your question, in terms

0:06:27.600 --> 0:06:29.919
<v Speaker 1>of the future forecast, we had no impact whatsoever.

0:06:33.440 --> 0:06:36.680
<v Speaker 4>Renee Ha, CEO of ARM, thank you for joining us.