WEBVTT - Advanced Micro Devices Chair, President, & CEO Lisa Su Talks AI Demand & the Future

0:00:02.400 --> 0:00:07.120
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news.

0:00:07.920 --> 0:00:11.440
<v Speaker 2>From robotaxis to AI chips, Bloomberg's at ludlowis standing by

0:00:11.520 --> 0:00:14.239
<v Speaker 2>with Advanced micro Devices CEO Lisa Sue, who joins them

0:00:14.240 --> 0:00:17.560
<v Speaker 2>live from AMD's Advancing AI event in California where they

0:00:17.720 --> 0:00:20.320
<v Speaker 2>just unveiled their latest generation of chips.

0:00:20.480 --> 0:00:20.560
<v Speaker 1>ED.

0:00:21.440 --> 0:00:24.920
<v Speaker 3>That's right, Scarlett and Lisa Sue celebrating ten years as

0:00:25.040 --> 0:00:29.280
<v Speaker 3>CEO of AMD. Lisa m I three twenty five X.

0:00:29.440 --> 0:00:33.640
<v Speaker 3>Everyone I speak to in the world of technology accepts

0:00:33.680 --> 0:00:38.600
<v Speaker 3>and sees that AMD is the number two rival behind Nvidia.

0:00:38.960 --> 0:00:43.639
<v Speaker 3>How does the latest accelerator change that competitive landscape in

0:00:43.680 --> 0:00:44.320
<v Speaker 3>that market?

0:00:45.720 --> 0:00:47.400
<v Speaker 1>Well, first of all, Ed, it's great to be here

0:00:47.440 --> 0:00:49.680
<v Speaker 1>with you today, Thank you for having us. It was

0:00:49.720 --> 0:00:52.800
<v Speaker 1>a big day for us. We talked about so many

0:00:52.840 --> 0:00:56.080
<v Speaker 1>new technologies. We launched a number of new products. We

0:00:56.200 --> 0:00:59.480
<v Speaker 1>launched our new fifth gen epic server product called Turn.

0:01:00.000 --> 0:01:02.200
<v Speaker 1>We launched MI I three twenty five and talked about

0:01:02.240 --> 0:01:05.160
<v Speaker 1>our future roadmap and I three fifty and beyond. We

0:01:05.240 --> 0:01:08.360
<v Speaker 1>talked about aipcs and networking. So it's been a great

0:01:08.480 --> 0:01:11.240
<v Speaker 1>day as it relates to you know, MI I three

0:01:11.280 --> 0:01:13.399
<v Speaker 1>twenty five and what we're doing in the data center

0:01:13.440 --> 0:01:16.560
<v Speaker 1>accelerator market. I mean, this is such an exciting market

0:01:16.640 --> 0:01:21.760
<v Speaker 1>right now. There's so much demand overall for new AI capability,

0:01:21.920 --> 0:01:24.520
<v Speaker 1>and you know, we are doing extremely well. We had

0:01:24.520 --> 0:01:26.960
<v Speaker 1>a number of customers on stage with us today too,

0:01:27.040 --> 0:01:32.640
<v Speaker 1>including Microsoft, Meta, Oracle, number of new AI companies as

0:01:32.640 --> 0:01:35.319
<v Speaker 1>well as all of the major OEMs, and everyone is

0:01:35.360 --> 0:01:38.880
<v Speaker 1>excited about MI three twenty five and the roadmap.

0:01:38.480 --> 0:01:44.800
<v Speaker 3>Ahead, and yet shares closed down four percent. The disappointment

0:01:44.920 --> 0:01:47.240
<v Speaker 3>is not in the long term, right you updated the

0:01:47.280 --> 0:01:50.880
<v Speaker 3>forecast for the market five hundred billion dollars in twenty

0:01:50.960 --> 0:01:54.120
<v Speaker 3>twenty eight. But what can you tell me or how

0:01:54.120 --> 0:01:57.120
<v Speaker 3>can you quantify the near term demand of those new

0:01:57.160 --> 0:01:58.680
<v Speaker 3>customers that you just outlined.

0:01:59.760 --> 0:02:02.360
<v Speaker 1>Well, you know, the best thing to say is this

0:02:02.440 --> 0:02:05.800
<v Speaker 1>is an incredibly fast growing market. I mean, if you

0:02:05.880 --> 0:02:08.160
<v Speaker 1>think about you know, what we've done just over the

0:02:08.240 --> 0:02:11.800
<v Speaker 1>last twelve months or so, we've seen three hundred demand

0:02:11.919 --> 0:02:14.680
<v Speaker 1>increase from when we started the year to In our

0:02:14.720 --> 0:02:17.600
<v Speaker 1>most recent earnings, we actually said we would be over

0:02:17.680 --> 0:02:21.000
<v Speaker 1>four and a half billion dollars in this important AI

0:02:21.040 --> 0:02:23.519
<v Speaker 1>accelerator market. You know, we view this as a multi

0:02:23.600 --> 0:02:26.600
<v Speaker 1>year opportunity for us, And what's really important. You know,

0:02:26.639 --> 0:02:29.280
<v Speaker 1>what our customers say is, you know, we've shown that

0:02:29.320 --> 0:02:32.839
<v Speaker 1>we have excellent hardware, but even more importantly, we've made

0:02:32.919 --> 0:02:36.480
<v Speaker 1>tremendous progress in the overall software roadmap. And what that

0:02:36.520 --> 0:02:39.120
<v Speaker 1>allows is that allows customers to get to market, you know,

0:02:39.160 --> 0:02:41.640
<v Speaker 1>as fast as possible. So we're seeing, you know, with

0:02:42.000 --> 0:02:44.160
<v Speaker 1>three twenty five, we expect the new chips to be

0:02:44.520 --> 0:02:48.320
<v Speaker 1>anywhere from twenty to forty percent higher performance on key

0:02:48.360 --> 0:02:51.600
<v Speaker 1>inferencing workloads than the competition. And that's a great place

0:02:51.639 --> 0:02:51.840
<v Speaker 1>to be.

0:02:53.280 --> 0:02:57.040
<v Speaker 3>Lisa, describe the demand for your AI products in just

0:02:57.080 --> 0:03:04.560
<v Speaker 3>a single word, amazing. Give me more than that. How

0:03:04.639 --> 0:03:07.640
<v Speaker 3>is this different to a year ago? You remember us

0:03:07.680 --> 0:03:11.240
<v Speaker 3>sitting in San Jose, right, we were sitting together and

0:03:11.280 --> 0:03:14.280
<v Speaker 3>it was the start of a very different time in

0:03:14.320 --> 0:03:17.560
<v Speaker 3>the investment cycle. How is it different now?

0:03:19.040 --> 0:03:22.560
<v Speaker 1>Well, I think so In a single word, the best

0:03:22.560 --> 0:03:26.000
<v Speaker 1>thing to say is it's an accelerating market. And you

0:03:26.040 --> 0:03:30.040
<v Speaker 1>know what we see is so many new applications, so

0:03:30.160 --> 0:03:34.400
<v Speaker 1>many new opportunities to use AI. We're at the very

0:03:34.440 --> 0:03:36.480
<v Speaker 1>beginning of the cycle. And one of the things that

0:03:37.040 --> 0:03:39.640
<v Speaker 1>I've been very very encouraged that is, you know, when

0:03:39.640 --> 0:03:42.040
<v Speaker 1>we started maybe twelve months ago, you would see that,

0:03:42.320 --> 0:03:44.360
<v Speaker 1>you know, there would be a lot of people trying things,

0:03:44.440 --> 0:03:46.320
<v Speaker 1>but it would actually take them a while to cut

0:03:46.400 --> 0:03:50.000
<v Speaker 1>over to their production environment. That's actually much faster now.

0:03:50.040 --> 0:03:52.160
<v Speaker 1>And what that means is, you know, there is more

0:03:52.240 --> 0:03:57.840
<v Speaker 1>demand for AI capability. But we've also significantly improved our

0:03:57.960 --> 0:04:01.160
<v Speaker 1>tools and overall software infrastructure. Sure. You know, we call

0:04:01.280 --> 0:04:04.280
<v Speaker 1>our software suite something called rock Them and it has

0:04:04.560 --> 0:04:08.280
<v Speaker 1>just made tremendous progress over the last ten to eleven months.

0:04:08.440 --> 0:04:12.440
<v Speaker 1>That really allows customers to adopt our solutions much much faster.

0:04:12.760 --> 0:04:14.880
<v Speaker 1>And that's the key for innovation is to try to

0:04:14.880 --> 0:04:17.320
<v Speaker 1>get as much in as fast as possible.

0:04:17.839 --> 0:04:21.120
<v Speaker 3>And CPU as well. If you're an investor on Wall Street,

0:04:21.240 --> 0:04:24.039
<v Speaker 3>or you are a retail investor, or you're a technologist

0:04:24.160 --> 0:04:27.800
<v Speaker 3>or dare I say it a technology journalist, everyone is

0:04:27.880 --> 0:04:32.120
<v Speaker 3>so over indexed to GPU. I really want to know

0:04:32.400 --> 0:04:36.880
<v Speaker 3>if you see a sort of revision back towards emphasis

0:04:36.880 --> 0:04:40.200
<v Speaker 3>on CPU with all of the data center investment and

0:04:40.240 --> 0:04:41.440
<v Speaker 3>build out that is to come.

0:04:42.839 --> 0:04:46.000
<v Speaker 1>I think the most important thing to think about when

0:04:46.000 --> 0:04:49.159
<v Speaker 1>you think about data center infrastructure is there is no

0:04:49.279 --> 0:04:52.239
<v Speaker 1>one size fits all in computing. You know, you need

0:04:52.320 --> 0:04:55.279
<v Speaker 1>great CPUs, you need great GPUs, you need to be

0:04:55.279 --> 0:04:57.800
<v Speaker 1>able to bring those together in systems. You need to

0:04:57.839 --> 0:05:01.760
<v Speaker 1>have all these different types of you need to have. Frankly,

0:05:01.800 --> 0:05:05.200
<v Speaker 1>AIPCS is another category that we find to be a

0:05:05.320 --> 0:05:08.640
<v Speaker 1>very interesting and exciting category. So when you put all

0:05:08.640 --> 0:05:11.400
<v Speaker 1>those things together, First of all, we're super excited about

0:05:11.480 --> 0:05:15.880
<v Speaker 1>our new generation of EPIC servers. This is our fifth

0:05:15.920 --> 0:05:20.400
<v Speaker 1>generation we called turn. We've seen tremendous performance in many cases,

0:05:20.920 --> 0:05:24.080
<v Speaker 1>our performance over the competition has increased substantially with this

0:05:24.120 --> 0:05:27.719
<v Speaker 1>new generation. And what CIOs are telling me is they're

0:05:27.760 --> 0:05:30.839
<v Speaker 1>looking to modernize their data centers. You know, the idea

0:05:31.040 --> 0:05:33.560
<v Speaker 1>of you know, there's a lot of data center infrastructure

0:05:33.600 --> 0:05:36.520
<v Speaker 1>out there that is, you know, three, four or five

0:05:36.640 --> 0:05:39.680
<v Speaker 1>years old. You know, those guys are not as performant,

0:05:40.160 --> 0:05:42.160
<v Speaker 1>They take up a lot of space, and frankly, they

0:05:42.520 --> 0:05:45.599
<v Speaker 1>cost a lot from an operational standpoint. And so we're

0:05:45.640 --> 0:05:49.520
<v Speaker 1>able to substantially modernize the data center by adopting our

0:05:49.600 --> 0:05:53.360
<v Speaker 1>new generation of EPIC CPUs. We're seeing anywhere from a

0:05:53.600 --> 0:05:57.479
<v Speaker 1>seven to one ratio. You can replace seven legacy servers

0:05:57.520 --> 0:06:01.760
<v Speaker 1>with a single new and fifth generation EPIC server and

0:06:01.800 --> 0:06:06.880
<v Speaker 1>that just results in tremendous Paybass for enterprises.

0:06:07.760 --> 0:06:11.440
<v Speaker 3>Lisa, this is a question from our Bloomberg Technology Audience direct.

0:06:12.040 --> 0:06:14.839
<v Speaker 3>Would you consider using Intel for foundry.

0:06:16.480 --> 0:06:20.200
<v Speaker 1>Well, look, when we look at the manufacturing landscape for semiconductors,

0:06:20.480 --> 0:06:25.080
<v Speaker 1>it's really important to have a strong manufacturing set of partners.

0:06:25.600 --> 0:06:28.359
<v Speaker 1>We're extremely happy with the work that we do with TSMC.

0:06:28.560 --> 0:06:32.240
<v Speaker 1>They're great partner. It is also super important for us

0:06:32.240 --> 0:06:35.279
<v Speaker 1>to have geographic diversity. I think what the Chips Act

0:06:35.320 --> 0:06:37.880
<v Speaker 1>has done has been excellent in terms of trying to

0:06:37.920 --> 0:06:41.360
<v Speaker 1>bring more manufacturing back to the US. We're excited about

0:06:41.400 --> 0:06:44.040
<v Speaker 1>the work that's being done in Arizona, and we're very

0:06:44.080 --> 0:06:47.240
<v Speaker 1>active in bringing some of our chips into the new

0:06:47.279 --> 0:06:51.120
<v Speaker 1>factories of TSMC and Arizona. And we're always looking at

0:06:51.400 --> 0:06:54.279
<v Speaker 1>the manufacturing landscape and we'll always think about how we

0:06:54.320 --> 0:06:56.560
<v Speaker 1>can have the most resilient supply chain.

0:06:57.720 --> 0:07:00.359
<v Speaker 3>Lisa. For the last month, Bloomberg's Ink and I have

0:07:00.440 --> 0:07:05.120
<v Speaker 3>been using aipcs with processes based on our architecture. I

0:07:05.160 --> 0:07:08.680
<v Speaker 3>feel like the GAT with XAV six is closing. When

0:07:08.760 --> 0:07:11.040
<v Speaker 3>does a MD wade into that arena?

0:07:12.240 --> 0:07:15.160
<v Speaker 1>Well, I need to send you an AIPC, that's for sure.

0:07:15.440 --> 0:07:20.200
<v Speaker 1>So look, aipcs are an exciting new category. It really

0:07:20.200 --> 0:07:24.240
<v Speaker 1>will change the way the PC form factor behaves, especially

0:07:24.240 --> 0:07:26.800
<v Speaker 1>for enterprises. When you think about, you know, all of

0:07:26.840 --> 0:07:30.400
<v Speaker 1>the business productivity, all the collaboration we do, all the

0:07:30.440 --> 0:07:34.280
<v Speaker 1>security that you need. AIPCS really fits that. You know,

0:07:34.320 --> 0:07:37.800
<v Speaker 1>we just announced that we'll have copilot plus enterprise PCs

0:07:38.560 --> 0:07:42.680
<v Speaker 1>with our new stricts rising AI capability. We had Microsoft

0:07:42.720 --> 0:07:45.360
<v Speaker 1>on stage to talk about some of the new experiences there.

0:07:45.760 --> 0:07:48.880
<v Speaker 1>You'll see plenty of aipcs from a MD this year,

0:07:49.120 --> 0:07:52.000
<v Speaker 1>and that even increases as we go into twenty twenty five.

0:07:52.080 --> 0:07:55.120
<v Speaker 1>So over you know, one hundred plus enterprise platforms with

0:07:55.240 --> 0:07:57.080
<v Speaker 1>best in class AI capability.

0:07:58.000 --> 0:08:00.960
<v Speaker 3>Lisa, let's end the conversation looking to the future. You

0:08:01.160 --> 0:08:05.480
<v Speaker 3>celebrated ten years a year ago. You set out aggressive

0:08:05.560 --> 0:08:09.960
<v Speaker 3>targets for your company and the industry. You sat next

0:08:09.960 --> 0:08:13.960
<v Speaker 3>to me and outlined them. Where do you go next?

0:08:15.000 --> 0:08:17.280
<v Speaker 1>Well, I would say, first of all, you know, this

0:08:17.440 --> 0:08:20.560
<v Speaker 1>is an incredibly exciting time to be in the industry

0:08:20.640 --> 0:08:23.360
<v Speaker 1>when I think about what we have in front of us,

0:08:23.800 --> 0:08:26.480
<v Speaker 1>the promise of AI, you know, the fact that high

0:08:26.520 --> 0:08:30.240
<v Speaker 1>performance computing is you know, at the center everything in

0:08:30.280 --> 0:08:32.839
<v Speaker 1>the world, and we're able to really use the technology

0:08:33.280 --> 0:08:35.920
<v Speaker 1>for good, you know. That's super exciting for me. And

0:08:35.960 --> 0:08:39.080
<v Speaker 1>then we have this humongous opportunity when you think about

0:08:39.160 --> 0:08:42.120
<v Speaker 1>AI overall, and AMD is the only one in the

0:08:42.160 --> 0:08:44.520
<v Speaker 1>industry that can really bring together all of these end

0:08:44.559 --> 0:08:47.640
<v Speaker 1>to end AI pieces. So it's a super exciting time

0:08:47.679 --> 0:08:50.200
<v Speaker 1>and I'm really looking forward. I'd like to say we're

0:08:50.320 --> 0:08:51.199
<v Speaker 1>just getting started.

0:08:52.520 --> 0:08:55.600
<v Speaker 3>A m D CEO, Lisa Sue, thank you for your time.

0:08:56.200 --> 0:08:57.160
<v Speaker 3>Scarlet Vonnie