CEOInterviews.AI
Start App
Get an email when Lisa Su gives a new interview
One email per new interview. Unsubscribe in one click.
Lisa Su
Chair, President & Chief Executive Officer, Advanced Micro Devices

Lisa Su: Why AMD Demand Exceeds Supply, Helios and the $1T Data Center Race | CNBC Full Interview

📅 Nov 05, 2026 Vampyre Drakul 13 MIN 1218 VIEWS 25 SEGMENTS · 3 SPEAKERS
AMD Chair and CEO Lisa Su joins CNBC to explain why demand for AI compute is accelerating faster than the company expected. In this full interview, Su discusses AMD data center growth, the launch of Helios racks at Anthropic, OpenAI, Microsoft, Oracle and Meta, and why AMD sees demand exceeding supply across the semiconductor ecosystem. Su explains how agentic workloads, personal AI assistants and frontier models are driving demand for both CPUs and GPUs. She describes the performance advantages of Helios and MI455 accelerators, the launch of the Venice server processor, and how AMD is workin...

What Lisa Su said

Written from the verified transcript and checked against it. Every figure links to the moment it was said.

Lisa Su, AMD's chair and CEO, said demand for compute is very strong, with data center revenue growing over 40% year-over-year in Q3 and expected to grow 50% sequentially in Q4. She said demand exceeds supply, and AMD has landed first racks of its new Helios products at customers including Anthropic, OpenAI, Microsoft, Oracle, and Meta. Su forecast data center business growth of over 80% next year, more than double the long-term model, with the majority under contract. She said the supply chain constraint is end-to-end, from wafers to data center infrastructure, and AMD is planning three to five years ahead. Su affirmed the $1 trillion data center total addressable market by 2030 is unchanged despite the election. On margins, she said Q3 exited at 56% and Q4 guided to 55%, and she feels good about the balance of the portfolio into 2027.

Key takeaways

  1. AMD data center revenue grew over 40% year-over-year in Q3 and is guided to grow 50% sequentially in Q4.
  2. AMD forecasts data center business growth of over 80% next year, with the majority under contract.
  3. First racks of Helios products have been deployed at Anthropic, OpenAI, Microsoft, Oracle, and Meta.
  4. Su affirmed the $1 trillion data center total addressable market by 2030 is unchanged.
  5. Gross margins exited Q3 at 56% and are guided to 55% in Q4.

Numbers and commitments

FigureWhat it refers toTypeAt
40% Year-over-year data center revenue growth in Q3 metric 0:22
50% Sequential data center growth guidance for Q4 guidance 1:24
80% Data center business growth forecast for next year guidance 1:24
100% Data center business growth this year guidance 2:40
100% Data center business growth forecast for next year guidance 2:40
56% Gross margin exiting Q3 metric 12:04

Chapters

  1. 0:00Strong Q3 and demand outlook
  2. 1:24Demand exceeds supply and customer deployments
  3. 2:40Revised growth forecasts and contract visibility
  4. 5:13Supply chain constraints and ecosystem
  5. 6:07Election impact and public-private partnership
  6. 7:26Competitive position in CPUs and AI accelerators
  7. 9:07Data center growth and local community value
  8. 10:21Neoclouds and long-term planning
  9. 12:04Gross margins and business model

Questions asked in this interview

8
  1. 2:20How would you revise that now looking at what you just did?
  2. 4:52And where is the tightest constraint then?
  3. 5:55Can you talk a little bit about how you and your team are discussing strategy in the wake of a shift in the balance of power?
  4. 6:54Can you speak to your portfolio versus the portfolio of your competitors?
  5. 8:36How worried are you that data center growth is going to slow dramatically as a result of this election?
  6. 9:44Didn't hear him speaking of the CoreWeaves, the Neoclouds, if they are the marginal builders and they don't have the capital to build, does that change your 2027 or 2028 growth rate predictions?
  7. 11:19No change to that in spite of what we saw last night in the election?
  8. 11:44Where are we when you look at gross margins?
Interviewer 0:00 ↗
And that's why it's so terrific to have now first on CNBC AMD chair and CEO Lisa Su. Lisa, so many positives, it's very hard to figure out which ones to start. How about the fact that year-over-year and quarter linked quarter, so to speak, dramatic surprises from where we thought would happen even three months ago and even one year ago.
Lisa Su 0:22 ↗
Well, first of all, Jim, David, Carl, it's great to be here this morning. It is a very, very strong time in terms of demand for compute. We had a great third quarter, you know, exceeded our expectations. You know, data center was extremely strong. We grew over 40% year-over-year. And as we look forward, we're seeing even stronger demand overall, Jim. So I think what we're seeing is, you know, again, compute is one of those things that everybody in the industry wants, and we are extremely well set up with our newest products ramping starting in the third quarter and going into Q4 and 2027.
Interviewer 1:00 ↗
I've been watching your stock from before the election after. It's obviously people feel that whatever you thought would be about data center build has got to be curtailed. I'm curious about what you think because frankly, what you have in the pipeline, I can't see a single deal or contract of consequence that would be cancelled, and what you have not in the pipeline for 2031, '32 could be also spoken for given the demand that we see.
Lisa Su 1:24 ↗
Well, we are in a place, Jim, where I think there is an intense need for more compute, and when we look at that, we are seeing a very strong order pipeline. We are actually seeing the demand being higher than supply, but we were able to guide, for example, our data center business is going to grow 50% sequentially, quarter-on-quarter, into the fourth quarter. We have actually landed our first racks at our top customers, including Anthropic, OpenAI, Microsoft, Oracle, Meta, and these customers are seeing great performance out of our newest products. So, you know, what I look at is this is an opportunity to continue to expand. We see our business growing over 80% next year, which is a very significant number. It's over double of our long-term model, and we see lots of opportunity to continue expanding because of our, you know, very strong customer partnerships as well as very strong supply chain partnerships.
Interviewer 2:20 ↗
I would also point out that you have chips that don't need as much DRAM, which I think would matter to a lot of people. I have the November AMD financial analyst day from 2025 in front of me. You talked about the compound annual growth rate of 60% for data center business. How would you revise that now looking at what you just did?
Lisa Su 2:40 ↗
Well, I think what we see is, again, you know, this year we're going to be 100% of our data center business in terms of growth. Next year we're also forecasting over 100% growth. Our server CPU business, you might have seen, for example, the newest capabilities with, you know, MetaMuse and these personal agents. I think we're seeing, again, we're at the very beginning of the adoption cycle where people are seeing where, you know, AI and super intelligence can really take us, and for that you need all kinds of compute. You need CPUs, you need GPUs, you need all of these accelerators to come together, and this is what we specialize in. So we're very, very excited about the long-term demand. You know, there is a lot of conversation about how do we ensure that we supply all of the demand that is out there.
Interviewer 3:27 ↗
You know, Lisa, I mean, you mentioned this, you're guiding to more than 80% growth next year, and it is more than double the model that you laid out, I think it was only a year ago. And so I am curious what has changed, particularly let's say even in the last 90 days for that target, and how much of it's already under contract versus just a forecast.
Lisa Su 3:45 ↗
Yeah, I would say, David, a couple things have changed. First of all, the demand for AI has continued to accelerate, so you see more compute is needed out there, and frankly we've also had our customers now also see the strength of our products. So, you know, launching Helios here in the third quarter, very, very, you know, big step up in terms of performance for us, being deployed at, you know, major frontier as well as major hyperscalers, has been a big vote of confidence. And we're also seeing, you know, CPUs, I will say, David, the agentic workloads were, you know, really just coming to fruition over the last, you know, three or six months, and even over the last six weeks you've seen just the tremendous momentum that, you know, Meta has had with Muse and the number of, you know, people who are adopting that. So all of that coming together gives us very high confidence into 2027. You should imagine that the majority of that is under contract with customers as well as under contract with our supply chain, because both sides are extremely important, and we see an opportunity to continue to, you know, ramp that demand as we go through 2027.
Interviewer 4:52 ↗
Yeah. And what we've mentioned, of course, the ratio of CPUs to GPUs when it comes to inference and agentics. All right. You've said a number of times that demand exceeds supply, Lisa. And so I'm just curious, if you got every wafer, every stack of memory you wanted, how much bigger would next year be? And where is the tightest constraint then?
Lisa Su 5:13 ↗
Well, I think, David, what we have seen is this supply chain is really about end to end. So you start from, you know, wafers to, you might have heard about substrates, advanced packaging, memory, into data centers, land, power, shell, data center infrastructure, all of that has to come together. I'm very pleased that we have tremendous partnerships across all of that ecosystem, and so we are, you know, confident in our ability to grow, you know, greater than 80% next year. I think you will see over time that we will continue to work each one of those constraints, so there's not any one thing. I think it's just an opportunity for us to continue to align customer demand with the supply chain and the data center infrastructure environment.
Interviewer 5:55 ↗
Lisa, it's obviously we led with the midterms in the election cycle, a big story of the morning. Can you talk a little bit about how you and your team are discussing strategy in the wake of a shift in the balance of power?
Lisa Su 6:07 ↗
Well, I actually think the work starts now, and it is one of these areas where, you know, I look at it as, you know, AMD as well as our peers are always working across both sides of the aisle. What I see is an opportunity for us as, you know, technology companies and technology CEOs to step up and be part of the conversation. And at the end of the day, what we need to do is ensure that everybody wins from AI. And, you know, there's no doubt in my mind that winning the AI race is extremely important for America, but I also believe it's extremely important, you know, for every region, every local community, and this is an opportunity for us to, you know, really open that dialogue. So I think it's a very clear recognition that public-private partnership is super important.

13 more exchanges in this transcript

Sign in free to read the rest of this interview. No card required.

Sign in to read the full transcript

Cite this transcript

APA, MLA, BibTeX
APA

Su, L. (2026, November 5). Lisa Su: Why AMD Demand Exceeds Supply, Helios and the $1T Data Center Race | CNBC Full Interview [Interview transcript]. Vampyre Drakul. CEOInterviews.AI. https://ceointerviews.ai/interview/3204691/

MLA

Lisa Su. "Lisa Su: Why AMD Demand Exceeds Supply, Helios and the $1T Data Center Race | CNBC Full Interview." Vampyre Drakul, 5 Nov. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3204691/.

BibTeX
@misc{su2026_3204691,
  author       = {Lisa Su},
  title        = {Lisa Su: Why AMD Demand Exceeds Supply, Helios and the $1T Data Center Race | CNBC Full Interview},
  howpublished = {Interview transcript, Vampyre Drakul. CEOInterviews.AI},
  year         = {2026},
  month        = {nov},
  url          = {https://ceointerviews.ai/interview/3204691/},
  note         = {Speaker-attributed transcript with timestamps}
}