So this notion of enterprise, our enterprise customer base growing, we’ve sold to education customers, manufacturing customers, governments. We’ve sold to financial services, business, engineering and consumer services companies. They’re seeing vast deployments, proving out the technology. And some cases are using the tooling of the public cloud. And then they quickly find that they want to run AI on-prem because they want to control their data. They want to secure their data. It’s their IP and they want to run domain specific and process specific models to get the outcomes they’re looking for. Hopefully that gave you a rich context of what we’re seeing across the customer base, the opportunity for us going forward. And this is at node scale, rack scale, and data center scale.
Ananda Baruah: Thanks a lot.
Jeff Clarke: Thanks for the color, Ananda.
Ananda Baruah: Thank you.
Jeff Clarke: You’re very welcome.
Operator: We will take our next question from Wamsi Mohan with Bank of America.
Wamsi Mohan: Yes, thank you so much. You called out pricing pressure in AI servers. How are you responding to that? You think that you’re leaving any revenue or orders on the table if the financial returns were not acceptable or are we not anywhere close to that point yet? And if you could just clarify this acceleration in ISG growth that we’re looking at in next quarter. This quarter, server revenue growth was still relatively weak within servers and networking despite a very strong 800 million in AI servers. So just hoping to think through how you’re expecting the AI or non-AI server component of that in that ISG growth over the next quarter. Thank you.
Jeff Clarke: Maybe a couple of market descriptors and then Yvonne can weigh in on the financial impact. Well, the competitive market in PCs increased quarter-over-quarter and particularly in low price bands. So when we made remarks about us being selective or focusing on profitability, that’s how to decode it in the PC business is that this sub $500 market opportunity is certainly less profitable and we were much more guarded of how much we participated in that, which led to our share. We actually took share in mid-range and high-end price bands in mature markets. So I think that’s a component of this. When I look at AI and then the comments around AI and margins, our margins actually improved quarter-over-quarter with AI. That’s an encouraging sign.
There’s still less than traditional servers on a rate basis, but improving it’s the second consecutive quarter of improvement. So that’s some color about what’s happening in the marketplace. We did see in traditional servers that in large bids, the competitiveness did increase quarter-over-quarter in Q4. We expect that to continue. That’s not uncommon. Yvonne?
Yvonne McGill: No, I would. As we look at the holistic server portfolio for Q1, we’re seeing obviously strong growth driven by the AI servers, but we’re seeing growth, as we’ve talked about, sequential growth over traditional servers. It’s close. We’re not seeing, we’re expecting growth to return year-over-year in traditional servers, but it’s a very competitive market, as Jeff mentioned. What I’m really excited about is the other thing that Jeff talked about on really getting more and more value out of our GPU servers, really with, as we move more and more into the enterprise and get more richly configured, more services, etcetera, attached to that.
Jeff Clarke: That’s the path. There’s storage, deployment services, pro support, our consulting services, networking, so in the entire basket of the solution.
Yvonne McGill: But it is a competitive environment out there. So, we’ll continue to maximize our opportunity.
Jeff Clarke: Yes, so modest growth in traditional, stronger growth in AI-enabled servers, and an opportunity with storage as the year progresses. Thanks, Wamsi.
Operator: We’ll now take our next question from Samik Chatterjee with JPMorgan.
Samik Chatterjee: Hi, thanks for taking my question, and Yvonne and Jeff, thanks for all the comments. Yvonne, maybe this is more for you. Just how should we think about the higher memory costs playing through the gross margins, particularly in the fourth quarter itself that you reported? And then as we think through fiscal 2025, are the headwinds more in the sort of early quarters? And how much of that is sort of baked in in relation to a headwind for the fiscal 2025 guide that you gave for 100 basis points moderation and gross margin? Thank you.
Yvonne McGill: Thanks. We have taken into account all the information that we have available and have embedded that into the guide, right? So we are expecting it to be an inflationary environment going forward, and we will price that to the best of our ability in the market. Obviously, it’s a competitive market. So the guide embeds those expectations. And, I don’t know if Jeff, there’s anything else you would add to that.
Jeff Clarke: Not at all. Well, we know we have planned that into the guidance that we provided both for the quarter and the year.
Yvonne McGill: And how we’re pricing deals.
Jeff Clarke: Absolutely.
Yvonne McGill: We have flexibility to it, and we’re doing what we need to do to maximize our profitability.
Jeff Clarke: Good. Thanks, Samik.
Operator: And we will now take our next question from Amit Daryanani with Evercore.
Amit Daryanani: Thanks for taking my question, and congrats on a nice set of numbers here. I’ll speak to the AI team, because that seems to what matters right now. And so, Jeff, as you think about sort of the AI backlog of 2.9 billion that you’re sitting on, how do you think about that converting to revenue? Sort of what’s the cadence to convert that to revenues as you go forward? And then as you folks have talked about this mid-teens growth in ISG, could you maybe parse out and tell us how much you think will be storage versus traditional servers versus what is going to be driven by the AI servers for the whole fiscal 2025? Thank you.
Jeff Clarke: Well, Amit, there’s no easy answer to your question, because the backlog is a mixed array, as I mentioned earlier, of H100s, H200s, H800s, as well as MI300Xs, with varying supply commitments and varying delivery commitments, as well as converting new demand of this five-quarter pipeline that I just mentioned. We believe we will ship more in Q1 than we shipped in Q4. As we look forward in our annual guidance, Yvonne [ph] has our best estimate of our demand and fulfillment of that demand that we’ve put into the annual guidance. To be able to parse it down in more detail, we have a product transition that’s in front of us that we have to work on, H100, H200 to be specific, and we’re taking orders on the new stuff as well as converting the current pipeline on the current product, the XC9680, with H100s in addition.
