주목해서 읽어야 할 부분
AI 반도체 수요와 변화 즉, 피크아웃 우려.
전력과 관련된 이야기가 나오는지
고객사의 이야기.
요약
중국에서의 수요 감소는 관세 부과 시작 전 일시적으로 증가한 주문과 중국 수출 라이센스 불허가 문제 때문임.
그러나 라이센스 문제가 왜 갑자기 이제와서 생겼는지에 대해서 경영진은 즉답을 피하며 충분히 설명하지 못함.
최첨단 로직과 메모리 쪽 수요는 여전히 매우 강함.
따라서 중국의 수요 감소가 반도체 사이클이 현재 피크임을 알리는 위험 신호의 ‘시작’인지 미중 분쟁으로 인한 중국에 국한된 현상인지 확인이 필요함.
아쉽게도 에너지 관련 이야기는 나오지 않았음.
하이라이트
Prepared Remarks
we expect revenue and earnings to be sequentially lower in our fourth quarter, primarily due to uncertainties in our China business.
For Applied’s business, there are 3 main factors that mute our outlook for the quarter ahead. First is digestion of capacity in China. Second is our large backlog of pending export license applications, where we have taken a conservative position and assumed none of these licenses will be issued in the next quarter. And third is nonlinear demand from leading-edge customers, which is primarily linked to market concentration and fab timing.
Compared to our expectations at our last earnings call, we saw slightly less-than- anticipated growth in leading-edge spending due to a slower fab build-out schedule and stronger-than-expected spending in ICAPS.
Compared to our expectations at our last earnings call, we saw slightly less-than- anticipated growth in leading-edge spending due to a slower fab build-out schedule and stronger-than-expected spending in ICAPS.
As a result, we are seeing customers take longer to commit to orders, leading to a shorter visibility window. With these shifts in mind for fiscal Q4, we expect total revenue of $6.7 billion, plus or minus $500 million, representing a 4.9% decrease year-over-year at the midpoint and non-GAAP EPS of $2.11, plus or minus $0.20, representing a 9% decrease year-over-year at the midpoint.
Q&A
James Edward Schneider
I was wondering if you could maybe talk to the incremental source of weakness in the outlook. Specifically in China, do you see the decrease in visibility there being extended well into 2026?
Brice A. Hill
And we actually expect this to continue for several more quarters as we look forward.
You’ve got CapEx going up at cloud service providers. You’ve got a lot of pull from AI-related technologies, especially on the leading edge in DRAM, in HBM also. And so we expected and modeled a relatively linear ramp that would accelerate through 2025 and into 2026. And we’re not seeing that in the order pattern for Q4.
Now with one customer being larger than the other customers, it’s more attached to their ramp and their loadings. So that’s a factor in the unevenness of the ramp going forward.
Stacy Aaron Rasgon
China is 35% of the revenue in the current quarter. I didn’t get the impression last quarter that you expected China to be that strong. I think it was 25% the quarter before. So I guess the question is, was China significantly stronger than you expected? And if that’s the case, it feels like the non-China business feels weaker than you were sort of alluding to in your comments. Just can you give me a little more color on what your expectations had been for China relative to where it came in, in the quarter relative to the rest of the business?
Brice A. Hill
And yes, from our expectation, you don’t know until the end what the percentage will be, but we expected it to be higher. And yes, no real change in the flow from that perspective during the quarter. The only thing that I highlighted in the script was there was a little bit less activity on leading edge in the quarter, which we highlighted.
So the change for us, even in our outlook, we expected China to be lower in Q4.
핵심 질문
Stacy Aaron Rasgon
Got it. I mean if I dig into that a little bit, so if you’re guiding China to like 29% of revenue, versus $35 million. So that’s — I don’t know, it’s a $600 million decline. Display is $100 million, give or take, stronger. So it’s something like a $500 million decline in China equipment. And then the total equipment guide is down $700 million plus. So that incremental $200 million to $250 million, is that like the lower leading edge? Is it all — so it’s all in like the foundry/logic leading? Like how do I think about like the components of that guide?
Brice A. Hill
I think you got your equation right. So approximately $500 million China, approximately $500 million leading edge, which I alluded to with a $4.5 billion instead of $5 billion. And then there’s some upsides in rest of world ICAPS that explain the difference that you’re looking for.
Stacy Aaron Rasgon
Do you think the non-China leading edge is down another $500 million sequentially? Or that’s what you’re saying? Or that was an annual number?
Brice A. Hill
That was not. So reconciling the down of $700 million, I would say it’s $500 million China, $500 million leading edge and then a plus factor for rest of world ICAPS.
Stacy Aaron Rasgon
Okay. Where is that coming from though?
Brice A. Hill
Well, we’re not going to be specific. It would be too indicative of a particular customer, but it’s not China.(아마도 TSMC!!!!!)
Vivek Arya
Gary, the question is for you. So both kind of near and medium term. So if you look at Q1, if these trends persist, should we be assuming any sequential growth into Q1, right, based on what we know.
Brice A. Hill
So you’ve got the leading-edge memory players all growing nearly 50%, as Gary said in his script. So DRAM is very strong.
Christopher James Muse
I guess I was hoping to probe the different areas of weakness that you cited for October. So for the first part, with China, it sounds like for many of your peers that the concentration is the 6 large players who don’t require licenses. So I’m curious why your shipments could be so robust through July, but now in October, licenses are now the challenge.
Brice A. Hill
But just so investors understand, there’s new fabs, there’s lots of new investment in China. It’s just not quite at that 2024 rate.
And I’ll just say that our model had a more linear assumption than reality. So no significant changes to point to in customer availability.
Harlan L. Sur
Can the team just throw us up like what type of growth are you expecting this fiscal year in advanced packaging?
Brice A. Hill
we’re seeing about a steady pace relative to last year, minus the burst capacity, initial burst capacity we saw in HBM at the end of last year. So I’d say similar to last year from a pacing perspective for that business. And I — we don’t see any change related to the leading-edge schedules and the purchasing behaviors there.
Gary E. Dickerson
And what I talked about earlier on the call is that we’re on track to more than double this business going forward to more than $3 billion per year.
Yu Shi
I noticed that you actually upsized your leading edge DRAM growth for the fiscal — current fiscal year from 40%, which was what you said last quarter to 50%. But the fiscal Q3 DRAM number, you said it is slightly better than you expected.
Brice A. Hill
Yes. Thanks for the question, Charles. Yes, I certainly don’t want to draw the conclusion that there’s any pop or some unusual increment there. We view DRAM as very strong.
Vijay Raghavan Rakesh
Just to follow up on the previous question. In terms of the ramps, given your more than 50% share on gate-all-around and backside power, when do you see that start to get more material, if you can give us some more color there?
Brice A. Hill
we do expect investment to increase — when we think across the long term, we have leading-edge logic and DRAM as the 2 fastest-growing end markets.
