Intel Looks Good Near Term: New Street Research

Bloomberg

Watch Intel Looks Good Near Term: New Street Research - Bloomberg on Dailymotion

Transcript

00:00What did you make of the numbers? Was there enough there to keep the share price rallying as it has

00:03been? Yeah, so the numbers for Intel on a day like yesterday are about relatively short to

00:13near-term trend. And so the PC market has been very weak for some time, meeting actually

00:19expectations that were very high because everybody was very excited about the AI PC and things like

00:25that, which didn't really move the needle for the market. And so now what we see is a more

00:30traditional driver, like Windows refresh cycle, helping enterprise demand. And that's great, of

00:35course. And then servers, we've seen the server market being weak for quarters and quarters, if not

00:41years, because you had so much focus on deploying AI servers that a lot of very large players were

00:48kind of deprioritizing the deployment of traditional servers. And now we are entering a phase of

00:54catch-up. So the near-term for Intel looks good. And it reminds you that it's a good business, that

00:59there is operating leverage, that even if they've lost a lot of market share to AMD, they are still

01:04like a market leader. So that's great. But these are very surprisingly positive, but relatively

01:11short-term. And that's the way I interpret the stock movement, you know. Yeah. Big risk on the

01:16news flow. And then you look back and you're like, well, but at the end of the day, what does really

01:20matter for Intel? Is that really like PC demand over the next couple of years? Or is that more like...

01:25So what matters, Pierre? What matters is 14A. What matters is, are they going to be the cutting edge of the

01:31next iteration of chips? Are they going to really rival TSMC and get others than Intel's own fabrication

01:37going? What did you hear from Litboutin and the team about that?

01:41So I heard two things. The first one is mixed signal, which is that 18A, the manufacturing is ramping,

01:50is going to get into volumes imminently, you know, like for the beginning of next year. But comments about

01:57the yield. So, you know, the quality of the process are still very, very mixed. Let's face it, the yields are not good

02:03today. So early yields not being good is almost business as usual. But I think 18A's performance in

02:10terms of manufacturing performance is disappointing. And then the second piece of news that really

02:16matters the most for the long term of Intel is like more positive comments, but 14A. Remember, three months

02:23ago, the CEO had very cautious comments even mentioned in writing in the 10K as a possibility of not

02:30developing 14A if Intel didn't find enough customer support to do it. And now they're talking

02:36positively about 14A on the technical front, like the roadmap is progressing well. And also, I don't

02:42even want to call that the commercial front, but more like the coalition front. This idea that if you

02:47really want Intel to be successful in manufacturing, you need more than clients. You need industrial

02:53policy. You need a coalition. You need partners that deeply engage with Intel, invest in Intel,

02:59commit to Intel to work with Intel. That's the only way the foundry is really going to do well.

03:04And honestly, on that front, it's very, very early signals, but I think they are positive.

03:10I tried to get a sense from Intel CFO, Dave Zinsner, when I spoke to him on the phone about

03:16that long term you were talking about. And in an AI world, AI, his answer was quite simple. They do see

03:23opportunity in accelerators. But the foundry bit is what I was like most surprised at. He's

03:28saying that the foundry opportunity for them in AI is in wafers and packaging.

03:33That doesn't sound hugely exciting, Pierre.

03:39Well, maybe it doesn't sound hugely exciting, but it sounds very realistic and really playing to

03:46Intel's strengths. The way I see Intel today is they have two major product franchises. One is in the

03:53the PC business, x86 CPUs for PCs and wasn't in server. So I'm pretty sure the management wants to

04:00maximize the value of these two franchises and concentrate on them. And it's good. You can't

04:05invent yourself, you know, like an AI accelerator designer like 15 years down the line or 10 years

04:13down the line. And then, yes, Intel has an edge in packaging. Intel is a strategic asset for this

04:22industry. And this is probably what management wants to build on. And I see that as maybe not as exciting

04:29as willing to compete against Nvidia, but very realistic and the right industrial approach, probably.

04:35You know, and for our audience, Pierre, I appreciate that answer because you also cover Nvidia, right?

04:40You were able to make the comparison. On x86, I think the argument Intel were making to us was

04:47the hyperscalers have woken up a bit to the value of having the latest CPU and infrastructure. They're

04:53willing to invest and refresh on CPU. So again, for them, Intel, they would argue that it's a great

04:59opportunity for them. Do you see that opportunity? Yes, absolutely no doubt. And it's very, very

05:08strong. You have like clients asking for it, the specifics of the x86 architecture and the way

05:15Intel implements it is very differentiated. I've always been a great believer in the quality of the

05:22products of Intel on that front, like the ability of an Intel x86 chip to deal with very unforeseen

05:28situation in very complex workloads is excellent. And you've had so much demand for it that it actually

05:37triggers Nvidia's move first to open up NVLink, like the high bandwidth connectivity with their GPUs so that

05:47x86 chip can be integrated into systems and a $5 billion investment in Intel and a partnership with

05:56Intel to co-develop chips that would be optimized to get integrated into Nvidia systems. And my read,

06:03having followed that for a few years, is that it's not just like a nice gesture from Nvidia under

06:09political pressure or anything like that. It is a genuine thorough and deep interest for Intel products

06:14in the industry for that role. Now, keep things to the scale they are out on NVL 72 server powered by

06:22x86 chips. The x86 chip is going to be single digit percentage of the cost of the overall server and

06:30the other 90 plus percent will be Nvidia. So, Pierre, with Nvidia's support, and we know the money's yet to

06:36come through and that'll settle in the next quarter, will that be enough to see Ohio break ground? Will we

06:42see an actual commitment not just to co-develop, but actually to eventually fabricate on site in the

06:48United States, not just relying on TSMC? It's a great question. You need more than that to have

06:55like a final commitment or to claim victory, basically. In order for Intel to successfully

07:01manufacture chips in the US, package chips in the US for US clients, you need the money and

07:10major progress made on that front with money from the government, from Nvidia, from Southbank,

07:14and I'm sure more will follow. You need also like the multi-year commitment of a coalition. So,

07:24you're not going to get like the right chip in the first go with Intel. It's a dialogue that needs to

07:30last very multiple years before you get to success. And last but not least, you need execution. You need

07:39intent to come up with a 14A node that is going to show up, like progressively be unveiled and better

07:48convinced and make potential adopters more confident. It's going to be the right not to do something with it.

Category

Transcript

Be the first to comment

Recommended