Q&A: Stacy Rasgon on Intel, Earnings, and Why This Crypto Cycle Is Different for the Chips Sector - Bloomberg

jumianta.blogspot.com

It’s been a while since we’ve done a semiconductors episode of the “Odd Lots” podcast. But one of our recurring guests, Sanford C. Bernstein analyst Stacy Rasgon, has helped us break down the sector and talked to us in the past about the struggles at Intel and the broader chip shortage.

So to get an update on the sector, I caught up with Stacy via an IB conversation to talk about Intel, the upcoming earnings season and other things he’s watching in the sector. Below is a barely edited version of our conversation.

Joe Weisenthal:
So to start: Where do things stand with chip shortages? If you zoom out, do strains generally appear to be easing, or are things still tight?

Stacy Rasgon:
Supply broadly is still tight. Though it varies by product, I think. For example, in PCs there is talk of shortages limiting supplies, but it's mostly trailing node stuff (power management chips, etc.) rather than processors. Auto supply is still tight, and seems mixed by vendor (for example Tata had some negative news on that front this week).

In general, though, there is hope that supply constraints will begin to ease a bit in the 2H overall

Joe:
So acute constraints aside, obviously one of the bigger themes is just tons of demand. And that flows through to scarcity at the foundries, like TSMC. And that of course flows through to huge plans for capex spend. Is there any sign of that momentum slowing down?

Stacy:
So far demand appears robust. Though of course it is a little hard to disaggregate (tight supply tends to stimulate more demand as customers scramble to secure what they need). But overall demand signs across the board, in general, still look good.

For some context, total semi industry sales rose 30% YoY in May … And are still on track for a very strong year.

We'll have to wait and see what next year brings of course.

Joe:
So Intel reports earnings in a couple of weeks. Its stock had rallied pretty nicely from November through April, but lately it's been more subdued. How are investors thinking about the company and its turnaround plans these days?

Stacy:

There have been some recent issues.

1) The company (just over the last week or two) delayed their next-generation server part (called "Sapphire Rapids"), the first time that has happened under Pat Gelsinger, and bringing into question the state of the roadmap.

2) Investors are getting somewhat worried about the sustainability of PC demand given how incredibly strong it's been (unit shipments this year are likely to meet or exceed the prior peak, which was hit in 2011, and are likely to be hugely above pre-Covid levels).

2a) I also think there is evidence that the supply chain has been stockpiling processors, given shortages in other areas (our analysis suggests that as strong as PC shipments have been, CPU shipments have been even stronger and appear to be overshooting).

3) The company has scheduled their analyst day in November, at which point we may get a better idea of what the "trough" economics during the transition might look like, with worries over a (potentially substantial) reset.

 4) On the more positive (such as it is) the company has been lobbying hard for governmental funding (a popular cause), and is likely to get a good amount of dollars at some point.

Overall, though, I suspect Gelsinger's honeymoon period is over. In particular the server pushout, while maybe necessary, is not a good look, and if Pat is pushing things out instead of pulling them in, it shows the roadmap is in bad shape. I like Pat a lot, but he's not a magician...

Joe:
Yeah, let me ask you about that a bit more. So obviously the first time we ever talked on the podcast, last year, it was about this specific issue. Intel's manufacturing woes. If I recall, the plan under Gelsinger has been to reinvigorate its production, which initially investors got excited about. What is the roadmap, and beyond this one server, how soon will we know whether the rest of it is on track?

Stacy:
I'm not sure the manufacturing issues are at the heart of the Sapphire Rapids pushout. It's made on 10nm, not 7nm, and while I still don't think 10nm is super, I don't think it's the root cause.

Joe:
Ah I see.

Stacy:
They talked about new features the chip had that required additional validation as the cause. But who knows? The way they announced it was kind of iffy.

They did it in a blog post, with the pushout buried in the last paragraph. It was kind of "Look at how great this new product is going to be." "But we need more time to deliver that greatness. But customers are going to love it.”

I suspect, however, that customers would rather have it on time.

Joe:
Ha, right. So what about the rest of the manufacturing agenda. Where does that stand? And how soon will we know whether it's bearing fruit?

Stacy:
The process issues overall they claim to have fixed. They are blaming the 7nm delays on not going big enough into EUV lithography. Frankly, I can kind of see why they didn't want to use it. Their capex is now going through the roof (even without foundry). But I suppose it's better than the alternative (not being able to manufacture at all).

But even if it's "fixed," 7nm won't be here till 2023.

So they'll still be way behind TSMC, who will probably be on 3nm (which might be equivalent to Intel 5nm) by then. They won't close that gap ever, unless TSMC screws something up.

To that end, there has been fairly credible speculation that Intel will themselves be using TSMC 3nm process for some products in that timeframe as well. As a backstop.

Joe:
Ah got it. So besides Intel, what are the other big stories that you're looking to watch in the upcoming earnings season?

Stacy:
A few things:

• Overall state of the industry, supply-demand across markets etc.
• ADI-MXIM deal (ADI is expecting it to close soon, with subsequent significant increase in capital return)
• State of semi capex and impact on semicap / WFE market, especially in the memory space near term
• Pace of China handset demand
• Crypto

Lots of stuff going on.

Joe:
Yeah, was just going to ask you about that. How do you think about crypto's role in the industry? Obviously, it seems unpredictable and fluctuates along with prices. And how much has it made capacity at the foundries even tighter?

Stacy:
Crypto semis come in a couple of flavors. Bitcoin is mostly mined using custom chips/rigs. During the last cycle those took up a material amount of foundry capacity. Not so much this time around...

The other is, of course, GPUs, which are typically used to mine Ethereum, as well as many of the altcoins. That is a controversy for the likes of NVDA of course, who have seen massive demand for their gaming GPUs, and who are supply constrained.

The worry is, of course, that it's the miners who could be driving all the demand. This is what happened during the last cryptocycle back in like 2018.

However, I do think there are a number of differences this time around:

First, crypto is a bit more mainstream today. Second, the last time NVDA's gaming cycle during that period (when they launched Turing) was fairly lackluster, so when the crypto demand eased there wasn't a lot of gamer demand to backstop it. This time around, however, gamers are falling all over themselves to buy the new gaming parts (Ampere from NVDA, and Navi2 from AMD), so as the miners (eventually) ease it seems far more likely that there will be genuine gaming demand to step in.

NVDA is also developing crypto-specific cards. They did this last time too, but are marketing them much more aggressively this time around, as well as trying to limit the mining efficiency of their new gaming cards (to try to push the miners to buy the crypto-specific ones.)

Joe:
I'm curious. Has basically every sell-side chip analyst been forced to get to know the crypto market?

Stacy:
I mean if you cover NVDA or AMD or the foundries, then probably. And even if you don't you are going to get questions on it from clients so you'd better have an awareness at least.

Joe:
If crypto were to crash and stay depressed for a long period of time, would that free up graphics cards for gamers? Obviously it's pretty well known how frustrated they are. Or does, say, NVDA's segmentation of the market with crypto-specific chips take care of that somewhat?

Stacy:
I think there is still a good amount of pent-up demand from gamers so that is the hope.

You can look at prices. They have come down somewhat, but most of the cards are still selling for WAY more than MSRP which suggests supply remains very tight.

Joe:
Ah yeah, I like checking out the after-market prices on StockX. OK, two more quick questions.

NVDA more broadly. Unlike Intel, this stock has been an absolute rocket ship. Just an incredible run. Are they unstoppable? What is it about what they're doing that has investors so convinced by their story?

Stacy:
I'm hesitant to ever use the word "unstoppable" in any context. But yes, they have an extremely strong position.

They own the market for AI training. They are extending that lead into other markets (inference, edge, etc). Building software layers to increase the value-add and usability to customers, as well as strengthening their moat. Aggressively going after new opportunities (the Arm deal is a bold effort, whether they manage to pull it off or not). Beginning to build a broader narrative around monetizing software.

Lots here that is appealing.

Joe:
Yeah, can see that. OK, last question then I'll let you go. You mentioned earlier the public investment. The money that will help Intel and others. What is the status there? How much is that going to move the needle and how do you see it shaping domestic investment in the space?

Stacy:
Current numbers here in the U.S. is $52B (about 80% for manufacturing) over 5 years.

It sounds like a lot, but frankly it's a rounding error in this industry.

If the goal is to materially increase the amount of manufacturing capacity that is installed in the U.S. (currently ~12%) it will do close to nothing :)

If we are really serious about significantly changing the status quo, it will require 10-20x that amount, and 10-20 years, and broader efforts (customer acquisition, immigration reform, education etc).

And I'm not sure there is appetite to go that big (yet). But you have to start somewhere. And this is a start.

Adblock test (Why?)

"cycle" - Google News
July 07, 2021 at 08:25PM
https://ift.tt/36kaUJC

Q&A: Stacy Rasgon on Intel, Earnings, and Why This Crypto Cycle Is Different for the Chips Sector - Bloomberg
"cycle" - Google News
https://ift.tt/32MWqxP
https://ift.tt/3b0YXrX

Bagikan Berita Ini