00:00It's kind of funny when we had this big rally in these chip names over the last few months,
00:04it was kind of the opposite effect. You've seen it had this euphoric sentiment and everybody was
00:07wondering, are the fundamentals going to match it? Now we're getting proof of life that the
00:11fundamentals are there, but everybody, for some reason, still seems to be looking for the door.
00:15Is this just profit taking or is there something more significant going on with regards to
00:20investor sentiment? The profit taking is a part of it, but investors just overall are concerned.
00:27And there's a notion in a sense that because we're spending so much, because earnings for
00:32these companies have gone up so much, that can't possibly continue. And we would argue that we are
00:38seeing signs that there is a lot of activity, a lot of real economic activity happening around AI.
00:45If you add up the revenue run rate for OpenAI and Anthropic and maybe pieces of Google and
00:51Meta that are relevant, XAI that are relevant, you get to about $100 billion revenue
00:56run rate. That number was zero two years ago. So clearly there is real economic activity happening,
01:03which is to say the companies that are buying chips will likely continue to buy them. And
01:09specifically in memory, we're not making nearly enough chips and we won't be making nearly enough
01:14chips at least for the next year or two. So the fundamentals are still very good. Indications are that
01:21they stay that way. But investors that have gone through this long ride are now worried that maybe
01:28the ride is coming to a net. Well, I am curious. I mean, we are I mean, we got the
01:32update from TSMC.
01:34We get the full earnings report in a couple of days. ASML over there in Europe also scheduled to sort
01:39of
01:39put its cards on the table, I believe, on Wednesday, US time. Is there a sense here that we can
01:45divine
01:46something from those earnings reports that feeds into what we would expect out of Nvidia and Qualcomm
01:53and Broadcom, et cetera? Yeah, absolutely. And those results are very good and continue to be very good.
01:59In fact, again, all semis, all the semis that are reported anytime recently have reported accelerating
02:05revenue growth because the purchasers of compute, the Amazon, Microsoft, Google, Meta, et cetera,
02:13are buying more and more. And they're likely to continue to say that. So in a couple of weeks,
02:18Google report, the Microsoft report, what they're going to probably say is, look, our CapEx is probably
02:23going to go up next year. It won't go up as much as it did this year, but it's still
02:28going to go up,
02:29which is to say the demand for semis will will continue at least into next year. Now, we what happens
02:36after that? Obviously, there's a big debate in the market. Are we close to a peak? Is the peak in
02:412030?
02:42We see the opportunities in the stocks that are priced as if the cycle is peaking. And we think
02:48the stocks that are overvalued are the ones where there's an assumption that when the cycle peaks
02:53in 2030. So Nvidia, Micron, SK Hynix are priced as if we're rolling over in the cycle. That makes
03:00them the opportunity. Intel, Cerebrus, a lot of the semi-cap optical nuclear names are trading as if
03:06the party's going to go on at least till 2030. That's where there's more risk.
03:11Well, Gill, you're touching on something that I've been wondering about. You know,
03:14once we roll into some of these reports from the big heavyweights in the tech space,
03:19you know, and you hear in your expectation that they're still committed to their CapEx
03:24plans, do you think that will be enough to reset sentiment? Because again, as Romain pointed out,
03:30as you've pointed out, the fundamentals are there right now, but it's just the stock that's not
03:36responding. Yeah, I think they need to thread the needle. So what I would expect, let's use Microsoft
03:42as an example, because it's their fiscal year end. So they're going to guide 12 months ahead
03:46on CapEx and revenue growth. If they say Azure is going to grow more than 40 percent, it's going to
03:53accelerate. And CapEx is going to decelerate, grow less than last year and less than 40 percent.
04:00That's a very positive message. That means AI is working. Microsoft's getting a return. And they
04:06don't need to continue to accelerate CapEx in order to achieve that. If they can deliver that message
04:11effectively, it should help the entire AI sector. Well, Gill, you also made the point that you think
04:19about how some of these names are trading. They're acting as if, you know, things are already
04:24rolling over. And I feel like one of the narratives that's haunting some of these memory stock names
04:29right now is that memory typically has been a very cyclical industry. And you've heard that from
04:36maybe some more bearish commentators on the street. And I wonder whether you think that the sector
04:42overall has been able to escape some of those patterns that it has historically had.
04:49All semis are cyclical. But I would argue that now, as we sit here today, memory may be the least
04:56cyclical, at least for a company like Micron that got that went up into its earnings call and said,
05:02look, we signed five-year deals. We exchanged a little bit of the upside to eliminate the downside.
05:07So all semis are cyclical. GPUs are cyclical. TPUs are cyclical. CPUs are cyclical. But memory is now
05:15being sold in five-year deals. So it is actually less cyclical than all those sectors. And yet,
05:21the long-term memory is persistent here, which is why these stocks are trading at such a discount.
05:27CPU stocks trading at 40 times earnings and more. Memory stocks trading at eight times earnings or less.
05:33In spite of the fact that I would argue CPUs are now more cyclical than memory.
05:39Well, it gets to this idea then. I mean, if we are seeing a real transformation here in that
05:45relationship, Gil, do you have confidence, though, that that is indeed, are we going to be talking
05:50about this 10 years from now in the same manner, meaning the idea that it's a bit more of an
05:55acyclical or less cyclical industry? Or are we kind of diluting ourselves into just what is
06:01basically still a cycle, just a much more prolonged one?
06:05Yeah, I'm not arguing that it's not a cycle. This is just the nature of things that when you
06:10build a lot of supply, sometimes you build it ahead of demand. There will be a cycle at some
06:14point. It's just the level that the cycle is happening at is much higher. And again,
06:19for a company like Micron that's using its current market power to lock in five-year contracts
06:25and eliminate the downside from the cycle, that's a much better view. So the cycle is at a much
06:32higher level, and they've eliminated the downside. That's why we find them so compelling right now.
Comments