00:00We're off ahead of these numbers, but a lot of optimism baked into Micron and memory makers
00:05right now. Absolutely. And the price shows it. And there is a pattern to Micron's earnings,
00:11right? That, oh, maybe a month before people, and it's probably in reaction to Micron's customers
00:18talking about, you know, what they're buying from Micron. The stock generally, and when things are
00:25good, this is a very cyclical stock, but when things are good, things are very, very good for
00:30Micron. And I think in the short term, things are going to be really, really good for Micron.
00:36But often the stock sells off right after, just because of that buildup that has happened,
00:42you know, in the recent past. So keep that in mind. I mean, here today, Kim, the stock's up 168%.
00:49So we've got a little bit of a dip into the earnings. It's kind of nothing in terms of
00:53its performance. But what are you looking for after the bell in terms of signals that the
00:58strength is there and that we don't keep questioning the AI bubble as we know it?
01:02So there are very few providers of the big product that AI wants, which is high bandwidth memory,
01:11DRAM. And there's only two other players in the market. Micron has had a nice lead. The other
01:17players are coming up, but it looks like for at least the foreseeable future, infinite demand
01:24is there for this product. And I'm teasing, okay? I don't really believe it's infinity,
01:30but it's a very high demand. So we would like to know what that's doing. And you can see how Micron
01:36is shifting away from its older products and its more consumer-related NAND devices. They announced,
01:43you know, one of their longtime franchises, they're going to be closing that in favor of putting all
01:49its effort behind DRAM. So that's what I'm looking for is what is the, why do they have that confidence
01:56and why should we as shareholders, you know, go along with the ride?
02:01Because right now there is shaky confidence around certain names about the AI infrastructure
02:07build-out. Do you have shaky confidence about certain newer players to cloud, newer areas and data
02:12centers? Well, this is going to sound like a Miss America kind of speech. I have a real,
02:21I know AI is going to work out. I just don't know how. The large language models seem to be consumers
02:28of tons of bandwidth, which you talked about on your last segment, or not, you know, computing power,
02:34not bandwidth. But I don't know that that goes on forever because this whole scaling thing where just
02:40throw more compute power at it seems like the brute force method. And it's not very computer science.
02:46We like to think. I'm an ex-practitioner. We like to think rather than just, you know, build. So I'm
02:52thinking that breakthroughs in the design of these things are going to help out and may make a lot of
02:59the demand that we assume is there for hardware, not irrelevant, but not infinite.
03:05Well, to that point, we've seen what China has been forced to do because it can't just throw ever
03:11more compute at the situation. It doesn't have access in terms of supply chain. What about Micron's
03:17own ability to the global opportunity here? And they're exposed to China in some ways.
03:23Sure. I mean, it is a concern because we don't really want to allow them to sell the really good
03:29stuff into China because of military issues and, well, just, you know, leadership. But
03:34that we're a capitalist society. So some of that kind of has to get pushed aside. And we let,
03:43you know, companies work throughout the world. That is a concern. But right now, for the foreseeable
03:51future, let's say 18 months to 36 months, there doesn't look to be another breakthrough other than
03:56scaling. And it looks like, all things being equal, that scaling is the path forward, especially for
04:04large language models. So unless information comes to light tonight, it looks like the companies that
04:13are offering products to these big data centers are going to continue to have their products very much
04:19in demand. What's interesting is we're trying to understand how much NVIDIA is still in demand at the
04:24same time as an Amazon builds out its Tranium, as the same time as TPUs come to the fore at Alphabet,
04:29and indeed Broadcoms used for ASICs. What do you make of where, if they can all win?
04:35Well, I think they can all win because the training, training, test, and validation are parts of this
04:42process. And they can better use different hardware setups. So I think that there's a place for
04:51everyone. What you're trying to figure out is how much of a place for each one of these items.
04:57I think competition always makes it better for the end consumer. And I'm not talking about the
05:03people using LLMs. I'm talking about the people building LLMs now. So I am very heartened to see
05:09that the leadership of NVIDIA isn't overshadowing everybody else with an idea. I think it's also kind
05:16of interesting, just from a performance standpoint, that companies like Google and Amazon have been
05:22able to quickly come out with something that's very usable. And we all had the assumption maybe
05:3018 months ago that that was, you know, the horse was out of the barn and over the hill and, you know,
05:36so far away. NVIDIA was the clear winner. So actually, I love it as a capitalist.
05:41I love it as a capitalist.
05:51NVIDIA
05:53JONI
05:53JONI
05:57JONI
05:58JONI
06:00JONI
06:01JONI
06:06JONI
06:09JONI
Comments