00:00You kind of nailed what's been a very recent change in psychology of the market. We've gone
00:04from a place in the middle of the week where Micron is absolutely surging and the market is
00:10cheering a situation where memory chip supply is so tight and the memory names have pricing power
00:16to now looking at it from completely the other perspective, from the seller of memory to now
00:21the buyer. And that Apple case study, you know, the way that Bloomberg put it is Apple raising
00:26prices because one single component cost is so high is an extreme measure. And so now
00:32extrapolating out and looking at what happened in Asia trading overnight, a little bit the trading
00:36of this morning's market. The concern is the next step is we come full circle and demand for consumer
00:43electronics is dampened because of high prices. And by extension, memory ironically sees demand
00:51dampen a bit. So it's a really sort of circular debate the market's having with itself here.
00:57Well, you know, when it comes to the memory chip shortage, is this something that is acute at the
01:03moment or is this going to be the new normal, especially with all the demand for chips from
01:08everywhere essentially? It's incredibly severe. So like in very simple terms, all these data centers
01:14that are getting built specifically with NVIDIA GPUs require lots of what's called high bandwidth
01:21memory. High bandwidth memory is just DRAM. Everyday DRAM goes into your own consumer electronics
01:26layered upon one on top of the other. For every NVIDIA GPU you have, there are eight corresponding
01:32high bandwidth memory chips, so lots of layers of DRAM. And because A, Micron and other memory names
01:38have pricing power, and B, that those building the data centers have lots of money, that's where the
01:43supply is being prioritized. The net result is that the consumer electronics industry, not just Apple,
01:49Xbox raised prices on its consoles too. They have no choice but to pay that higher pricing in order
01:55to secure supply. It is a very severe situation. And, you know, the weirdness was that in the moment,
02:03again, there was euphoria in the markets Wednesday into Thursday. Good for you, Micron. Stock surges.
02:09It's a really enviable position to be in. And what Micron said is that this situation is going to last
02:14for at least 18 months before it improves. So it's a lot for the market to decide, like, what is
02:20good
02:20news and what is bad news here? Ed, when we had those higher egg prices coming out of the pandemic,
02:25my genius solution was to just make more chickens. And that basically is what happened and egg prices
02:31came down. That logic, I don't think it's going to apply to memory chips, will it?
02:36So historically, before this whole AI thing, memory was highly cyclical, highly commoditized. It was an
02:44industry that went through boom and bust. When consumers got disinterested in PCs and smartphones,
02:49the memory market crashed, both in terms of pricing. And so that is an industry that's so sensitive to
02:56getting supply right. The story under this administration in particular, but for a longer
03:01period of time has been, let's get these companies to build more capacity. The problem is, what if they
03:06build all these memory chip fabs and then demand evaporates and they're left with very large facilities
03:11that aren't doing anything? Well, that would be a really less than prudent financial move. That is
03:17still there. The difference is, is that the data center demand seems to be here forever. It has staying
03:24power. It's not as cyclical as the consumer electronics industry has been. It's really hard
03:29one for them to get right. All right. And let's switch gears and talk about this OpenAI IPO, which
03:34may not be happening now until 2027. What can you tell us about that? I'm just looking at my bio
03:39to
03:40see if we've published yet. We haven't. What I can say is that it's consistent with what OpenAI said on
03:48June 8th, when it disclosed its confidential filing, that they would stay private for quite a long time
03:53because they're things that they want to do that are more achievable as a private company than a public
03:58company. There is a lot going on in capital markets, as you know. We had the biggest IPO of all
04:02time in
04:03SpaceX. There is a lot of activity in the bond market, not just SpaceX, tapping the bond market, NVIDIA
04:10as well being the most recent. And there is a finite pool of capital. It's consistent that OpenAI would take
04:17its time. But I think what I can say safely on air is like, I think OpenAI is also pretty
04:21calm about
04:22the idea that Anthropic goes public before them. You know, if you put it in the context of a race
04:27between two AI companies.
Comments