Latest / In the Loop AI / Memory Could Be the Last Great Investment of this AI Cycle

Transcript

Memory Could Be the Last Great Investment of this AI Cycle · In the Loop AI

0:00
Ejaaz: If you invested $25,000 in SanDisk last year, you'd have a million dollars today.
0:05
Ejaaz: Memory stocks have absolutely skyrocketed, and the reason behind that is simple.
0:09
Ejaaz: It's AI. Memory has become the core component behind GPUs, CPUs.
0:13
Ejaaz: It actually makes up 50% of the materials cost to actually build AI models today.
0:18
Ejaaz: And just six weeks ago, a new memory ETF launched, giving you exposure to the
0:22
Ejaaz: top three memory manufacturers, SK Hynix, Samsung, and Micron.
0:26
Ejaaz: And just under a month from launching, it is up over 100%. It took in over $6
0:33
Ejaaz: billion in assets in the first five weeks of it operating.
0:37
Ejaaz: In fact, Goldman Sachs referred to it as the fastest growing ETF that they've
0:41
Ejaaz: seen with $1.1 billion worth of inflows in a single day. So this begs the question,
0:46
Ejaaz: Is memory stocks or AI memory specifically a good investment to make right now
0:51
Ejaaz: or has the bubble grown too much?
0:53
Ejaaz: If we look historically at memory, it's gone under very volatile boom and bust cycles.
0:58
Ejaaz: So in this episode, we're going to explore whether this makes sense at all.
1:01
Ejaaz: And I think I would personally argue that I think this time is different.
1:05
Ejaaz: Memory has undergone a significant shift of demand where AI is just consuming all available supply.
1:11
Josh: So maybe just take a step back before we get into why. We could talk about what
1:14
Josh: memory is as it is referred to around AI, which is basically high-speed chips
1:19
Josh: that sit physically next to the GPU.
1:21
Josh: And they hold the model weights and the data that the chips need in order to
1:24
Josh: access during the inference and training run.
1:27
Josh: So this matters because these large models are bottlenecked by not really how
1:30
Josh: fast the GPU can do the math, but how fast it can pull data from that memory.
1:34
Josh: So the speed and capacity of that memory directly determines how big and how
1:39
Josh: fast a model can run. So if you want to publish AI tokens, if you want to generate
1:43
Josh: tokens, memory is the single critical kingpin.
1:46
Josh: And this is different than anything we've ever seen before, because for the
1:49
Josh: last 40 years of time, memory was the pretty much textbook example of a horrible business.
1:54
Josh: There are three suppliers that make these ships. There's Samsung,
1:57
Josh: SK Hynix, Micron, and they make identical products, which are essentially a commodity.
2:01
Josh: And it turned into this thing called a pig cycle, which is basically high prices.
2:05
Josh: And then everyone builds the same fabs. And then there's oversupply,
2:08
Josh: which results in price collapse. and this cycle has kind of run itself every
2:11
Josh: couple of years since the 1980s.
2:13
Josh: But Ijaz, like you said, this time is different. There are these two key traits
2:17
Josh: that are converging that make it different this time.
2:20
Ejaaz: So the first one is...
2:22
Ejaaz: The demand sync that comes from AI for memory has vastly outpaced any other
2:27
Ejaaz: demand sync that memory has had before.
2:30
Ejaaz: So if you look at pre-AI memory stock charts and post-AI, it looks like completely
2:35
Ejaaz: like a penny stock. It's just gone parabolic, right?
2:38
Ejaaz: And so the question everyone's asking is, is this real? Well,
2:41
Ejaaz: the answer is, it's true.
2:42
Ejaaz: If you look at NVIDIA specifically, when the AI cycle really kicked off,
2:47
Ejaaz: their GPUs were in incredible amounts of demand.
2:50
Ejaaz: Now, back then, the models used a significant chunk of memory,
2:54
Ejaaz: typically around 30% to 40% of the bill of materials cost for GPUs back then
2:58
Ejaaz: for NVIDIA's early models went to memory, basically.
3:02
Ejaaz: Now, if you fast forward to today, it's 50%. And if you look at every future
3:06
Ejaaz: iteration, including Rubin, Rubin Ultra, and Feynman, that is going to be launched later in 2028.
3:12
Ejaaz: Doesn't just require the same amount of memory, it requires 2,
3:15
Ejaaz: 3, and 4x the amount of memory for every single iteration.
3:19
Ejaaz: So the point being is, there is an insatiable amount of demand for memory,
3:24
Ejaaz: and there are different types of memory.
3:25
Ejaaz: In fact, AI has created a completely new novel type of memory structure known
3:30
Ejaaz: as HBM, high bandwidth memory,
3:32
Ejaaz: which functions above the status of local genuine DRAM, which is like your standard
3:37
Ejaaz: DRAM, which a lot of other consumer devices use, just for AI specifically.
3:43
Ejaaz: So the point being is there are very limited amounts of memory available and
3:48
Ejaaz: AI is eating up pretty much the majority of it right now.
3:52
Josh: Yeah, so we have the network effects that are occurring just from AI demand,
3:56
Josh: but the high bandwidth memory is a key structural difference that I think a
3:59
Josh: lot of people aren't really aware of.
4:00
Josh: If you imagine a single traditional stick of RAM, something that you might see
4:04
Josh: plugged into a computer, you could think of it, that's like a one-story warehouse.
4:07
Josh: What these AI chips require right next to the GPU is this thing called high
4:11
Josh: bandwidth memory and that is a warehouse but stacked vertically
4:15
Josh: mostly right now it's about 8 to 12 stacks high soon there's
4:18
Josh: going to be 16 stacks high and that requires a lot of
4:21
Josh: very precise materials a lot of engineering a lot of
4:24
Josh: resources that make these chips far more complex but also
4:27
Josh: constrain the supply far more i mean each gpu or each
4:30
Josh: stick of memory is now approximately 8 to 12 sticks
4:33
Josh: so you have this convergence of a few things one is just the
4:36
Josh: need for them in order to generate tokens but specifically the
4:39
Josh: need for that high bandwidth memory that makes a huge difference so
4:42
Josh: now not only do we have the supply constraints coming in terms of the actual
4:46
Josh: structural advantages that um hbm has but you also only have three people that
4:51
Josh: actually produce this there are three manufacturers and the constraint is so
4:55
Josh: tight that they have actually some of them stopped making commercially available
4:58
Josh: ram because they're so focused on building
5:01
Josh: purely memory for AI.
5:03
Ejaaz: Yeah. And there's also a recent shift that has happened within AI in general
5:08
Ejaaz: where previously it was very focused on AI training. And so it was very GPU heavy.
5:14
Ejaaz: That's why NVIDIA is now like the most valuable company in the world.
5:17
Ejaaz: Now that recently shifted towards inference. Inference is like where you prompt
5:22
Ejaaz: or call on a model and you get an answer back.
5:25
Ejaaz: Instead of humans doing the majority of the inference, we're finding that AI
5:28
Ejaaz: models themselves or AI agents specifically are spending a lot of time thinking.
5:34
Ejaaz: Talking amongst each other and coming up with a better output and a better answer.
5:38
Ejaaz: That's what a lot of cutting edge models and AI products today actually do behind the scenes.
5:43
Ejaaz: That requires a lot of inference. Now, if you want to do all that amount of
5:47
Ejaaz: inference, you need not just a little bit more memory, you need a load more memory.
5:52
Ejaaz: So the argument that's being made right now is in a future where the entire
5:56
Ejaaz: economy is facilitated or propped up by AI agents, you're going to need like
6:01
Ejaaz: 10 to 50x more memory than we have available today.
6:05
Ejaaz: And Ben from Stratechery did an amazing breakdown and I kind of like summarized
6:10
Ejaaz: it here where basically the value of the opportunity is like 10 to 50x from
6:13
Ejaaz: now and we don't currently have enough memory to do that.
6:15
Ejaaz: And it immediately reminded me of another company, which was NVIDIA,
6:19
Ejaaz: back when they were building out GPUs for gaming processing,
6:23
Ejaaz: graphics processing specifically, And then they stumbled across this AI thing
6:27
Ejaaz: where they were like, huh, it also needs parallel computing.
6:30
Ejaaz: And now, like, the demand sync was crazy from AI. The same thing is happening
6:33
Ejaaz: with memory, and these memory manufacturers are going to benefit the most.
6:37
Josh: So we're going to get into what companies are best positioned for memory.
6:41
Josh: But prior to that, it's probably makes sense to explain why this time is different,
6:45
Josh: why the clock is ticking, where there's a perhaps a limited opportunity before
6:49
Josh: the market starts to realize this structural change.
6:53
Josh: Now, we mentioned the problems that are occurring in terms of just supply side
6:56
Josh: due to the invention of HBM and just the requirement of that in every single GPU.
7:00
Josh: But there's also a software side demand vector as well, which is the rise of agents.
7:04
Josh: And we've mentioned this in a previous episode, but I think it's important to
7:07
Josh: double down on, is the idea that you're not just talking to a singular chatbot anymore.
7:11
Josh: And when you send a query to an LLM, like ChatGPT or like Claude,
7:15
Josh: you create this thing called the KV cache.
7:17
Josh: And that uses a lot of memory. It stores a lot of the context inside of this narrow window.
7:21
Josh: Now, what happens with agents, which is the new paradigm, which is what a lot
7:25
Josh: of these companies are shifting to, and a lot of people who are doing productive
7:27
Josh: work are actually using, those agents, each one consumes their own KB cache,
7:32
Josh: which means it needs, if you have 20 agents, you need 20 instances of memory.
7:36
Josh: And there is no sign of that trend slowing down where there's more instances
7:40
Josh: or more demand for agents, which means more context that needs to be remembered,
7:44
Josh: which means significantly more of that HBM, which is a ready supply constraint.
7:49
Josh: In fact, if you want to get some memory now, lead times are extended currently
7:53
Josh: to the end of 2027 because there's just this inability to scale production.
7:58
Josh: And I think when we answer the question of why this time is different,
8:01
Josh: there are these unique structural changes, both on the hardware demand side,
8:05
Josh: but also on the software supply side that both point to
8:09
Josh: A pretty far out window in which you should start to worry about demand because
8:13
Josh: there is just a very clear trend in one direction and a very clear under-indexing
8:18
Josh: of the ability to supply all of the memory required to fulfill that trend.
8:21
Josh: So I think that's something worth noting is that there are demands for this
8:25
Josh: coming from all these different areas.
8:27
Ejaaz: There's a physical constraint to building memory.
8:30
Ejaaz: The top three providers cannot physically build out more capacity.
8:36
Ejaaz: In fact, I was reading a story earlier this week where apparently
8:39
Ejaaz: a bunch of different AI customers have been offering Micron and the likes of
8:45
Ejaaz: Samsung and SK Hynix to help them build out or purchase equipment in exchange
8:50
Ejaaz: for guaranteed capacity or supply in 2027 and 2028, right? You're not exaggerating that.
8:57
Ejaaz: Currently, they're booked up until the end of 2027.
9:01
Ejaaz: And so there's a physical constraint there. Typically in a bubble,
9:05
Ejaaz: whether it's a memory bubble or a tech bubble in general,
9:07
Ejaaz: you have some version of these companies faking demand, and they're levering
9:13
Ejaaz: up to create a supply which ends up becoming an oversupply.
9:17
Ejaaz: So there's too much of the product out there, and so prices kind of like crash
9:20
Ejaaz: demand kind of like sinks. That isn't the case here.
9:24
Ejaaz: If you look at every quarterly earnings report from each of these three manufacturers,
9:28
Ejaaz: who are all public, by the way,
9:30
Ejaaz: you can see that these back orders, that these bookings are not only legitimate,
9:35
Ejaaz: but they're already paid for
9:36
Ejaaz: in advance, which typically doesn't happen with any of these contracts.
9:39
Ejaaz: So you have legitimate cash on the books that you can then use as a cash projection
9:44
Ejaaz: or a forward projection for a lot of these companies. In fact,
9:47
Ejaaz: if you look at Micron, their forward earnings right now is under 10, which.
9:52
Ejaaz: Isn't definable within a bubble where if you look in like past historical cycles,
9:57
Ejaaz: that typically hits around 30 to between 30 to 50 affords earnings ratio.
10:01
Ejaaz: So the point is, it's legitimate as far as we can say.
10:05
Ejaaz: Now, skeptics are saying, maybe it's the earnings that is the bubble.
10:10
Ejaaz: So the earnings that they're reporting isn't actually real, but that's harder
10:14
Ejaaz: to pick apart than what we're seeing right now.
10:17
Josh: Yeah, I was actually going to mention, I think that's such an important point. the idea
10:20
Josh: that although these companies are trading so much higher sandisk is
10:22
Josh: up 1500 percent the forward pe ratio which
10:25
Josh: is the price to earnings basically what premium it trades for on a
10:28
Josh: relative basis versus what it's worth today is only 9 to 12x and it's lower
10:33
Josh: for samsung at 7x and it's even lower for sk hynix at 5 to 6x so they're not
10:38
Josh: really trading at a premium relative to what they're guiding for their estimates
10:42
Josh: and revenue to be if you take a company like tesla i think that's currently trading at 40 to 50x.
10:47
Josh: And companies like Microsoft or Amazon are like 15 to 20x.
10:51
Josh: So these memory stocks on a relative basis are actually valued pretty in line
10:55
Josh: with what you'd expect for a industry that's doing very well.
10:58
Josh: And that's noteworthy. So let's get into the actual stocks, the companies that
11:02
Josh: people can participate in, or at least look at if they're interested in participating
11:06
Josh: in this. There are three large players.
11:08
Josh: There is SK Hynix, their Samsung, and their SanDisk.
11:11
Josh: I'm seeing another ticker here, which perhaps you could walk us through,
11:15
Josh: which is MU up 864%. Why is that up so much?
11:18
Ejaaz: Yeah, I mean, it's Micron. Micron is up over 8x over the last year.
11:23
Ejaaz: The primary reason for that is the other two companies, the biggest memory manufacturers,
11:28
Ejaaz: which is SK Hynix and Samsung electronics are based in Korea.
11:32
Ejaaz: So if you're in the West, or if you're in America specifically,
11:35
Ejaaz: and you want to get exposure to these stocks, it's not as easy as going on to
11:39
Ejaaz: your Chase stock account and purchasing these, you can't exactly do that.
11:42
Ejaaz: There's a few more hoops to jump through, and most people aren't willing to do that.
11:46
Ejaaz: So another way of getting exposure to this thing is to buy the main American memory manufacturer.
11:52
Ejaaz: And there is one clear leader, which is Micron. So their stock has absolutely set.
11:56
Ejaaz: And to be fair, in Micron's defense, they produce a really good product and
12:00
Ejaaz: they are one of the advanced suppliers.
12:02
Ejaaz: But if you want to talk about who the king is, it is SK Hynix.
12:06
Ejaaz: They have been the sole and main provider for memory, for NVIDIA and all their future GPUs.
12:12
Ejaaz: In fact, in the same way that NVIDIA has booked out the majority of capacity
12:16
Ejaaz: for TSMC to build their GPUs, they have done the same with SK Hynix.
12:20
Ejaaz: So the reason why those people that I mentioned earlier are trying to like pay
12:23
Ejaaz: off SK Hynix for guaranteed supply in the future is because NVIDIA owns the
12:27
Ejaaz: monopoly on all capacity for SK Hynix going forward.
12:30
Ejaaz: So SK Hynix is like the strongest. Samsung Electronics is next.
12:33
Ejaaz: Samsung Electronics does a ton of other stuff, which is why it's not a pure
12:36
Ejaaz: concentrated bet on memory.
12:38
Ejaaz: And then you have Micron, which is like the American bet. Now,
12:41
Ejaaz: what I want to speak about is that ETF I mentioned earlier, Josh,
12:45
Ejaaz: that launched six weeks ago.
12:47
Ejaaz: That is like the fastest-growing ETF, I think is like $5 billion worth of inflows
12:52
Ejaaz: over the last five weeks or $6 billion.
12:54
Ejaaz: Um, it's this thing called DRAM. It's a fantastic ticker and it's a memory ETF.
13:00
Ejaaz: And the reason why it's so unique is two reasons.
13:03
Ejaaz: One, it's accessible in the West, but two, it's a basket exposure to those two
13:08
Ejaaz: Korean companies that I mentioned earlier on. It owns SK Hynix shares.
13:13
Ejaaz: It owns Samsung Electronics shares, but it also owns Micron.
13:18
Ejaaz: In fact, all three of these companies, I think make up around 70% of the entire
13:22
Ejaaz: basket. So it's a very concentrated play on memory, AI memory specifically.
13:26
Ejaaz: And then it also owns smaller stakes in the likes of SanDisk and Westgate,
13:30
Ejaaz: which is a different type of memory that is still equally important for the AI trade.
13:35
Josh: We have a great visual on this i think this is probably what most people will
13:38
Josh: be interested in participating in just showing the holdings within this i mean
13:42
Josh: this is the access that you want if you want access to the korean companies
13:46
Josh: it has 24 sk hynix 25 micron technology 5 in wd 6
13:52
Josh: seagate samsung is 20 so this is probably the most
13:55
Josh: balanced basket of exposure this is personally what i'm most
13:58
Josh: interested in if you don't want to think about it too much you just set it
14:01
Josh: and forget it uh ejash you mentioned this to me a little while
14:04
Josh: ago and since then it has gone absolutely nuclear this seems
14:06
Josh: to be the singular place in which people are converging on
14:09
Josh: if they don't want to do all the hard work of picking winners and losers
14:12
Josh: in fact this has recently gotten enough
14:15
Josh: attention to where they have leveled it up even once
14:18
Josh: more so now there is a ticker that goes by the
14:21
Josh: name of just ram instead of dram which is the etf that
14:24
Josh: covers basically the entire memory market and this is a 2x
14:28
Josh: leverage long memory etf that is going
14:31
Josh: to be launching very soon so if you really want to go risk on
14:33
Josh: if you really believe in the memory trade over a long period of time there are
14:37
Josh: tools available to be able to enable you to gamble on this to the full extent
14:42
Josh: i'll probably just stick with the regular dram etf but for those ambitious this
14:46
Josh: is probably the best way to get access to those individuals particularly if
14:49
Josh: you listen in the united states like most of us do and you want access to those Korean powerhouses,
14:54
Josh: this is a pretty good way of doing it.
14:55
Ejaaz: Now, I'm sure what a lot of listeners are thinking as you're hearing what we're
15:01
Ejaaz: saying is, okay, well, all these stocks are up
15:04
Ejaaz: 5 to 8x, we've completely missed the trade.
15:08
Ejaaz: Why on earth would we potentially want to invest in this? Is this a bubble? Is this going to pop?
15:13
Ejaaz: And there is feasibility to argue that that might be the case.
15:17
Ejaaz: This time isn't actually different.
15:19
Ejaaz: In fact, if we look at the historical context of these memory suppliers,
15:24
Ejaaz: on the left here in 2000, there were 11 main manufacturers of memory.
15:30
Ejaaz: And it's important to understand that memory has gone through boom and bust
15:33
Ejaaz: cycles for the better part of the last two to three decades.
15:37
Ejaaz: And with every boom and bust cycle, you end up with fewer players,
15:41
Ejaaz: which brings us over to the right over here, which this is funny,
15:44
Ejaaz: this was in 2013, Josh. So these three have kind of stuck it out.
15:48
Ejaaz: You have three companies which basically dominate memory manufacturing in its entirety.
15:53
Ejaaz: So the question now is, is this gonna happen in the same way?
15:58
Ejaaz: Now, there's two sides of this argument.
16:00
Ejaaz: The side that argues against this thesis is,
16:04
Ejaaz: Um, AI has created a demand sync for the reasons that we mentioned earlier,
16:09
Ejaaz: Genetic AI, the fact that you need it for all successive GPUs in larger quantities
16:12
Ejaaz: than we've seen in any other technology shift. And that is the case.
16:16
Ejaaz: Um, AI is pretty pervasive across every single industry.
16:20
Ejaaz: So if you assume that every single industry is going to have some form of AI
16:23
Ejaaz: model that changes the way that the industry works or AI agents that can automate, say,
16:28
Ejaaz: I don't know, 10 to 40% of working roles there, you can then extrapolate to
16:33
Ejaaz: think, well, okay, that's going to need a bunch of memory, a bunch more GPUs,
16:36
Ejaaz: a bunch more CPUs, a bunch more agents to be able to facilitate that.
16:40
Ejaaz: So you can then kind of mark up the amount of memory that is required.
16:43
Ejaaz: And that's why these stocks have probably a longer way to go.
16:46
Ejaaz: Now, the skeptic would argue, okay, that's the case for now.
16:50
Ejaaz: But what if we launch an AI model or a GPU, which requires much less memory?
16:54
Ejaaz: Won't that mean that memory oversupply will happen? What happens when Micron's
16:58
Ejaaz: fab gets launched and we have a heck ton more memory?
17:02
Ejaaz: Will that decrease prices in demand? You could argue yes and no.
17:05
Ejaaz: Jevon's paradox, people will just maybe want more of the things.
17:08
Ejaaz: So you may need even more amounts of memory. But those are both sides of the opposing argument.
17:13
Josh: Okay, so I guess I have something that I'd want to challenge you on in kind
17:17
Josh: of supporting why this time is different, if you think this time is truly different.
17:20
Josh: Because when we look at the historical patterns of what these memory cycles
17:23
Josh: are like, they generally last about 18 months and then experience a pullback
17:28
Josh: of anywhere from like 40 to 75%. This happened in 1995, 99, 2017, 2023.
17:34
Josh: And now this current cycle started actually in the fourth quarter of 2024.
17:37
Josh: So we're about month 18 now, which is where things should be either taking a
17:43
Josh: turn for the worse or changing for the better.
17:45
Josh: So if you're thinking about how you allocate personally and how you believe
17:49
Josh: this to go, this is a pretty pivotal time. If you're looking at it as if there's
17:54
Josh: like resistance on the chart, we're at that resistance right now. We're in month 18.
17:57
Josh: Does it seem like it's more probable that this actually is different or is this
18:01
Josh: the natural cycle? Because I mean, all the signs point to it being different this time.
18:05
Josh: But you mentioned, I mean, the ETF had doubled in five weeks.
18:08
Josh: So a lot of that like easy beta that people had is gone.
18:11
Josh: Is there still opportunity for the upside? And what timescale are you thinking about?
18:14
Ejaaz: If you ask me personally, I do think that there is a long term, a lot more upside.
18:21
Ejaaz: But it wouldn't surprise me if we see a short term retrieval from stock price
18:26
Ejaaz: charts. We have had, I think, six weeks of constant uptick in this DRAM ETF
18:32
Ejaaz: and in just memory stocks in general.
18:35
Ejaaz: And that has been on the basis of these quarterly earnings and a bunch of other
18:38
Ejaaz: partnership announcements.
18:40
Ejaaz: That can't keep happening every single week. At some point, we're going to run
18:44
Ejaaz: out of news to kind of push prices up further.
18:46
Ejaaz: And I think we're going to see a recession. I think people are kind of overthinking it.
18:50
Ejaaz: Now, the argument against that would be, well, forward earnings ratios are still
18:53
Ejaaz: quite low. And some people might see that, but I just don't think we can see
18:57
Ejaaz: a chart go up exponentially every single week. I think we're going to have some back pull.
19:01
Ejaaz: The thing that gives me optimism is those backlog of orders that I mentioned
19:07
Ejaaz: earlier that aren't just signed, sealed, but before they even delivered.
19:12
Ejaaz: People have paid for it up front. There was a recent deal that was struck between,
19:17
Ejaaz: I believe it was Google and Micron for their TPUs, and they paid 40% up front.
19:24
Ejaaz: So the point is, these companies really want memory, and the companies that
19:30
Ejaaz: really want memory are not unknown companies. They're the mag-7 who have the
19:36
Ejaaz: cash flow and free cash flow to pay for it and back it up.
19:39
Ejaaz: So if you have some of the smartest, most intelligent companies,
19:42
Ejaaz: most valuable companies in the world, putting up money and big risks like that,
19:46
Ejaaz: you could argue that this time is different and more legitimate.
19:49
Ejaaz: Now, of course, there is the non-zero chance that all of this blows up,
19:52
Ejaaz: that the demand is actually fake, that Anthropic and OpenAI on their demand
19:56
Ejaaz: side actually end up inflating their numbers too much.
19:58
Ejaaz: And that could lead to a waterfall cascade of all these things crashing.
20:01
Ejaaz: But I don't see the structural context for that happening right now.
20:05
Josh: 12 months from now, higher or lower?
20:07
Ejaaz: Higher.
20:07
Josh: I think higher too. That feels like the vibe.
20:10
Josh: There is like very clear trajectory and sure, there will be a bit of volatility
20:15
Josh: along the way, but higher feels like an optimistic version of the future that
20:19
Josh: we have enough information to go on that I could feel comfortable about betting on.
20:24
Josh: So I think that's probably most of the case for memory.
20:28
Josh: That's why it's important what unique market structures are now available that
20:32
Josh: weren't in the past version of these 18-month cycles and where the opportunity
20:36
Josh: lies if you are interested in participating.
20:39
Josh: I guess the ask for you, the person who's listening to this, is...
20:43
Josh: What do you think on that 12-month timescale? Or what is the best way for you to get exposure?
20:47
Josh: There's this chart that really devastates me personally, EGIS,
20:50
Josh: as someone who loves to create my own custom PCs and build hardware.
20:55
Josh: The cost, yeah, if you scroll down all the way to the bottom even,
20:58
Josh: the cost of these memories or memory systems for consumer products is through the roof.
21:05
Josh: And that's been a little disappointing because if you want to just go buy a
21:08
Josh: PC, if you want to get a new Xbox, even the new Steam machines,
21:11
Josh: they've all been delayed or the prices have been increased.
21:13
Josh: In fact, if you want to go buy a PlayStation 5, it costs now today than it did,
21:18
Josh: what was it, five years ago when it first came out because of these supply constraints.
21:22
Josh: So we're starting to see it permeate out into the general broad consumer market.
21:27
Josh: And that feels a little discouraging. I wonder how high that can go.
21:30
Josh: I wonder how much tolerance the consumer market has before it starts to bend.
21:33
Josh: So that's something I'm going to be looking at also, is following these prices
21:36
Josh: as it relates to just general consumer hardware.
21:39
Josh: Is the iPhone going to suffer are those prices going to go up the average smartphone
21:42
Josh: costs things of this nature are also on the watch list to see if they could
21:45
Josh: if they could bear it and if they can then we're probably good if not it's something
21:49
Josh: to just take some note of personally
21:51
Ejaaz: Memory is going to be one of the biggest bets in the same way that gpus were
21:55
Ejaaz: and it saw nvidia you know breach of above five trillion dollar market cap i
22:00
Ejaaz: think the same thing is going to happen in um memory this time and we have three concentrated players.
22:06
Ejaaz: One in America, Micron, that is going to do the same thing for memory.
22:10
Ejaaz: So I'm excited about it. I am slightly nervous about the boom and bust of these
22:15
Ejaaz: different cycles, but I guess we'll see if this time is different.
22:18
Ejaaz: The fundamentals. It's getting high, man. Things are going up.
22:21
Ejaaz: Well, at some point, it's going to break, right?
22:24
Ejaaz: I'm looking at this consumer PC memory chart, and I'm like, at some point,
22:28
Ejaaz: people are just going to throw their hands up in the air.
22:30
Ejaaz: It's going to be too expensive and something will need to shift,
22:34
Ejaaz: whether it is these consumer device makers like Apple.
22:38
Ejaaz: Pushing back and saying, listen, we're just not going to do this,
22:41
Ejaaz: or we need to find an alternative.
22:43
Ejaaz: I don't know what that is. But one thing is for sure is the demand for AI models,
22:48
Ejaaz: the demand for AI agents is just going up and to the right.
22:52
Ejaaz: If it's not coming from consumer, it's coming from enterprise.
22:54
Ejaaz: So I just don't know how this conundrum is going to be fixed.
22:57
Ejaaz: Maybe we create an alternative type of model. But either way, we will be tracking it.
23:01
Ejaaz: And I'm bullish memory at least for the next six to 12 months.
23:05
Ejaaz: But that is it. I'm curious whether the listeners of the show have any other takes.
23:10
Ejaaz: We've released, I think, two other investment-themed episodes over the last
23:14
Ejaaz: week and a half, and the feedback has been tremendous.
23:17
Ejaaz: Both comments telling us why we're right and why we are explicitly wrong.
23:22
Ejaaz: I want to hear from both sides because it helps us kind of like figure out where
23:26
Ejaaz: we're going next and what to cover next.
23:28
Ejaaz: If there's anything that we missed on the memory set of things, also let us know.
23:32
Ejaaz: And finally, we heard you on wanting to talk about or unpack the layers of substrates
23:39
Ejaaz: and infrastructure that we covered on our previous episode.
23:42
Ejaaz: We are going to plan and deliver that episode fairly soon. But this week,
23:46
Ejaaz: later on, we're getting Leopold's firings.
23:48
Josh: Happy Leopold week. This is huge.
23:50
Ejaaz: This is huge.
23:52
Josh: Our last two episodes, those were among the biggest ever because it was covering Leopold.
23:56
Josh: I don't want to say, i don't want to take any credit but if you search up leopold's
23:59
Josh: portfolio or if you go on x we're very much the beginning of
24:02
Josh: the leopold narrative and now it is the next chapter in
24:05
Josh: that so leopold's uh fun situational awareness has
24:08
Josh: gone from 1.5 billion to like 6 billion in the course of
24:11
Josh: two years and they publish these things called 13f filings every
24:14
Josh: quarter that reveals the new set of positions that 13f filing is going live
24:18
Josh: sometime by the end of this week it's due by friday so we will be very ambitiously
24:23
Josh: monitoring this and as soon as it releases we'll have an episode ready to go
24:26
Josh: to talk about all the new investment thesis around situational awareness for the next quarter.
24:31
Josh: So stay tuned for that. We are locked in and ready to go as soon as that one drops.
24:35
Ejaaz: Exactly. But aside from that, I think that is it. And we will leave you guys
24:40
Ejaaz: until the next episode. But see you then.