记忆可能是本轮人工智能周期最后的伟大投资

如果你去年投资了两万五千美元买闪迪的股票，现在早就值一百万美元了。

If you invested $25,000 in SanDisk last year, you'd have $1,000,000 today.

存储类股票简直涨疯了，原因很简单：全是人工智能带火的。

Memory stocks have absolutely skyrocketed, the reason behind that is simple: it's AI.

现在内存已经成为GPU和CPU背后的核心部件。

Memory has become the core component behind GPUs, CPUs.

如今，光是构建AI模型，内存就占了材料成本的五成。

It actually makes up 50 of the materials cost to actually build AI models today.

就在六周前，一款新的存储行业ETF上线了，让你能直接投资到三大内存巨头：SK海力士、三星和美光。

And just six weeks ago, a new memory ETF launched, giving you exposure to the top three memory manufacturers SK Hynix, Samsung, Micron.

上市还不到一个月，它的涨幅已经超过了一倍。

And just under a month from launching, it is up over 100%.

在头五个星期里，它就吸金超过六百亿美元。

It took in over $6,000,000,000 in assets in the first five weeks of it operating.

连高盛都称，这是他们见过的增长最快的ETF，单日资金流入就高达十一亿美元。

In fact, Goldman Sachs referred to it as the fastest growing ETF that they've seen with $1,100,000,000 worth of inflows in a single day.

所以这就引出一个问题：现在投资内存股，尤其是AI相关的内存，是个好主意吗？还是说泡沫已经吹得太大了？

So this begs the question, is memory stocks or AI memory specifically a good investment to make right now, or has the bubble grown too much?

从历史来看，内存行业一直经历着剧烈的涨跌周期。

If we look historically at memory, it's gone under very volatile boom and bust cycle.

所以在这期节目中，我们来好好聊聊，这种投资到底有没有道理。

So in this episode, we're gonna explore whether this makes sense at all.

我个人觉得，这一次情况真的不一样了。

And I think I would personally argue that I think this time is different.

现在内存的需求发生了根本性变化，AI几乎把所有产能都吞掉了。

Memory has undergone a significant shift of demand where AI is just consuming all available supply.

在深入分析原因之前，我们先澄清一下，大家常说的AI内存到底是什么——其实就是紧挨着GPU的高速芯片，用来存储模型权重和推理、训练过程中芯片需要快速调用的数据。

So maybe to take a step back before we get into why, we could talk about what memory is as it is referred to around AI, which is basically high speed chips that sit physically next to the GPU, and they hold the model weights and the data that the chips need in order to access during the inference and training run.

这很关键，因为大型模型的瓶颈不在于GPU的计算速度，而在于它从内存中读取数据的速度有多快。

So this matters because these large models are bottlenecked by not really how fast the GPU can do the math, but how fast it can pull data from that memory.

因此，内存的速度和容量直接决定了模型能跑得多大、多快。

So the speed and capacity of that memory directly determines how big and how fast a model can run.

所以如果你想发布AI相关的芯片，想生产这些内存颗粒，内存就是最关键的核心环节。

So if you want to publish AI tokens, if you wanna generate tokens, memory is the single critical kingpin.

这跟我们过去见过的任何情况都不同，因为在过去四十年里，内存行业几乎就是个典型的糟糕生意。

And this is different than anything we've ever seen before because for the last forty years of time, memory was the pretty much textbook example of a horrible business.

目前只有三家公司在生产这些芯片。

There are three suppliers that make these chips.

分别是三星、SK海力士和美光，它们生产的产品几乎一模一样，本质上就是大宗商品。

There's Samsung, SK Hynix, Micron, and they make identical products, which are essentially a commodity.

结果就形成了所谓的‘猪周期’——价格一高，大家都疯狂扩产，很快供过于求，价格就崩盘，这种循环从八十年代起每隔几年就重演一次。

And it turned into this thing called a pig cycle, which is basically high prices and then everyone builds the same fabs and then there's oversupply which results in price collapse, and this cycle has kind of run itself every couple of years since the nineteen eighties.

但正如伊贾兹你说的，这次真的不一样了。

But Ejaaz, like you said, this time is different.

现在有两个关键因素正在交汇，让情况跟以往截然不同。

There are these two key traits that are converging that make it different this time.

第一个因素是，AI对内存的需求增长速度，远远超过了历史上任何一次需求爆发。

So the first one is the demand sync that comes from AI for memory has vastly outpaced any other demand sync that memory has had before.

你看一下AI出现前后的内存股票走势，简直就像个仙股一样。

So if you look at pre AI memory stock charts and post AI, it looks like completely like a penny stock.

直接飙升到天上去，对吧？

It's just gone parabolic, right?

所以大家都在问，这波行情是真的吗？

And so the question everyone's asking is, is this real?

答案是，确实是真的。

Well, the answer is it's true.

拿英伟达来说，AI热潮刚起来的时候，他们的显卡需求大到爆棚。

If you look at NVIDIA specifically, when the AI cycle really kicked off, their GPUs were in incredible amounts of demand.

那时候，模型用的内存占了显卡物料成本的很大一部分，大概三到四成，都花在内存上了。

Now back then, the models use a significant chunk of memory, typically around 30 to 40% of the bill of materials cost for GPUs back then for NVIDIA's early models went to memory basically.

但到了今天，这个比例已经涨到五成了。

Now, if you fast forward to today, it's 50%.

再看未来所有新版本，包括Rubin、Rubin Ultra和Feynman——这些都计划在2028年晚些时候推出——它们需要的内存不只是跟现在一样多。

And if you look at every future iteration, including Rubin, Rubin Ultra, and Feynman, that is gonna be launched later in 2028, it doesn't just require the same amount of memory.

每一次迭代，内存需求都要翻两到四倍。

It requires two, three, and four X the amount of memory for every single iteration.

所以关键在于，内存的需求量巨大，而且内存类型也不尽相同。

So the point being is there is an insatiable amount of demand for memory and there are different types of memory.

事实上，AI催生了一种全新的内存结构，叫HBM，也就是高带宽内存，它的性能远超普通的DRAM——那种大多数消费设备用的标准内存，专为AI设计的。

In fact, AI has created a completely new novel type of memory structure known as HBM, high bandwidth memory, which functions above the status of local genuine DRAM, which is like your standard DRAM, which a lot of other consumer devices use just for AI specifically.

因此，目前可用的内存非常有限，而AI几乎吞噬了其中绝大部分。

So the point being is there are very limited amounts of memory available and AI is eating up pretty much the majority of it right now.

是的。

Yeah.

AI带来的需求正在形成网络效应，但高带宽内存才是很多人没意识到的关键结构性变化。

So we have the network effects that are occurring just from AI demand, but the high bandwidth memory is a key structural difference that I think a lot of people aren't really aware of.

如果你想象一根传统内存条，就像你平时插在电脑里的那种，可以把它看作一栋单层仓库。

If you imagine a single traditional stick of RAM, something that you might see plugged into a computer, you could think of it that's like a one story warehouse.

而AI芯片需要紧挨着GPU的，是一种叫高带宽内存的东西，它就像一栋竖着堆起来的多层仓库。

What these AI chips require right next to the GPU is this thing called high bandwidth memory, and that is a warehouse but stacked vertically.

目前主要是八到十二层堆叠。

Mostly right now, it's about eight to 12 stacks high.

很快就要达到十六层了，这需要大量精密的材料、大量的工程投入和资源，使得这些芯片变得复杂得多，但也进一步限制了供应。

Soon there's gonna be 16 stacks high, and that requires a lot of very precise materials, a lot of engineering, a lot of resources that make these chips far more complex, but also constrain the supply far more.

我的意思是，现在每个GPU或每条内存条大约相当于八到十二条传统内存条。

I mean, each GPU or each stick of memory is now approximately eight to 12 sticks.

所以，这是几个因素交汇的结果。

So you have this convergence of a few things.

一方面是为了生成 token 而需要它们，但关键在于这种高带宽内存带来的巨大差异。

One is just the need for them in order to generate tokens, but specifically the need for that high bandwidth memory that makes a huge difference.

所以现在，我们不仅面临供应方面的限制，因为 HBM 本身具有结构上的优势，而且实际上只有三家厂商在生产这种内存。

So now not only do we have the supply constraints coming in terms of the actual structural advantages that, HPM has, but you also only have three people that actually produce this.

全球只有三家制造商，而且产能限制非常紧张，以至于其中一些厂商已经停止生产面向普通市场的 RAM，因为他们把全部精力都投入到为 AI 专门制造内存上了。

There are three manufacturers, and the constraint is so tight that they have actually, some of them have stopped making commercially available RAM because they're so focused on building purely memory for AI.

确实。

Yeah.

而且最近AI领域整体上出现了一个转变，以前大家特别关注AI训练。

And there's also a recent shift that has happened within AI in general, where previously it was very focused on AI training.

所以那时候对GPU的需求特别大。

And so it was very GPU heavy.

这就是为什么英伟达现在成了全球最有价值的公司。

That's why NVIDIA is now like the most valuable company in the world.

但最近这个重心已经转向了推理阶段。

Now that recently shifted towards inference.

推理就是你给模型一个提示，然后它给你返回答案的过程。

Inference is like where you prompt or call on a model and you get an answer back.

现在不再是人类主要来做推理，而是AI模型本身，尤其是AI代理，花大量时间在思考、彼此交流，从而得出更好的结果。

Instead of humans doing the majority of the inference, we're finding that AI models themselves or AI agents specifically are spending a lot of time thinking, talking amongst each other and coming up with a better output and a better answer.

如今很多前沿的AI模型和产品，背后其实都在这么做。

That's what a lot of cutting edge models and AI products today actually do behind the scenes.

而这需要大量的推理计算。

That requires a lot of inference.

现在，如果你要完成这么多推理任务，光多一点内存可不够，得要多出好几倍的内存。

Now, if you want to do all that amount of inference, you need not just a little bit more memory, you need a load more memory.

所以现在有人提出，未来如果整个经济都靠AI代理来支撑，那我们需要的内存可能是现在的十到五十倍。

So the argument that's being made right now is in a future where the entire economy is facilitated or propped up by AI agents, you're gonna need like 10 to 50X more memory than we have available today.

Strachery的本做了一个特别精彩的分析，我在这儿简单总结一下：这个机会的价值未来可能增长十到五十倍，但我们现在的内存根本不够用。

And Ben from Strachery did an amazing breakdown, and I kind of like summarized it here where basically the value of the opportunity is like 10 to 50X from now, and we don't currently have enough memory to do that.

这让我立刻想到了另一家公司——NVIDIA，当年他们刚做游戏显卡、图形处理的时候。

And it immediately reminded me of another company, which was NVIDIA back when they were building our GPUs for gaming processing, graphics processing specifically.

后来他们偶然发现，AI也需要并行计算。

And then they stumbled across this AI thing where they were like, it also needs parallel computing.

结果AI带来的需求一下子爆炸了。

And now like the demand sync was crazy from AI.

现在同样的事情正在内存领域上演，这些内存制造商将是最大受益者。

The same thing is happening with memory and these memory manufacturers are gonna benefit the most.

接下来我们要聊聊，哪些公司最有可能在内存领域占得先机。

So we're gonna get into what companies are best positioned for memory.

但在那之前，我们得先说清楚，为什么这次不一样，为什么时间紧迫——在市场真正意识到这场结构性变化之前，机会窗口可能很短。

But prior to that, it's probably makes sense to explain why this time is different, why the clock is ticking, there's a perhaps a limited opportunity before the market starts to realize this structural change.

我们之前提到过，由于HBM的出现以及每块GPU都必须配备它，供应端已经吃紧；但需求端同样在上升，那就是智能代理的兴起。

Now we mentioned the problems that are occurring in terms of just supply side due to the invention of HBM and just the requirement of that in every single GPU, but there's also a software side demand vector as well, which is the rise of agents.

这一点我们在上一期聊过，但我觉得有必要再强调一下：现在你不再只是跟一个单独的聊天机器人对话了。

And we've mentioned this in a previous episode, but I think it's important to double down on, is the idea that you're not just talking to a singular chatbot anymore.

当你向ChatGPT或Claude这样的大模型提问时，系统会创建一个叫KB缓存的东西，这会占用大量内存。

And when you send a query to an LLM like ChatGPT or like Claude, you create this thing called the KB cache, and that uses a lot of memory.

它把上下文信息都存进这个狭窄的窗口里。

It stores a lot of the context inside of this narrow window.

而智能代理是全新的模式，很多公司正在转向这个方向，很多人实际工作时也在用——每个代理都会独立占用自己的KB缓存，也就是说，如果你同时运行20个代理，你就需要20份独立的内存。

Now what happens with agents, which is the new paradigm, which is what a lot of these companies are shifting to and a lot of people who are doing productive work are actually using, those agents, each one consumes their own KB cache, which means it needs if you have 20 agents, you need 20 instances of memory.

这个趋势完全没有放缓的迹象：代理数量越来越多，需求越来越大，需要记住的上下文也越来越多，因此对HBM的需求也急剧上升，但产能根本跟不上。

And there is no sign of that trend slowing down where there's more instances or more demand for agents, which means more context that needs to be remembered, which means significantly more of that HBM, which is a ready supply constraint.

事实上，现在想买内存，交货期已经拖到2027年底了，因为产能根本扩不上去。

In fact, if you want to get some memory now, lead times are extended currently to the end of twenty twenty seven because there's just this inability to scale production.

我觉得，当我们回答‘这次为什么不同’这个问题时，关键在于硬件需求端和软件供应端都出现了独特的结构性变化，两者都指向一个相当长远的窗口期——你该开始担心需求了，因为趋势非常明确，而供应端满足这种需求的能力却严重滞后。

And I think when we answer the question of why this time is different, there are these unique structural changes both on the hardware demand side, but also on the software supply side that both point to a pretty far out window in which you should start to worry about demand, because there is just a very clear trend in one direction, and a very clear under indexing of the ability to supply all of the memory required to fulfill that trend.

值得注意的是，这种需求正从各个领域同时涌现。

I think that's something worth noting is that there are demands for this coming from all these different areas.

制造内存本身存在物理限制。

There's a physical constraint to building memory.

三大供应商在物理上已经无法再扩大产能。

The top three providers cannot physically build out more capacity.

这周早些时候我读到一篇报道，说不少AI客户正在主动找美光、三星和SK海力士，愿意出钱帮他们买设备，只为了换得2027到2028年的产能保证，对吧？

In fact, was I reading a story earlier this week where apparently a bunch of different AI customers have been offering Micron and the likes of Samsung and SK Hynix to help them build out or purchase equipment in exchange for guaranteed capacity or supply in 2027 and 2028, right?

你没夸张，现在他们的订单确实排到了2027年底。

You're not exaggerating that like currently they're booked up until the end of twenty twenty seven.

所以，这里确实存在物理上的瓶颈。

And so there's a physical constraint there.

通常在泡沫时期，不管是内存泡沫还是科技泡沫，这些公司都会虚增需求，加杠杆去扩大产能，结果最后导致供过于求。

Typically in a bubble, whether it's a memory bubble or a tech bubble in general, you have some version of these companies faking demand and they're like levering up to like create a supply, which ends up becoming an oversupply.

市场上产品太多了，价格因此暴跌，需求也跟着下滑。

So there's too much of the product out there and so prices kind of like crash demand kind of like sinks.

但这次情况不一样。

That isn't the case here.

如果你看看这三家制造商——它们都是上市公司——每季度的财报，就会发现这些订单不仅真实，而且已经提前付款了，这在行业里通常根本不会发生。

If you look at every quarterly earnings report from each of these three manufacturers who are all public, by the way, you can see that these back orders, that these bookings are not only legitimate, but they're already paid for in advance, which typically doesn't happen with any of these contracts.

所以这些公司账上都有实实在在的现金，可以作为未来收入的可靠预测依据。

So you have legitimate cash on the books that you can then use as a cash projection or a forward projection for a lot of these companies.

以美光为例，它现在的市盈率还不到10，而在过去的周期里，这种泡沫阶段的市盈率通常会飙到30到50之间。

In fact, if you look at Micron, their forwards earnings right now is under 10, which isn't definable within a bubble where if you look at in like past historical cycles, that typically hits around between 30 to 50 forwards earnings ratio.

所以从目前来看，这波需求是真实的。

So the point is it's legitimate as far as we can say.

不过也有怀疑者说，也许真正泡沫的是这些公司的盈利数据。

Now skeptics are saying, maybe it's the earnings that is the bubble.

他们报告的利润可能并不真实，但要戳穿这一点，比我们现在看到的情况要难得多。

So the earnings that they're reporting isn't actually real, but that's harder to pick apart than what we're seeing right now.

我本来就想提一下，这个观点真的特别关键。

I was actually gonna mention, I think that's such an important point.

虽然这些公司的股价涨得很高，比如闪迪涨了15倍，但它们的市盈率——也就是股价和盈利的比值，反映的是当前估值相对于实际价值的溢价——却只有9到12倍。

The idea that although these companies are trading so much higher, SanDisk is up 1500%, the forward PE ratio, which is the price to earnings, basically what premium it trades for on a relative basis versus what it's worth today, is only nine to 12 x.

三星的市盈率更低，只有7倍，而SK海力士更是低到5到6倍。

And it's lower for Samsung at seven x, and it's even lower for SK Hynix at five to six x.

所以，它们的定价其实并没有高估，和公司自己预测的营收和利润水平是匹配的。

So they're not really trading at a premium relative to what they're guiding for their estimates and revenue to be.

拿特斯拉来说，现在它的市盈率在40到50倍之间，微软和亚马逊也都在15到20倍左右。

If you take a company like Tesla, think that's currently trading at 40 to 50 x, and companies like Microsoft or Amazon are are, like, 15 to 20 x.

相比之下，这些内存芯片公司的估值其实很合理，和一个业绩强劲的行业应有的水平相当，这一点很值得注意。

So these memory stocks on a relative basis are actually valued pretty in line with what you would expect for a industry that's doing very well, and that's noteworthy.

那我们来具体看看这些股票，也就是大家如果想参与这场行情，可以关注的公司。

So let's get into the actual stocks, the companies that people can participate in or at least look at if they're interested in participating in this.

目前主要有三家大公司。

There are three large players.

主要有三家，分别是海力士、三星和闪迪。

There is SK Hynix, there's Samsung, and there's SanDisk.

我还看到另一个股票代码，也许你可以给我们讲讲，就是美光科技（MU），它涨了864%。

I'm seeing another ticker here, which perhaps you could walk us through, which is MU up 864%.

为什么

Why is that up

涨这么多？

so much?

美光。

Mean, Micron.

是美光。

It's Micron.

过去一年里，美光的股价涨了八倍多。

Micron is up over eight x over the last year.

主要原因在于另外两家公司，也就是最大的两家内存制造商——SK海力士和三星电子，它们都位于韩国。

The primary reason for that is the other two companies, the biggest memory manufacturers, which is SK Hynix and Samsung Electronics are based in Korea.

所以如果你在西方，尤其是美国，想投资这些股票，可没法像登录花旗的股票账户直接买入那么简单。

So if you're in the West or if you're in America specifically and you wanna get exposure to these stocks, it's not as easy as going on to your Chase stock account and purchasing these.

你没法直接这么操作。

You can't exactly do that.

中间得绕不少弯子，大多数人也不愿意费这个劲。

There's a few more hoops to jump through and most people aren't willing to do that.

所以另一种方式就是投资美国主要的内存制造商。

So another way of getting exposure to this thing is to buy the main American memory manufacturer.

其中有一个明显的领头羊，就是美光。

And there is one clear leader, which is Micron.

他们的股价确实涨得厉害，不过说句公道话，美光的产品确实不错，也是技术领先的供应商之一。

So their stock has absolutely said, and to be fair in Micron's defense, they produce a really good product and they are one of the advanced suppliers.

但你要说谁才是真正的王者，那还是SK海力士。

But you wanna talk about who the king is, it is SK Hynix.

他们是英伟达及其所有未来显卡内存的独家主要供应商。

They have been the sole and main provider for memory, for Nvidia and all their future GPUs.

事实上，就像英伟达包下了台积电大部分产能来生产GPU一样，他们也把三星海力士的全部产能都预定了。

In fact, in the same way that Nvidia has booked out the majority of capacity for TSMC to build their GPUs, they have done the same with SK Hynix.

所以之前我提到的那些人，之所以愿意提前付钱给海力士锁定未来供应，是因为英伟达已经垄断了海力士今后所有的产能。

So the reason why those people that I mentioned earlier are trying to like pay off SK Hynix for guaranteed supply in the future is because NVIDIA owns the monopoly on all capacity for SK Hynix going forward.

海力士是最强的，三星电子排第二。

So SK Hynix is like the strongest, Samsung Electronics is next.

三星电子做的业务太杂了，所以投资它并不是纯粹押注内存市场。

Samsung Electronics does a ton of other stuff, which is why it's not a pure concentrated bet on memory.

然后就是美光，算是美国这边的选择。

And then you have Micron, which is like the American bet.

接下来我想聊聊我之前提到的那个ETF，乔什，就是六周前推出的那个，目前增长最快的那只。

Now, what I wanna speak about is that ETF I mentioned earlier, Josh, that launched six weeks ago, that is like the fastest growing ETF.

过去五周，它的资金流入已经达到了50亿美元，甚至可能接近60亿美元，这只基金叫DRAM。

I think it's like $5,000,000,000 worth of inflows over the last five weeks or $6,000,000,000 It's this thing called DRAM.

这个代码太棒了，它是一只专注内存的ETF。

It's a fantastic ticker and it's a memory ETF.

它这么特别，主要有两个原因。

And the reason why it's so unique is two reasons.

第一，它在西方市场很容易买到；第二，它让你同时接触到我前面提到的那两家韩国公司。

One, it's accessible in the West, but two, it's a basket exposure to those two Korean companies that I mentioned earlier on.

它持有SK海力士的股票。

It owns SK Hynix shares.

它也持有三星电子的股票，同时还持有美光的股票。

It owns Samsung Electronics shares, but it also owns Micron.

实际上，这三家公司加起来，大概占了整个投资组合的七成左右。

In fact, all three of these companies, I think, make up around 70% of the entire basket.

所以这是一个非常聚焦于内存、特别是AI内存的投资标的。

So it's a very concentrated play on memory, AI memory specifically.

此外，它还少量持有闪迪和西部数据这类公司的股份，它们生产的是另一种同样对AI产业至关重要的存储器。

And then it also owns smaller stakes in the likes of SanDisk and Westgate, which is a different type of memory that is still equally important for the AI trade.

我们这里有张很好的图可以展示这一点。

We have a great visual on this.

我觉得大多数人会感兴趣的就是这个，直接看看它具体持有哪些股票。

I think this is probably what most people will be interested in participating in, just showing the holdings within this.

如果你真想投资韩国的这些公司，这才是你该选的途径。

I mean, this is the access that you want if you want access to the Korean companies.

它确实有。

It has Yeah.

其中SK海力士占24%，美光科技占25%，西部数据占5%，希捷占6%，三星占20%。

24 SK Hynix, 25% Micron Technology, 5% in WD, 6% Seagate, Samsung is 20%.

所以这大概是目前最均衡的配置组合了。

So this is probably the most balanced basket of exposure.

我个人最看好的就是这个。

This is personally what I'm most interested in.

如果你不想花太多心思研究，买进来就放着，别管它。

If you don't wanna think about it too much, you just set it and forget it.

伊贾兹，你之前跟我提过这个，从那以后，它的表现简直炸了。

Ejaaz, you mentioned this to me a little while ago, and since then, it has gone absolutely nuclear.

如果不想费劲去挑哪些公司能赢、哪些会输，这里似乎是唯一的选择。

This seems to be the singular place in which people are converging on if they don't wanna do all the hard work of picking winners and losers.

事实上，最近这个话题热度飙升，连产品本身都升级了一次。

In fact, this has recently gotten enough attention to where they have leveled it up even once more.

现在出现了一个叫 RAM 的股票代码，而不是以前的 DRAM，这是一个覆盖整个内存市场的ETF，而且很快就要推出一款两倍杠杆做多内存的基金。

So now there is a ticker that goes by the name of just RAM instead of DRAM, which is the ETF that covers basically the entire memory market, and this is a two x leverage long memory ETF that is going to be launching very soon.

如果你真想大胆押注，而且长期看好内存行业，现在已经有工具让你能彻底放手一搏。

So if you really wanna go risk on, if you really believe in the memory trade over a long period of time, there are tools available to be able to enable you to gamble on this to the full extent.

我可能还是会选择普通的 DRAM 基金，但如果你野心更大，这确实是进入这个市场的最佳途径——尤其是像我们大多数人一样关注美国市场的人，想布局韩国这些巨头的话。

I'll probably just stick with the regular DRAM ETF, but for those ambitious, this is probably the best way to get access to those individuals, particularly if you listen in in The United States like most of us do, and you want access to those Korean powerhouses.

这是个相当不错的方式。

This is a pretty good way of doing it.

我猜不少听众听到我们这么说，心里会想：这些股票都涨了五到八倍，我们是不是已经错过机会了？

Now I'm sure what a lot of listeners are thinking as you're hearing what we're saying is, okay, well, all these stocks are up five to 8X, we've completely missed the trade.

那我们为什么还要考虑投资这个呢？

Why on earth would we potentially want to invest in this?

这是不是泡沫？

Is this a bubble?

这会崩盘吗？

Is this gonna pop?

其实也不难理解，这种情况确实有可能发生。

And, you know, there is feasibility to argue that that might be the case.

这次真没什么不一样。

This time isn't actually different.

如果我们看看这些内存供应商的历史，比如2000年左边这张图，当时有十一家主要的内存制造商。

In fact, if we look at the historical context of these memory suppliers, on the left here in 2000, there were 11 main manufacturers of memory.

要明白，过去二三十年里，内存行业一直经历着起起落落。

And it's important to understand that memory has gone through boom and bust cycles for the better part of the last two to three decades.

每次繁荣之后都是萧条，最后剩下的玩家越来越少，这就引到了右边——这挺有意思。

And with every boom and bust cycle, you end up with fewer players, which brings us over to the right over here, which this is funny.

这是2013年的情况，乔什。

This was in 2013, Josh.

所以这三家坚持了下来。

So these three have kind of stuck it out.

现在有三家公司几乎垄断了整个内存制造市场。

You have three companies which basically dominate memory manufacturing in its entirety.

那么问题来了，这次会不会重演同样的情况？

So the question now is, is this gonna happen in the same way?

关于这一点，其实有两种不同的看法。

Now there's two sides of this argument.

反对这种观点的一方认为，人工智能创造了前所未有的需求，就像我们之前提到的，生成式AI需要比以往任何技术变革都要多得多的内存，用于每一代新的显卡。

The side that argues against this thesis is AI has created a demand sink for the reasons that we mentioned earlier, Genetic AI, the fact that you need it for all successive GPUs in larger quantities than we've seen in any other technology shift.

这确实是事实。

And that is the case.

人工智能已经渗透到每一个行业。

AI is pretty pervasive across every single industry.

如果你假设每个行业都会引入某种形式的人工智能模型，改变行业运作方式，或者部署能自动化——比如说，10%到40%工作岗位的AI代理，那就可以推断，这将需要大量的内存、更多的显卡、更多的CPU，以及更多支持这些系统的代理。

So if you assume that every single industry is gonna have some form of AI model that changes the way that the industry works or AI agents that can automate, say, I don't know, 10 to 40% of working roles there, you can then extrapolate to think, well, okay, that's gonna need a bunch of memory, a bunch more GPUs, a bunch more CPUs, a bunch more agents to be able to facilitate that.

所以你可以估算出所需内存的总量，这也是为什么这些股票可能还有很长的路要走。

So you can then kind of mark up the amount of memory that is required, and that's why these stocks have probably a longer way to go.

但持怀疑态度的人会说，现在确实是这样，但如果未来推出了一款对内存需求低得多的AI模型或GPU，那岂不是会导致内存供过于求？

Now, skeptic would argue, okay, that's the case for now, but what if we launch an AI model or a GPU which requires much less memory, won't that mean that memory oversupply will happen?

当美光的晶圆厂投产，市场上突然涌进海量内存时，会发生什么？

What happens when Micron's fab gets launched and we have a heck ton more memory?

这会让价格下跌吗？

Will that decrease prices in demand?

你可以说会，也可以说不会。

You could argue yes and no.

根据杰文斯悖论，人们可能会想要更多相关产品，结果反而需要更多的内存，这两种观点正好代表了对立的两方。

In Jevan's paradox, people will just maybe want more of the things, so you may need even more amounts of memory, but those are both sides of the opposing argument.

好吧，我想就你支持‘这次不一样’这个观点提出一点不同看法，前提是你真觉得这次不同。

Okay, so I guess I have something that I'd want to challenge you on in kind of supporting why this time is different, if you think this time is truly different.

因为从历史来看，内存周期通常持续大约十八个月，然后会回调百分之四十到七十五。

Because when we look at the historical patterns of what these memory cycles are like, they generally last about eighteen months and then experience a pullback of anywhere from like forty to 75%.

这种情况在1995年、1999年、2017年和2023年都出现过，而当前这一轮实际上是从2024年第四季度开始的。

This happened in 1995, '99, 2017, 2023, and now this current cycle started actually the fourth quarter of twenty twenty four.

现在我们已经到了第18个月，按以往规律，这时候市场要么开始恶化，要么迎来转机。

So we're about month 18 now, which is where things should be either taking a turn for the worse or changing for the better.

所以，如果你在考虑个人的投资配置，以及你对未来的判断，现在正是非常关键的时刻。

So if you're thinking about how you allocate personally and how you believe this to go, this is a pretty pivotal time.

如果你把市场看成一张图表，现在我们就站在了那个关键阻力位上。

If you're looking at it as if there's like resistance on the chart, we're at that resistance right now.

我们正处于第18个月。

We're in month eighteen.

你觉得这次真的不一样，还是说这只是周期的自然表现？

Does it seem like it's more probable that this actually is different or is this the natural cycle?

因为从各种迹象来看，这次确实不同，但你也提到，那个ETF在五周内就翻了一倍。

Because it I mean, all the signs point to it being different this time, but you mentioned, I mean, the ETF had doubled in five weeks.

所以，之前那种容易赚的收益，现在基本都没了。

So a lot of that like easy beta that people had is gone.

还有上涨的空间吗？你预期的时间跨度是多久？

Is there still opportunity for the upside and what timescale are you thinking about?

如果问我个人看法，我觉得长期来看还有很大的上涨空间，但短期股价回调我一点也不意外。

If you ask me personally, I do think that there is long term, a lot more upside, but it wouldn't surprise me if we see a short term retrieval from stock price charts.

这几个月以来，DRAMETF和整个内存板块几乎每周都在涨，已经连续六周了。

Like we have had like, I think six weeks of constant uptick in this DRAM ETF and in just memory stocks in general.

这波上涨主要是靠季度财报和一堆合作公告推动的。

And that has been on the basis of these quarterly earnings and a bunch of other partnership announcements.

这种势头不可能每周都持续下去。

That can't keep happening every single week.

总有一天，能推动股价的新消息会用光的。

At some point, we're gonna run out of news to kind of push prices up further.

我觉得接下来会迎来经济衰退。

And I think we're gonna see a recession.

我觉得大家想得有点太多了。

I think people are kind of overthinking it.

但反对观点会说，现在的远期盈利倍数其实还很低，有些人可能会看到这一点，但我真不觉得股价能每周都这么持续暴涨。

Now, the argument against that would be, well, forwards earnings ratios are still quite low and some people might see that, but I just don't think we can see a chart go up exponentially every single week.

我觉得迟早会回调一下。

I think we're gonna have some backpull.

让我觉得乐观的是，我之前提到的那些订单 backlog，不只是签了合同，甚至在产品还没交付之前，客户就已经预付了款项。

The thing that gives me optimism is those backlog of orders that I mentioned earlier that aren't just signed sealed, but before they even delivered, people have paid for it upfront.

最近谷歌和美光就达成了一笔交易，为他们的TPU芯片，对方直接预付了40%的款项。

There was a recent deal that was struck between, I believe it was Google and Micron for their TPUs, and they paid 40% upfront.

这说明这些公司对内存的需求非常迫切，而且真正有需求的都不是小公司。

So the point is like these companies really want memory and the companies that really want memory are not unknown companies.

而是那些现金流充裕、自由现金流充足的科技巨头，也就是所谓的‘七巨头’。

They're the mag seven who have the cash flow and free cash flow to pay for it and back it up.

如果全球最聪明、最有价值的公司都愿意押这么大笔钱，那我们确实有理由相信，这次的情况可能真的不一样，也更靠谱。

So if you have some of the smartest, most intelligent companies, most valuable companies in the world putting up money and big risks like that, you could argue that this time is different and more legitimate.

当然，也不能排除一种可能：这一切最终会崩盘，需求其实是虚假的，Anthropic和OpenAI那边夸大了需求数据，结果引发连锁反应，导致整个市场崩塌。

Now, of course, there is the non zero chance that all of this blows up, that the demand is actually fake, that Anthropic and OpenAI on their demand side actually end up inflating their numbers too much, and that could lead to a waterfall cascade of all these things crashing.

但我目前看不到这种情况会发生的结构性基础。

But I don't see the structural context for that happening right now.

一年后，是更高还是更低？

Twelve months from now, higher or lower?

更高。

Higher.

我也觉得会更高。

I think higher too.

没错。

Yeah.

这感觉就是主流趋势。

That feels like the vibe.

趋势非常明确，当然中间会有波动，但未来走向更高，是基于我们现有信息能合理预期的乐观前景，我有信心押注这一点。

There is like very clear trajectory and sure, there will be a bit of volatility along the way, but higher feels like feels like an optimistic version of the future that we have enough information to go on that, I could feel comfortable about about betting on.

所以我觉得，这大概就是内存行业最核心的论据了。

So I think that's probably most of the the case for memory.

没错。

That's Mhmm.

为什么这很重要？现在有哪些过去十八个月周期里没有的独特市场结构？如果你有兴趣参与，机会在哪里？

Why it's important, what unique market structures are now available that weren't in the past version of these eighteen month cycles, and where the opportunity lies if you are interested in participating.

我想问问你，正在听这段话的你，对于这十二个月的时间尺度，你怎么看？或者你认为最好的参与方式是什么？

I guess to ask for you, the person who's listening to this, is what do you think on that twelve month time scale, or what is the best way for you to get exposure?

有一张图表让我个人特别受打击，Ejaaz，因为我一直喜欢自己组装电脑、搭建硬件。

There's this chart that really devastates me personally, Ejaaz, as someone who loves to create my own custom PCs and build hardware.

成本啊。

The cost yeah.

如果你往下翻到底，就会发现消费类产品所用的内存或内存系统的成本已经高得离谱。

If you scroll down all the way to the bottom even, the cost of these memories or memory systems for consumer products is through the roof.

这让人有点失望，因为如果你想买台新电脑，或者想入手一台新的Xbox，甚至最新的Steam设备，全都推迟了，或者涨价了。

And that that's been a little disappointing because if you wanna just go buy a PC, if you wanna get a new Xbox, even the new Steam machines, they've all been delayed or the prices have been increased.

事实上，现在想买一台PlayStation 5，价格比它五年前刚上市时还高，就是因为这些供应链问题。

In fact, if you wanna go buy a PlayStation five, it costs now today than it did, what was it, five years ago when it first came out because of these supply constraints.

所以我们开始看到这种影响蔓延到整个大众消费市场，这让人有点沮丧。

So we're starting to see it permeate out into the general broad consumer market, and that feels a little discouraging.

我不禁想知道，这还能涨到多高。

I wonder how high that can go.

消费者能承受多久，才会开始撑不住呢？

I wonder how much tolerance the consumer market has before it starts to bend.

接下来我也会关注这些价格变化，特别是对普通消费电子产品的影响。

So that's something I'm gonna be looking at also is following these prices as it relates to just general consumer hardware.

iPhone 会受到影响吗？

Is the iPhone going to suffer?

它们的价格会上涨吗？

Are those prices going to go up?

智能手机的平均价格。

The average smartphone cost.

类似这样的东西，我也在留意，看它们能不能扛得住。

Things of this nature are also on the watch list to see if they could they could bear it.

如果能撑住，那我们就没什么好担心的了。

And if they can, then we're probably good.

如果撑不住，那也值得我们留意一下。

If not, it's something to just take some note of.

我个人觉得，内存会像当年的GPU一样成为最大的赌注，那时候英伟达的市值突破了五万亿美元。

Personally, memory is gonna be one of the biggest bets in the same way that GPUs were, and it saw NVIDIA breach above $5,000,000,000,000 market cap.

我觉得这次内存也会经历同样的事情。

I think the same thing is gonna happen in memory this time.

现在市场上有三家主要玩家，美国的美光就是其中之一，它也会在内存领域复制同样的奇迹。

And we have three concentrated players, one in America, Micron, that is gonna do the same thing for memory.

我对这个前景非常看好。

So I'm excited about it.

我对这些周期性的繁荣与萧条确实有点担心，但也许这次真的不一样了。

I am slightly nervous about the boom and bust of these different cycles, but I guess we'll see if this time is different.

基本面一直都很强啊。

Fundamentals- It's been high, man.

东西都在上涨。

Things are going up.

但总得有个尽头吧？

Well, at some point it's gonna break, right?

我看着消费级PC内存的图表，心想迟早有人要放弃。

Like I'm looking at this like consumer PC memory chart and I'm like, at some point, people are just gonna throw their hands up in the air.

价格太高了，总得有人调整，要么是苹果这样的设备厂商站出来说，算了，我们不玩了，要么就得找别的办法。

Like it's gonna be too expensive and something will need to shift, whether it is these consumer device makers like Apple pushing back and saying, listen, we're just not gonna do this, or we need to find an alternative.

我不知道那会是什么，但有一点是肯定的——AI模型的需求只会越来越大。

I don't know what that is, but one thing is for sure is the demand for AI models.

AI代理的需求也是持续攀升。

The demand for AI agents is just going up and to the right.

如果消费者这边撑不住，那肯定就是企业端在扛着。

If it's not coming from consumer, it's coming from enterprise.

所以真不知道这个难题最后怎么解决。

So just don't know how this conundrum is gonna be fixed.

也许我们可以开发一种新型的模型，但不管怎样，我们都会持续关注，至少未来六到十二个月，我对内存市场还是持乐观态度。

Maybe we create an alternative type of model, but either way, we will be tracking it and I'm bullish memory at least for the next six to twelve months.

就这些了。

But that is it.

我想知道，节目的听众们有没有其他看法？

I'm curious whether the listeners of the show have any other takes.

过去一周半，我们已经推出了另外两期以投资为主题的节目，反响非常热烈。

We've released, I think, two other investment themed episodes over the last week and a half, and the feedback has been tremendous.

听众们既支持我们的观点，也明确指出我们哪里错了。

Both comments telling us why we're right and why we are explicitly wrong.

我想听听双方的声音，这能帮我们更好地判断下一步该往哪个方向走，该聚焦哪些话题。

I wanna hear from both sides because it helps us kind of like figure out where we're going next and what to cover next.

关于内存这块，我们有没有漏掉什么重要的点？

Does anything that we missed on the memory side of things?

也欢迎你们告诉我。

Also let us know.

最后，我们注意到大家希望深入聊聊上期节目中提到的底层技术和基础设施。

And finally, we heard you on wanting to talk about or unpack the layers of substrates and infrastructure that we covered on our previous episode.

我们很快就会策划并推出这一期内容。

We are gonna plan and deliver that episode fairly soon.

不过这周晚些时候，我们要迎来利奥波德的行情了。

But this week later on, we're getting Leopold's fires.

恭喜利奥波德周到来。

Happy Leopold week.

这事儿可太重要了。

This is This is huge.

我们上两期节目，都是迄今为止播放量最高的，就是因为讲了利奥波德。

Our last two episodes, those were those were among the biggest ever because it was covering Leopold.

我可不是想邀功，但如果你去搜利奥波德的投资组合，或者上X平台看，我们会是最早推动这个话题的人。

I don't wanna say I don't wanna take any credit, but if you search up Leopold's portfolio or if you go on x, we're very much the beginning of the Leopold narrative.

现在，这正是那个故事的下一章。

And now it was the next chapter in that.

两年时间里，Leopold基金的态势感知规模从15亿美元涨到了60亿美元左右。

So Leopold's fund situational awareness has gone from 1,500,000,000 to, like, 6,000,000,000 in the course of two years.

对。

Mhmm.

他们每个季度都会发布一种叫13F的文件，里面会披露最新的持仓情况。

And they publish these things called 13 f filings every quarter that reveals the new set of positions.

这份13F文件预计这周末之前就会公布。

That 13 f filing is going live sometime by the end of this week.

截止日期是周五。

It's due by Friday.

我们会密切跟进，一旦文件发布，马上推出一期节目，深入分析下一季度关于态势感知的所有新投资逻辑。

So we will be very ambitiously monitoring this, and as soon as it releases, we'll have an episode ready to go to talk about all of the new investment thesis around situational awareness for the next quarter.

敬请期待。

So stay tuned for that.

我们已经准备就绪，等文件一出来就立刻上线。

We are we are locked in and ready to go as soon as that one drops.

是的

Yep.

没错

Exactly.

不过除此之外，应该就这些了，我们先到这里，下一期再见。

But aside from that, I think that is it, and we will leave you guys until the next episode.

下期见。

Let's see you then.