$LRCX · Chip Stock Investor · 2026-11-03
Lam Research 是存储供应链中持久的瓶颈：它深度参与制造工艺本身的设计，而不只是设备供应商，并通过设备、升级和服务获利。这一地位无法靠从别处采购或自行设计轻易解决，这也是长期投资者反复关注它的原因。
逐段跳播
当你想到不仅需要被设备填满的新洁净室空间，还想到设备升级时，你自然而然会想到谁？答案就是——Fab Five，那经典的五大设备厂商。尤其是像 Lam Research 这样的公司，它历史上一直专注于存储市场。
嗯哼。而且它们全都协同运作，Applied Materials 在这里也扮演着非常重要的角色。
And who should you automatically think of when you think of not just new clean room space that needs to be filled up with equipment, but also equipment upgrades? Ta-da. The Fab Five.
The good old Fab Five. Especially company like Lam Research, which has historically been focused in the memory market. Mm-hmm. And all of them work together. Applied Materials plays here very heavily as well.
这些资料由 Lam Research 提供。我们先来看看这栋房子或现有公寓可以怎样被分割。这是一层老式 DRAM 芯片的侧视图。DRAM 需要供电才能保持内存中的数据，它用一个电容来保持这个电荷。
一个电容加一个晶体管就构成一个存储位。而把这些电容和晶体管对分隔开的，是字线和位线，中间标着 BL 和 WL。右边再通过示意图看一遍。这里的挑战在于，如何把它从二维结构变成三维结构？
这正是“分割房子”或“在现有地皮上盖多层楼”这个类比开始有点站不住脚的地方，因为这里发生的是：左边那张电容、晶体管以及字线和位线的示意图被侧过来放倒。然后它们就可以把这些结构垂直堆叠起来。
但这仍然解决不了问题，因为你如何把所有这些不同的存储位互连起来？这时类比又说得通了。中间那条粗大的垂直线 BL 就是位线。把位线想象成一部中央电梯，然后左右两侧每一层的每个停靠点就是你的字线。
那里有一小段走廊，然后是公寓。所以他们算是把传统结构侧放过来，在中间插了一部大电梯，而这是正在进行中的工作。设计出来是一回事，真正能够制造出来则完全是另一回事。但这就是为什么作为长期投资者，我们最终总是回到那些更持久的瓶颈上，那些不那么容易解决的瓶颈。
因为 Lam Research 不只是为制造商提供设备。它们实际上深度参与了工艺本身的设计和开发，然后当然通过出售机器或升级及其他服务来变现。但你明白这个意思。这不是一个容易解决的瓶颈。
你不能只是决定说，Lam Research，我们决定去别处买设备，或者自己设计设备，或者嘿，你们直接扩大设备产能，问题就解决了。那些设备制造商开发了行业里如此多的东西，如此多的设计工作和制造工作，资金也流经它们，所以这在供应链行业流程图中就是一个相当持久的位置。
These are provided by Lam Research. Let's look at how that house or existing apartment can be subdivided first. This is a side view of a one-layer good old-fashioned DRAM chip.
DRAM requires power to hold the data that's in memory, and it holds that charge in with a capacitor. One capacitor and one transistor makes up a bit for memory. And separating those pairs of capacitors and transistors, you have wordlines and bitlines, BL and WL there in the middle.
Another look at that via schematic on the right. The challenge here is how do you take this from a two-dimensional structure to a three-dimensional one? This is where the analogy of subdividing the house, or building on top of your existing real estate to build a multi-story building starts to break down a little bit because what's happening here is same schematic on the left of the capacitor and transistor and the wordlines and bit lines getting tilted on its side.
And then they can stack those structures vertically on top of each other. But that still doesn't do the trick because how do you interconnect all of those different bits in storage?
Now the analogy works again. There's the bit line there in the middle, the big fat vertical line BL. There's the bit line. Think of that like a central elevator, and then each stop on the floor to the left and the right, there's your wordlines.
There's a little hallway and then the apartment. So they've kind of tipped the traditional structure on its side, stuck a big elevator down the middle of it, and this is ongoing work.
Designing it is one thing. Actually being able to manufacture it is something else entirely. But this is why we always ultimately, as long-term investors, end up coming back to the more durable bottlenecks, ones that are not so easy to solve.
It's because Lam Research doesn't just provide equipment for manufacturers. They actually are integral to designing and coming up with the process itself, and then they monetize it, of course, via the sale of the machine or upgrades and other services.
But, you get the idea. This is not a bottleneck that is easily solved. You don't just decide, Lam Research, we've decided to go get equipment somewhere else or design our own equipment or, hey, just expand your equipment capacity and problem solved.
Those equipment makers develop so much of what the industry does, so much of the design work and manufacturing work, and the money flows through them that this is just a pretty durable spot in the supply chain industry flowchart.