安费诺受益于AI基础设施的建设，因为每机架的内容含量高且业务模式多元化。
逐段跳播
这家公司并非家喻户晓，名为安费诺（Amphenol），其股票代码为 AP。因此，这是一家非常专注于科技基础设施物理层的企业，确切地说，安费诺设计并制造微型引脚、精密插头、高速铜缆。
这些不起眼但至关重要的硬件连接器，他们销售给全球的企业服务器制造商以及汽车制造商。因此，他们的业务在人工智能和数据中心热潮之外也是多元化的。
company, not a household name, a company called Amphenol, and that the ticker for that is AP. So this is a business that's very much focused on the physical layer of tech infrastructure quite literally. So Amphenol designs and manufactures uh microscopic pins, precision plugs, uh high-speed copper cables. So this very unglamorous but essential hardware connectors that they sell to the enterprise server builders as well as auto manufacturers globally. So they have a a diversified business outside of you know the AI and data center boom.
我们之前讨论过，建设过程中遇到的物理限制是人工智能繁荣的一种守门人，但数据传输的物理限制也是如此。当云巨头购买英伟达显卡时，这些芯片如果没有非常专业、精确的电缆和连接器配置就无法通信，这正是安费诺所销售的。
他们在这个硬件细分领域长期以来一直是领导者，这使得他们成为当前以及我认为未来不可或缺的公用事业提供商。
um something we've talked about you know the physical limitations that we are seeing in the buildout are one sort of gatekeeper of the AI boom but it's also the physical limitations on data transmission so you know when cloud giants are buying Nvidia graphics cards those chips can't communicate without very uh specialized precise cables connectors configurations that's what Amphanol sells they are the leader and have been for a long time uh in this hardware uh niche and so that makes them a essential utility provider um in this current time and I think as well uh moving forward
我认为他们的商业模式另一个真正引人注目的方面是，随着芯片性能越来越强大，每个服务器机架的总美元含量也在增加。在旧的服务器设置中，传统的铜线足以处理标准网络流量，但在当今的数据传输需求下，这已经不够了。
如果单个连接器出现哪怕轻微的信号衰减，整个训练模型可能会立即退化。这赋予了安费诺定价权，因为很明显，超大规模企业不会在保护其计算集群的关键硬件接口上偷工减料。
the other thing that's I think really um compelling about their business model is that their total dollar content per server rack scales up as chips get more powerful so with the old server setups you had traditional copper wiring it was more than enough to handle standard web traffic but in today's uh you know data movement requirements um that is not enough um if a single connector experienc experiences even minor signal degradation, you can have an entire training model that degrades instantly. And so this is giving amphenol pricing power because you know obviously hyperscalers they're not going to cut corners on the vital hardware interfaces that are protecting their compute clusters.
所以，我觉得有趣的一点是，无论数据中心运营商决定围绕英伟达芯片、定制云处理器还是替代平台来构建其基础设施，他们仍然需要像安费诺这样的公司，而且往往是安费诺，来购买铜缆和光纤路径，以连接整个设施。
So, you know, one of the things that I find fascinating about this is whether a data center operator decides to build their infrastructure around Nvidia silicone and custom cloud processors or even alternative platforms, they're still going to require a company like Amphenol and often Amphanol to purchase the copper the fiber optic pathways to tie that facility together.
为了提供一些数据，最近一个季度他们实现了约 40 亿美元的收入，同比增长 18%，营业利润率在 20% 左右。
Just to put a few numbers on this, the recent quarter they delivered revenue of about $4 billion. That was an 18% year-over-year growth rate, operating margin in the low 20s.
你可以将人工智能模型从封闭式改为开放式，或者改变地理位置，但你无法改变高速信号必须在芯片、服务器和机架之间传输的物理现实。作为英伟达平台等架构的主要合作伙伴，安费诺实际上对部署的每一件物理基础设施都收取通行费。
Um, you know, you can change the AI model to a closed one to an open one or the geographic location, but you can't change the physical reality that high-speed signals have to move between the chips, the servers, and the racks. And as a primary partner for architectures like Nvidia's platforms, empanel essentially gets paid a toll on every piece of physical infrastructure that's deployed.
先进节点不仅需要更多的线缆，它们使用非常密集的高速布线、光纤和先进的配电硬件。安费诺在每个 AI 机架上的组件价值或含量远高于传统云服务器。
Um, and advanced nodes, they don't just require a few more briars. They use very dense, high-speed cabling, fiber optics, and advanced distribution hardware. And Amphanol sells vastly more component value or content per AI rack than they ever did for uh the traditional cloud servers.
他们还拥有全球制造足迹，因此能够供应大型科技公司，也能无缝供应本地化的主权供应链，因为主权人工智能是我们正在看到的巨大趋势。
Um, and they also have a global manufacturing footprint. So they're able to supply uh the big tech companies but also the localized sovereign supply chains seamlessly as sovereign AI is a huge trend that we're seeing.
这是一家有趣的企业，也许在人工智能革命中较少被提及，是人工智能竞赛中那些悄然必不可少的基础设施投资之一。
So interesting business maybe one that's less talked about in the AI revolution. One of those kind of quietly essential infrastructure plays on the AI race.
是的，我对它不太熟悉，所以以为这是一家非常小的公司，但这家公司市值接近 2000 亿美元，过去 12 个月营收达 290 亿美元，所以绝对不是一个小角色。
>> Yeah, you know I wasn't very familiar with it. So I thought this would be a very small company, but this is a company worth almost $200 billion generating over the last 12 months $29 billion. So definitely not a a small player,
但正如你所说，随着下一代服务器机架系统变得越来越大、越来越复杂，每个机架的内容更多意味着更多的资金流向像安费诺这样的公司。
but yeah, like you said, it's a very crucial player as nextg uh server rack systems become bigger and bigger, more complex. Guess what? More content per rack means more money flowing to a player like Amphanol.
而且，正如你所说，他们是多元化的，涉及电力、国防系统、汽车、重型工业设备，这意味着在现金流方面，他们有稳定稳定的现金流，而且显然他们一直在收购一些较小的公司并成功整合，这意味着他们的竞争力不断增强，当然这之所以有效，是因为他们设法找到了合适的公司，以合适的价格收购，并以合适的方式整合。
Also, like you said, they are diversified. their power, defense systems, cars, heavy industrial equipment, which means cash flow wise. They have steady steady cash flow and apparently they've been acquiring certain smaller companies and integrating them as well quite successfully which means that their competitiveness grows and grows and grows and of course this works because they managed to find the right company, buy them at the right price and integrate them in the right way.
所以，如果英伟达增长如此之快，那么可以肯定的是，安费诺现在和可预见的未来都在赚很多钱。
So if Nvidia is growing this quickly rest assured that Amphanol is is making quite a lot of money right now and in the foreseeable future.
这个频道对 $AP 的历次判断
The Motley Fool 在这只股票上只有这一条判断。