一份可能撼动市场的财报

有一份财报可能会影响市场走势。

There's one earnings report that could move the market.

《Motley Fool 隐藏宝石投资》现在开始。

Motley Fool Hidden Gems Investing starts now.

欢迎收听《Motley Fool 隐藏宝石投资》。

Welcome to Motley Fool Hidden Gems Investing.

我是乔恩·奎斯特。

My name is Jon Quast.

今天由我主持，还有我们的嘉宾马特·弗兰克尔和瑞秋·沃伦。

I'm your host today, and I'm joined by our guests, Matt Frankel and Rachel Warren.

今天我们会多次打开听众来信，聊聊数据中心，还有并购的话题。

Today, we're gonna dive into our mailbag a couple of times to talk about data centers, also talk about mergers and acquisitions.

不过首先，我们想先说说本周的新闻。

But first, we wanted to get to our kind of news of the week.

本周，英伟达将公布季度财报。

This week, Nvidia is going to report quarterly earnings results.

顺便说个周末的小故事，居然还有人不知道英伟达是什么，我还得跟人解释。

And just to share an anecdote from over the weekend, it's amazing that there are some people who still don't know what Nvidia is, and I had to explain it to somebody.

所以我想在播客里也讲一下，不假设所有人都知道英伟达。

So I wanna do that here in the podcast, not take for granted that everybody knows what Nvidia is.

这是一家市值五万亿美元的公司。

This is a $5,000,000,000,000 company.

所以非常重要，非常非常重要。

So mass very, very important.

它其实是从游戏起家的，但它生产的那些GPU正是推动AI革命的核心。

Really kinda got its start in gaming, but those GPUs that it makes are what is powering the AI revolution.

这些GPU正被疯狂抢购，用来填满你可能听说过的全国各地的数据中心。

And these are what are being bought up like crazy to fill the data centers that you might have heard about that are going in around the country.

所以这是一家非常重要的公司，它将在本周晚些时候，准确说是周三，公布财报。

So very, very important company, and it is reporting its earnings later this week, Wednesday to be precise.

那么在我们开始之前，马特，给我们讲讲英伟达，以及这次财报我们该期待些什么。

And so just as we get started here, Matt, tell us about Nvidia and what we should look forward to in this report.

对。

Yeah.

嗯，我是说，为了让你说的更有背景一点，英伟达其实发明了GPU，他们在数据中心GPU市场大约占有95%的份额。

Well, I mean, just to put what you said in a little more perspective, Nvidia actually invented the GPU, and they have roughly a 95% market share in the data center GPU space.

所以他们是一个非常主导的玩家。

So they're they're a a dominant player.

这就是为什么所有那些数据中心，大家都在推动建设的，里面装的全是他们的芯片。

That's why all, you know, all these data centers that everyone's pushing back being built in their talents that they're it's their chips that are filling them.

所以他们预计本季度营收约为920亿美元。

So they're expected to report about $92,000,000,000 in revenue this quarter.

是十亿吗？

Billion would it be?

他们的管理层给出了910亿美元的指引，但说实话，投资者现在基本上就是假设他们会超出预期。

Their management guided for 91,000,000,000, but, honestly, investors kind of just simply assume that they're gonna beat expectations at this point.

考虑到过去几个季度，这个假设挺合理的。

It's a pretty fair assumption given the the past few quarters.

所以这大约是同比100%的增长，而且还有环比加速，也就是说每个季度的增长率预计会提升。

So that would be roughly 100% year over year growth as well as a sequential acceleration, meaning that the growth rate from quarter to quarter is gonna is is expected to pick up.

而且这还是建立在一个已经非常庞大的营收基础之上。

And that's off of an already pretty enormous revenue base.

所以，当然，数据中心是重点关注的领域。

So, I mean, of course, data center is the big piece to watch.

他们也做其他业务，但说实话，英伟达做的其他所有事情，比如你提到的游戏芯片、专业可视化，也就是图形设计芯片之类的。

They do other things, but quite frankly, everything else Nvidia does, their gaming chips that you mentioned, pro visualization, which is like graphic design chips and things like that.

还有汽车部门，他们为汽车用途制造芯片。

The auto division, they make chips for automotive use.

这些业务跟数据中心业务相比，基本上可以忽略不计了。

They're essentially rounding errors at this point compared to the data center business.

对。

Yeah.

如果它们是独立公司的话，那它们会是非常庞大的公司，但我想回到你刚才说的那个点。

And they would be enormous stand alone companies if they were stand alone, but I I just wanna circle back to what you just said here.

我们说的是全球最有价值的公司，年增长率达到100%，也就是一年翻一番。

We're talking about the world's most valuable company growing revenue at a 100% year over year, doubling year over year.

这简直太惊人了。

I mean, this is just absolutely astonishing.

但如果说这还不够让人震惊的话，另一个让人震惊的点是英伟达过去十年的利润率。你知道，十年前毛利率大概是58%，这已经不错了。

But one of the other astonishing things, if that wasn't astonishing enough, is Nvidia's margin over the last decade, you know, ten years ago, a 58% gross margin more or less, and that's good.

但现在毛利率达到了74%，基本上他们每卖出100美元的产品，直接成本只有36美元。

But right now, sitting at 74% gross margin, basically, every $100 a product that they sell, it only costs them $36 to make it in in direct costs.

当然还有运营成本，但每卖出100美元就能赚74美元的毛利润。

Obviously, there's operational costs as well, but $74 gross profit per 100 that they sell.

在即将发布的财报中，投资者应该关注这一点吗？

Is this something that investors should watch in the upcoming report?

是的。

Yeah.

当然。

For sure.

这绝对是我会持续关注的事情。

And it's something I'll definitely be keeping an eye on.

这个利润率，不仅仅是因为英伟达在过去十年里规模大了很多。

The the margins, it's not just because Nvidia got a lot bigger over the past ten years.

那肯定是其中的一部分原因。

That's definitely part of it.

你知道，公司随着规模扩大，效率会越来越高。

Like, you know, companies get more efficient as they scale.

很大一部分是因为新建的大型数据中心。

A lot of it is because of the new big data center build out.

英伟达有很大的定价权。

Nvidia has a lot of pricing power.

他们想收多少钱就收多少钱。

They can charge whatever they want.

他们数据中心芯片未来几年的产能基本上都卖光了。

They're essentially sold out of chips for data centers for the next couple of years.

所以现在，他们想怎么定价都行。

So right now, they could charge whatever they want.

所以我会密切关注这个，因为它是定价能力的一个重要指标。

So I'm gonna be really watching that because it's a great indicator of pricing power.

而且我特别感兴趣，因为AMD刚刚推出了他们首个面向数据中心的全规模机架系统。

And I'm especially interested because, AMD just rolled out its first full scale rack system for data centers.

所以竞争正在加剧，争夺那95%的市场份额。

So competition is heating up that 95% market share.

AMD正试图从中分一杯羹。

AMD is trying to take some of it.

所以，你知道，利润率将是判断他们是否成功的一个好指标。

So, you know, the the margins are gonna be a good indicator of whether or not they're successful.

所以Rachel，我们请你来谈谈，因为显然毛利率高是好事，但这也可能开始下降。

So Rachel, let's bring you in here because obviously higher gross margin, good, and that could start to come down feasibly.

假设需求没那么大，或者竞争开始加剧，毛利率会降到什么水平，你才会开始担心市场的竞争性？

Let's say that there's just not as much demand or if competition starts coming in, what would be a level of gross margin that it comes down to that you would start to be concerned about the competitive nature of the market?

对。

Yeah.

嗯，首先我想多聊一下是什么在推动这些毛利率。

Well, first I wanna talk a little bit more about what's driving these gross margins.

我是说，Matt 刚才简单提了一下，但要理解，你知道，英伟达是怎么做到大约75%的毛利率的？

I mean, Matt hit on it briefly, but to understand, you know, how is Nvidia commanding these roughly 75% gross margins?

你真的得更多看看我们现在看到的这种供需失衡和高性能计算的情况。

You really have to look more at that supply demand imbalance and high end computing that we're seeing right now.

所以现在，超大规模云服务商，对吧，微软、亚马逊、Alphabet，他们订购这些下一代芯片的速度，比英伟达的制造合作伙伴台积电实际生产的速度还要快。

So right now, the hyperscalers, right, Microsoft, Amazon, Alphabet, they are ordering these next generation chips faster than Nvidia's manufacturing partner, TSMC, can actually produce them.

而且，你知道，这是一个典型的场景，当需求远远超过供应时，你基本上就拥有了完全的定价权。

And, you know, it's a classic scenario of when demand heavily outstrips supply, you have essentially total pricing power.

英伟达可以把这些上涨的投入成本，比如来自SK海力士等供应商的高带宽内存价格飙升，转嫁出去。

Nvidia can pass these rising input costs like the surging prices of high bandwidth memory, from suppliers like SK Hynix.

他们可以把这些成本直接转嫁给客户，而不会影响订单量。

They can pass these costs right under their customers without hurting order volumes.

但同样重要的是，要知道，他们有数百万开发者被锁定在他们专有的CUDA软件生态系统中，而且，为其他任何东西构建或优化AI模型都需要数月工程时间。

But it's also important to note, you know, they have millions of developers locked into their proprietary CUDA software ecosystem and, you know, building or optimizing an AI model for anything else takes months of engineering work.

真正缺乏的是能直接使用的可行替代方案，

There's a real lack of viable alternatives that work out of

开箱即用。

the box.

所以很多

And so a lot of

科技巨头宁愿支付英伟达的高价，也不愿冒险在AI竞赛中落后。

the tech giants choose to pay Nvidia's premium prices rather than to risk falling behind in the AI race.

他们从英伟达购买的，实际上是一个完整的生态系统，而不仅仅是芯片。

And they're buying in so doing from Nvidia really what's an entire ecosystem, not just the silicon.

对我来说，如果毛利率开始下降到70%的底线附近，我可能会开始有点紧张，或者至少质疑幕后发生了什么。

Now for me, where I would maybe start getting a little bit nervous or at least questioning what's happening behind the scenes is if gross margins were starting to fall down towards that 70% floor.

这或许能让我们了解一些实际情况。

It kind of might tell us a bit of a story about what's happening on the ground.

这可能表明供应已经赶上或超过了市场需求，也可能意味着那些更便宜的竞争架构，比如AMD的MI 300系列，或者超大规模数据中心内部的自研芯片——这也是需要考虑的另一部分——可能已经实现了某种软件兼容性，从而绕过了那道护城河。

It could indicate that supply would have caught up with or exceeded market demand, could meet the also mean that some of those cheaper competitive architectures, like AMD's MI 300 series or or hyperscalers internal custom chips, which is another piece, as well to consider might have achieved some software compatibility that bypasses that that moat.

不过我觉得我们离那种情况还远着呢。

Now I do not think that we are anywhere close to that reality.

而且我也要说明，我不认为这是赢家通吃的局面。

I also don't think to be clear that this is a winner takes all scenario.

但随着我们深入AI竞赛和AI革命，这些确实是值得关注的点。

But those are some things to watch as we get deeper into the AI race and AI revolution.

显然，在这个播客里，我们不会对所有公司做财报预览。

Obviously, on this podcast, we don't do an earnings preview for all the companies.

我们今天之所以讨论英伟达，并不完全是为了英伟达本身。

And the reason that we're doing it for Nvidia today is not for so much completely Nvidia's sake.

显然，我们是在谈英伟达的股票，但在我看来，英伟达如果有什么好消息，那对股市肯定会有经济上的连锁反应，反过来，如果英伟达出了什么坏事，我们也会看到相应的连锁反应。

Obviously, we're talking about Nvidia stock, but there is a, in my opinion, a 0% chance that something good would happen for Nvidia that wouldn't have economic ripples throughout the stock market, or conversely, something bad would happen with Nvidia and there we wouldn't see the ripple effects from that as well.

所以，我想问你们的是，面对这份财报，第一，你们个人持有英伟达的股票吗？

And so I I guess here my question to you guys is, going into this earnings report, one, do you own Nvidia stock personally?

那你会关注市场上哪些相关的领域，来看这种连锁反应？

And what are the kind of the connected places in the market that you'll be watching for that ripple?

好，先回答你的第一个问题，乔恩，没有。

Well, to answer your first question, Jon, no.

我目前没有持有英伟达的股票，在财报发布前。

I don't currently own Nvidia stock heading into earnings.

部分原因是估值问题。

Part of it's the valuation.

我现在就只想在场外看着。

I'm right now just happy to watch this one from the sidelines.

我觉得，你知道，管理层任何能引发整个行业连锁反应的声明。

I think, you know, any statement from management that would trigger sort of an effect across the entire sector.

我觉得不会是像制造延迟或者轻微利润率这种问题，实际上，说清楚点，回到我们之前的对话。

I don't think it would be about things like a manufacturing delay or even a slight margin this actually to be clear, to go back to our prior conversation.

我觉得可能是黄仁勋的任何评论，暗示超大规模数据中心的资本支出有减速或趋于平稳的迹象。

I think it would be maybe, any commentary from Jensen Huang that would indicate some type deceleration or plateauing in hyperscaler CapEx.

不过短期内我们不太可能看到这种情况发生。

Now we're not looking at a scenario anytime soon where this is, you know, likely to happen.

我觉得我们看到的是，超大规模企业的资产负债表非常强劲。

I I think what we are seeing is the hyperscaler balance sheets are very strong.

我们预计明年AI资本支出还会再增长两位数。

We're looking for another double digit increase in AI CapEx spending heading into next year.

说实话，科技巨头之间在基础设施建设上的竞争压力实在太大了。

And, honestly, the competitive pressure among the tech giants to build out infrastructure is so intense.

他们真的不能停下来或者放慢建设速度。

They can't really afford to blank or to slow down the build out.

我们看到所有大型科技公司的管理层都说，在AI基础设施上投资不足的风险远远大于过度建设的风险。

We've seen all the big tech management teams saying that the risk of underinvesting in AI infrastructure vastly outweighs the risk of overbuilding.

还有一点也很重要，微软、Alphabet、Meta这些公司，它们的核心广告和云业务带来了巨大的现金流。

And I think it's also important to note, you know, the Microsoft, Alphabet, Meta, these are companies that are generating tremendous cash flow from their core advertising and cloud businesses.

它们把这些钱又投入到数据中心，以巩固市场份额。

They're putting that back into data centers to secure market shares.

云服务商手里积压了大量企业客户的订单，他们也在等着算力资源，所以短期内的需求管道还是满的。

The cloud providers are seeing huge backlogs of enterprise customers also waiting for compute capacity so that near term demand pipeline remains filled.

所以如果黄仁勋说建设要放缓，那会是关键信号，但我觉得短期内不太可能看到这种情况。

So any commentary from Jensen Huang about a slowdown in the build out would be key, but I don't think we're gonna be seeing that anytime soon.

对。

Yeah.

我没有持有英伟达的股票。

I don't own Nvidia.

至少不是直接持有。

At I mean, at least not directly.

通过ETF间接持有的话，我算过。

By ETF ownership, I I've calculated it.

英伟达间接占我整个投资组合的差不多3%。

Nvidia is something like 3% of my total portfolio just indirectly.

但我会密切关注财报，因为英伟达的表现会波及很多其他公司，不只是那些超大规模云服务商，那肯定也是其中一部分。

But I'll be watching the results really closely because Nvidia's performance can have ripple effects on so many other companies and not just the hyperscalers, which that's definitely part of it.

我是说，英伟达的业绩数字、未来的指引，还有他们给出的评论，都能让我们感受到AI建设推进的速度。

I mean, Nvidia's numbers, their future guidance, the commentary they give, it gives a sense of the pace of the AI build out.

这对像思科和Arista Networks这样的网络公司影响很大，比如说，当这些英伟达芯片安装并需要连接起来时，它们就是直接的受益者。

That could have a big implications for, you know, networking companies like Cisco and Arista Networks, for example, that, you know, are are direct winners when these Nvidia chips are installed and have to be linked together.

英伟达对未来需求的表态，也会影响到像应用材料这样的公司，它们制造半导体生产所需的设备。

Nvidia's tone on future demand, it affects how companies like, say, Applied Materials, which builds the equipment that semiconductors are made with.

所以它们可以据此预测未来的需求，投资者也可以根据英伟达的动向预测未来的需求。

So they can kind of forecast future demand, and investors investors can forecast future demand based on the what Nvidia is doing.

这只是几个例子。

So those are just a couple examples.

我是说，我们可以用一整集来讨论所有以某种方式受英伟达影响的公司，而且我甚至不确定那样够不够。

I mean, we could spend a whole episode on all the companies that are affected by Nvidia in one way or another, and I don't even know if that would be enough.

但在英伟达发布公告之后，我们会看到很多连锁反应。

But there there there are a lot of ripple effects that we're gonna see in the in the wake of Nvidia's announcement.

英伟达可能甚至不是在其公告发布时股价波动最大的股票。

Nvidia might not even be the biggest stock to move or the stock movement, on on its announcement.

对。

Yeah.

当我们谈论的是一家市值五万亿美元的公司时，你最好相信市场肯定会有一些波动。

When we are talking about a $5,000,000,000,000 company, you better believe that there will be some movement around the market.

所以，谢谢大家分享你们的看法。

And so thank you all for sharing your thoughts on that.

但我们还没结束，因为休息之后，我们还会回答一个关于数据中心的听众来信问题。

But we are not done because after the break, we're gonna take a mailbag question regarding data centers.

你正在收听的是《Motley Fool 隐藏宝石投资》。

You're listening to Motley Fool Hidden Gems Investing.

那些看起来比别人完成更多事情的人，并不是工作时间更长或者喝更多咖啡。

The people who seem to get more done than everyone else, they're not working longer hours or running on more caffeine.

他们只是不再浪费时间在那些不能推动工作进展的事情上。

They've just stopped wasting time on the stuff that doesn't move work forward.

切换应用、重新解释背景、寻找文件，这些都不是小效率问题。

Switching apps, re explaining context, hunting for files, those aren't small inefficiencies.

这些时间每周都白白浪费掉了。

They're hours wasted every week.

Superhuman Go 能帮你把这些时间找回来。

Superhuman Go gives you those hours back.

作为 Grammarly 的出品方，Go 是一款 AI 聊天工具，它嵌入在你已经使用的每个标签页和工具中，随时待命，帮你处理手头的工作。

From the makers of Grammarly, Go is an AI chat that sits inside every tab and tool you already use, always available and ready to help you with what you're working on.

你可以让它起草内容、总结长对话、调出文件，或者帮你为会议做准备。

Ask it to draft something, summarize a long thread, pull up a file, or prep you for a meeting.

Go 都能搞定，而你完全不需要离开当前页面。

Go handles it without you ever leaving the page you're on.

这就是 AI 真正融入工作，而不是增加负担的样子。

This is what it looks like when AI actually fits into your work instead of adding to it.

就像有一个队友，唯一的任务就是帮你把工作做得更好。

It's like having a teammate whose only job is to help you be better at yours.

Go 跟上你的节奏，让你继续前进。

Go keeps up so you can move forward.

有了Go的协助，你就能尽情展现你的强项。

And with Go working with you, you can show off what you do best.

去superhuman.com看看Superhuman Go能做什么。

See what Superhuman Go can do at superhuman.com.

网址是superhuman.com。

That's superhuman.com.

欢迎回到Motley Fool Hidden Gems Investing节目。

Welcome back to Motley Fool Hidden Gems Investing.

我们喜欢从听众来信中挑选问题，我现在就来读一封。

We love to take questions from our mailbag, and I'm gonna go ahead and read this one.

这封信来自哥伦比亚波哥大的一位听众。

This comes from a listener in Bogota, Colombia.

非常感谢你收听我们的节目。

So thank you for listening to our show.

这封信的核心观点是，七十年代时，一台电脑要占满整个房间，而现在我们可以把它装进口袋随身携带。

Basically, the premise of the question is pointing out that back in the seventies, a computer used to fill a room, and now we can carry it around in our pockets.

所以计算技术一直是在用更少的资源做更多的事情。

So computing has a history of doing more with less.

那么问题来了。

So here's the question.

如果数据中心也走同样的路线，这些支出会变成什么样？

What happens to all this spending if data centers follow the same path?

如果芯片和冷却技术变得足够高效，同样的工作量所需的物理基础设施大大减少，那么今天的建设规模最终会不会显得过度？

If chips and cooling get efficient enough that the same workloads need far less physical infrastructure, does today's build out end up looking overbuilt?

还是说需求的增长速度足以消化掉任何效率提升带来的影响？

Or does demand grow fast enough to absorb whatever efficiency gains show up?

很想知道你们的想法。

Would love to hear your thoughts.

谢谢。

Thanks.

这是来自尼科的提问。

And that's from Nico.

Matt，我想让你先回答这个问题。

Matt, I want to I want you to answer this question first.

所以本质上，问题是，过去计算机变得更高效了。

So essentially, the question is, in the past, computers became more efficient.

我们可以用更少的资源做更多的事。

We could do more with less.

因此，有理由认为未来有了AI，我们也能用更少的资源做更多的事。

Therefore, it's reasonable to assume with in the future with AI, we can do more with less.

所以问题是，如果是这样的话，我们是不是在建造太多物理基础设施了？

And so the question is, are we building way too much physical infrastructure if that's the case?

你这里有一个有趣的观察，关于两种过度建设。

So you have an interesting observation here about two kinds of overbuilding.

一种呢，是供需暂时失衡，另一种呢，是需求完全消失。

On one, you kind of have supply and demand temporarily out of balance, and the other, you have an evaporation of demand completely.

跟我们说说你的想法吧。

Just walk us through what you're thinking.

对。

Yeah.

所以我对商业房地产行业非常了解，这其实不是技术问题。

So I know the commercial real estate industry very well, and that's really what this is not a technical question.

这是个房地产问题。

This is a real estate question.

在房地产行业，你会看到两种过度建设。

There there are two kinds of overbuilding you see in real estate.

先说第一种，几年前，疫情期间自助仓储的需求激增。

So for just the first one, a few years ago, self storage had a surge of demand during the pandemic.

大家都想清理家里的空间，因为被困在家里。

Everyone wanted to go declutter their space because they were stuck in their homes.

到2023年，市场供应过剩。

And by 2023, markets had too many too much supplies.

你知道，这些设施很容易建。

You know, these were very easy to build.

大多数情况下，它们不过是预制建筑而已。

They're little more than prefab buildings in most cases.

但经过几年几乎零开发之后，市场开始趋于平衡。

And but after a couple years of little to no development, the market started to reach equilibrium.

我们现在也看到仓库领域正在发生同样的情况，因为电商需求确实被提前释放了。

We're also seeing that happen in the the warehouse space right now as ecommerce demand was really pulled forward.

另一种情况就是写字楼所经历的。

The other type is what happened with office space.

写字楼存在供应过剩的问题，原因是永久性的结构性变化。

Office space has oversupply issues because of a permanent structural change.

人们都在家办公。

People are working from home.

我们三个人录制这期节目时，也都是远程工作的。

The three of us are working remotely as we record this.

我的意思是，对办公空间的需求比以前少多了，而且这种情况不会达到平衡，除非办公楼被拆除或改作他用。

I mean, there's a lot less need for office space than there used to be, and that that's not gonna reach equilibrium until, you know, offices are demolished, turned into other things.

所以，我们到底会属于哪一种情况，这真是个值得思考的问题。

And so it it it's a great question of what basket we're gonna be in.

说到数据中心，我们迟早会看到一些过度建设的情况。

So when it comes to data centers, we are going to see some overbuilding at some point.

你知道，即使只是暂时的，比如开发激增，然后出现芯片短缺或者电力短缺之类的，总会有某个时刻出现供需不平衡。

You know, even if it's very temporary, like there's a surge in development and, you know, there's a chip shortage or a power shortage or whatever, there's gonna be some some supply demand imbalance at some point.

但真正的问题是，这些供需问题会不会只是暂时的，还是说会像办公楼那样，变成结构性的问题，比如芯片和冷却技术变得更高效，需要的空间更少？

But it's a really great question of whether these are gonna be temporary supply demand issues or a more structural, like, office type problem if chips and cooling do become more efficient and less space is needed?

嗯，这确实是个关键问题。

Well and and that's really the question.

对吧？

Right?

我们会属于哪一种情况呢？

Which basket do we fall in?

因为这两种答案带来的影响差别非常大。

Because the there are profound differences in the implications of those answers.

而且，你知道，我觉得一方面，可以说现在并没有失衡，因为我今天早上刚看到一些研究，说数据中心的空置率只有1%，而历史上通常接近5%。

And, you know, I I would say that on on the one hand, can make the argument that there is no imbalance right now because I just saw some research this morning saying that data center vacancy is only at 1%, whereas more historically, it's closer to 5%.

所以，现在即将上线的项目都已经售罄了。

So and are now already sold out with what is coming online.

所以看起来需求还是相当高的。

So it seems like demand is still pretty high.

但假设未来我们能用更少的资源做更多的事情，借助AI，我想问的是，我们是在根据今天的AI来建设物理基础设施，还是为未来的AI建设基础设施？

But assuming we can do more with less with AI in the future, I guess my question is, are we building the physical infrastructure based on AI today, or are we building the infrastructure for the AI of the future?

我觉得这里会出现类似自助仓储的情况。

I think we're gonna see a self storage situation unfold here.

所以，我的意思是，让我稍微展开一下。

So, I mean and and let me unpack that a bit.

从历史上看，技术效率越高，消费量就越大。

So historically, the more technologically efficient something gets, more consumption happens.

所以，问题在于预测芯片会变得更高效，并随时间推移占用更少的数据中心空间，这个预测很可能是准确的。

So the the question is predicting that chips will become far more efficient and take up less data center space over time, a prediction that's likely to be accurate.

但是，过度建设论的前提是，像Anthropic、OpenAI、谷歌这些需要数据中心空间的公司，只想用更少的空间完成同样的工作量，因为它们更高效。

But the overbuilt thesis assumes that companies that need data center space like the Anthropics, the OpenAI's, the Google's, just wanna do the same amount of work with less space because they're more efficient.

但更高效的算力会解锁以前不经济的应用场景。

But more efficient compute will unlock workloads that aren't economical before.

我的意思是，你觉得苹果的工厂面积是变大了还是变小了？从个人电脑占满整个房间，到现在能装进口袋。

I mean, do you think Apple's factory space has gotten bigger or smaller since PCs took up a whole room and now can fit in your pocket?

你觉得他们现在需要更多还是更少的工厂空间？

Do you think they need more or less factory space now?

我这么问，你肯定知道答案。

And the I I mean, the way I'm asking that, I'm sure you know the answer.

我相信未来一二十年也会发生同样的情况，虽然中间会有供需平衡的波动。

I'm sure the same thing is gonna apply here over the next decade or two, although we're gonna see supply demand balances along the way.

但我也在假设，而且这是个很大的假设，就是芯片和设备不仅会变得更高效，而且会变得更便宜，就像个人电脑那样。

But I'm I'm also assuming, and it's a pretty big assumption, that that chips and equipment will not only become far more efficient but will become cheaper based on the amount of compute, kinda like how PCs did over time.

现在数据中心的建设非常资本密集，说实话，这是对我刚才说的所有话最大的看空理由。

Now the data center build out right now is very capital intensive, and, honestly, that's the biggest bear case to everything I just said.

但是，Rachel，Matt刚才给我们讲的这些，其实有一个正式的术语。

But, Rachel, there is an official term for what Matt has just walked us through and just introduced to us.

这个术语是什么？它又是怎么运作的呢？

What is that official term, and how does it work?

对。

Yeah.

这其实把我们带回到了经济学里一个基础概念，叫做杰文斯悖论。

I mean, this is very much kind of bringing us back to this foundational concept in in economics known as the Jevon's Paradox.

在十九世纪，有个经济学家叫威廉·斯坦利·杰文斯，他观察到，当蒸汽机变得更省燃料时，煤炭消耗量并没有下降。

So in the nineteenth century, there was an economist named William Stanley Jevon, and he observed that when steam engines became more fuel efficient, coal consumption didn't actually decrease.

反而飙升了。

It skyrocketed.

因为蒸汽动力变得更便宜、更实用，全新的行业开始采用它，甚至围绕它发展起来。

And because steam power became cheaper and more practical, there were entirely new industries that adopted it and even formed from it.

我们现在在AI数据中心上看到的，正是这个现象的重演。

And we are seeing that play out with AI data centers right now.

当然，时代不同了，但在我看来，这个概念确实很有道理。

Now, obviously, it's a different time, but this is very much a concept that, in my view, rings true.

我的意思是，当我们看到这些公司想办法提高AI芯片或冷却系统的效率时，这会大幅降低单次AI计算或单个token的成本。

I mean, when we are seeing these companies find ways to make AI chips or cooling systems more efficient, it drastically lowers the cost of a single AI computation or token.

而更低的成本让AI在经济上变得可行，从而催生出一波新的应用，这些应用以前可能因为太贵而无法运行。

And lower costs make AI economically viable for a new wave of of applications that maybe used to be too expensive to run.

所以，比如说，如果推理成本大幅下降，一家公司就可以从偶尔使用AI，转变为在整个供应链中运行真正复杂、持续的AI代理。

So, you know, for instance, if inference costs drop significantly, a company can pivot from using AI occasionally to running really complex, continuous AI agents across their entire supply chain.

这只是一个简单的例子。

That's just a basic example.

所以，当你审视这个概念，并将其带入AI和数据中心时代时，你会发现，效率非但没有缩小物理足迹，反而可能成为需求的加速器。

So what you see when you look at this concept and you bring it forward into the AI and data center era is that rather than shrinking the physical footprint, efficiency can act as an accelerator for demand.

科技巨头们并不真的把效率提升看作是缩小数据中心规模的手段。

And tech giants aren't really looking at efficiency gains as a way to downsize their data centers.

他们将其视为从当前正在建设的基础设施中提取更多能力的方式。

They see it as a way to extract vastly more capability out of the infrastructure they're currently building.

而且我们看到对算力的需求非常旺盛，似乎完全没有放缓的迹象。

And we are seeing a huge appetite for compute that doesn't appear anywhere close to slowing down.

在我看来，我认为目前的建设不太可能导致产能过剩，因为需求的增长速度远快于硬件缩小的速度。

In my view, I think that the current build out is unlikely to result in overcapacity because demand is scaling at a pace that is much faster than hardware is shrinking.

当然，对杰文斯悖论的反驳观点是，尤其是当我们回顾历史例子时，当时有煤炭，但煤炭价格下降的同时，也有大量企业带着真实需求排队等着。

Of course, the counterargument to Jevon's Paradox is, and especially when we look at the historical example, you had coal, but you also had a ton of businesses lined up with real demand on the other side of that coal price coming down.

我认为这里对AI的反驳观点是，很多需求可能是被补贴支撑的，而这种补贴是有期限的。

And I think that kind of the counterargument here with AI is that a lot of the demand perhaps being subsidized, and that does have a finite lifespan.

最终，这些需求需要用现金流来支撑。

At some point, it will need to be financed with cash flows.

所以，需求真的存在吗？

And so is the demand really there?

我认为这个问题对这次讨论非常关键。

I think that's a question that is really pertinent to this discussion.

不过，如果非要选边站的话，我会说我们现在可能并没有过度建设太多，因为需求实在太高了，但这只是我的个人看法。

However, if I'm going to stake my claim on one side of this or the other, I would say we're probably not overbuilding by a whole lot right now because demand is so high, but that's just my take.

休息之后，我们会再次打开听众来信，聊一聊一些收购的事情。

Coming up after the break, we are going to dip back into the mailbag a second time, and we're gonna talk about some acquisitions.

你正在收听的是 Motley Fool 的隐藏宝石投资节目。

You're listening to Motley Fool hidden gems investing.

欢迎回到 Motley Fool 的隐藏宝石投资节目。

Welcome back to Motley Fool hidden gems investing.

我想提醒大家，这个播客确实有听众提问环节。

And I do want to point out that we do have listener questions on this podcast.

你可以发邮件到 podcast@fool.com。

You can email us at podcast@fool.com.

保持简短，保持愚蠢。

Keep it short, keep it foolish.

记住，我们不能提供个性化的投资建议。

And remember that we can't give personalized investing advice.

如果你能满足这三个要求，那就发邮件到 podcast@fool.com。

But if you can meet all those three requirements, then email us at podcast@fool.com.

我们会很乐意考虑你的问题，并可能在这期播客中讨论。

And we'd be happy to consider your question for this podcast.

今天我们还会再回答一个问题。

And we're going to take a second one today.

他做得非常好，问题简短又有点傻气。

And he's done such a good job at keeping it short and foolish here.

这是来自大卫的问题。

Here's the question from David.

我注意到，当收购或可能的收购被宣布时，被收购的公司，比如华纳兄弟探索或PayPal，股价会飙升。

I've noticed that when acquisitions or possible acquisitions are announced, the company being purchased, I.

也就是，

E.

华纳兄弟探索或PayPal，股价会飙升。

Warner Bros Discovery or PayPal, has a stock price surge.

对于持有这些公司的投资者来说，收购完成后有什么好处？如果收购被法院否决，又会发生什么？

What are the benefits for investors holding those companies post acquisition, and what happens if the acquisition is not allowed by the courts?

瑞秋。

Rachel.

对。

Yeah.

这里有几个非常好的问题。

So a a few really great questions here.

所以，收购后持有股票的好处，实际上取决于交易的结构。

So the benefits of holding a stock post acquisition, it really depends on how the deal was structured.

如果是全现金交易，交易完成后你实际上不会持有任何东西。

So if it was an all cash deal, you're not gonna actually hold anything once the transaction finishes.

你的股份会被清零。

Your your shares are wiped out.

它们会按照最终的收购价转换成现金。

They're converted into cash at that final buy up price.

如果是换股交易，你的股份实际上会变成新合并公司的股权。

Now if it's a stock for stock swap, your shares are actually turning to equity in the new combined company.

我们经常听到管理层在宣布收购时谈论实现协同效应。

Now we hear a lot about management when they announce an acquisition talk about the realization of synergies.

听起来像是一个非常时髦的流行词。

Sounds like a very nice fancy buzzword.

这是什么意思呢？

What does it mean?

其实，这个想法是，合并后的企业可以消除很多重复的公司开支。

Well, the idea is, you know, the combined businesses can eliminate a lot of the duplicate, corporate expenses.

他们可以合并销售团队。

They can merge their sales teams.

他们可以利用合并后的规模，可能获得更低的公司债务利率。

You know, they can use their combined size to get maybe much lower interest rates on corporate debt.

那么这对你作为投资者意味着什么呢？

So what does that mean for you as an investor?

你基本上是在押注这两家公司合并后的价值会远远超过它们分开时的价值。

Well, you're essentially betting that these two companies together will be worth far more than they ever were apart.

这对你的长期投资组合来说可能是个好处。

Maybe that's a benefit for your long term portfolio.

现在这里还有另一个问题。

Now there is another question here.

如果收购被法院阻止了，会发生什么？

What happens if the acquisition is blocked by the courts?

通常，目标公司的股票会回吐收购溢价。

Now, typically, the target company stock will give back its acquisition premium.

我们会看到股价下跌，但背后可能还有一些损害在发生。

We'll see declines, but there can be some damage that happens in the background.

你知道，你可能会看到一家公司陷入企业僵局。

You know, you can see a company that's stuck in corporate limbo.

管理层显然要应对一场可能旷日持久的法律战。

Management is obviously dealing with a potentially protracted legal battle.

这可能会成为竞争对手利用这个不确定窗口趁机介入的领域。

This can be an area where competitors will use that window of uncertainty to sort of swoop in.

其实，这样的例子有很多。

There's actually, you know, a lot of examples of this.

比如，几年前Adobe试图以200亿美元收购设计平台Figma。

One would be, when Adobe tried to buy the design platform Figma a few years back for a $20,000,000,000 price tag.

显然，这笔交易最终没有成功。

That was obviously a deal that did not come to fruition in the end.

我记得这件事拖了大概十五个月。

There was I think it dragged on about fifteen months.

在最终取消之前，受到了严格的监管审查。

There was heavy regulatory scrutiny before it was ultimately called off.

而且，Figma当然继续运营日常业务，但他们受到标准合并条款的法律约束，这些条款限制他们进行重大的独立变革或融资操作，导致产品发布放缓。

And, you know, Figma, of course, kept running its day to day business, but they were legally bound by the standard merger covenants that restricted them from executing major independent shifts or financing moves, slows down product launches.

当交易失败后，Figma利用Adobe支付的10亿美元分手费，重新积极扩张。

When the deal collapsed, Figma used the $1,000,000,000 cash breakup fee from Adobe to aggressively, grow again.

但当然，那对他们来说是一段艰难的时期，而Adobe也不得不支付那十亿美元。

But of course, that was a major period of friction for them, and Adobe, of course, had to pay out a billion dollars.

这只是一个例子。

That's that's just one example.

对。

Yeah.

我是说，直接回答这个问题的第一部分，是的。

I mean, to directly answer the first part of that question, yes.

目标公司通常在交易后股价会飙升，因为收购方几乎总是要支付溢价，才能让公司董事会和股东同意被收购。

The the target generally spikes after the deal because the acquirer almost always has to pay a premium in order to get the company's board and shareholders to say yes to a takeover.

我是说，如果你的股票交易价格是100美元，然后Adobe突然出现说，我们愿意以每股100美元收购整个公司，那你就没什么动力了。

I mean, there's not much motivation if your stock's trading for a $100, and then, you know, Adobe comes in and swoops in and says, well, we'll give you a $100 a share for the entire company.

你懂吧？

You know?

为什么？

Why?

你为什么要这么做？

Why would you do that?

所以，是的，这取决于它是全现金交易，还是现金加股票的交易。

So, yeah, it it depends on if it's an all cash deal, if it's a cash and stock deal.

这才是你真正需要做决定的地方。

That's really what you have to where you have a decision to make.

如果是现金交易，通常会有我所说的监管差距，就是股价最初跳涨的价格和收购价之间的差距。

When it's a cash deal, you generally have what I call a regulatory, you know, gap between what the stock price immediate initially jumps to and what the the acquisition price is.

一旦你跨过这个监管门槛，也就是这笔交易能否获批，你就会看到这个差距开始真正缩小，股价会逐渐向交易的现金价格靠拢。

And once you get over that regulatory hump of will this deal be approved, that's when you'll see that gap really start to close, and it'll really gravitate toward the cash price of the deal.

如果是现金加股票的交易，就像瑞秋提到的，你之后会接触到合并后的公司。

With a cash and stock deal, as Rachel kinda mentioned, you have to you'll you'll have exposure to the combined company after.

通常收购方规模更大，所以你真的需要决定是否想在股票交易后持有收购方的股票。

Usually, the acquirer is bigger, so you really need to decide if you wanna own the acquirer after the stock.

有时候对我来说，答案是肯定的。

Sometimes for me, this answer has been yes.

比如，当Rocket公司收购Redfin时，我就是Redfin的股东。

Like, for when Rocket companies acquired Redfin, I was a Redfin shareholder.

现在我是Rocket的股东，因为我喜欢他们的业务。

Now I'm a Rocket shareholder because I like their business.

其他时候，答案是否定的。

At other times, it's been no.

我想强调一下Rachel在Figma和Adobe交易最后提到的内容。

And I wanted to kind of, you know, emphasize something Rachel mentioned at the end with the Figma and Adobe deal.

很多这类交易都有分手费。

A lot of these deals have breakup fees.

有时候如果出现竞购战，比如像问题里提到的华纳兄弟探索公司，你会看到一笔相当可观的分手费，这有点像交易的甜头。

And sometimes if there's like a bidding war happening, like with Warner Bros Discovery, like the question mentioned, you'll see a pretty hefty breakup fee, which kind of is like a a deal sweetener.

比如，如果派拉蒙和华纳兄弟的交易失败，分手费高达70亿美元。

Like, there's a $7,000,000,000 breakup fee if if the Paramount Warner Bros deal falls through.

所以有些交易内置了相当大的安全网，所以在这种情况下，我不会预期华纳兄弟的股价会完全回落到公告前的水平，因为有这笔费用。

So some deals have pretty big safety nets baked in, so they might in that case, I wouldn't expect Warner Bros to fall, like, all the way back to its, you know, preannouncement price because of that fee.

但这完全取决于具体交易，而且确实值得了解。

But that's very deal by deal, and it's really worth knowing.

对。

Yeah.

我觉得沃伦·巴菲特有一次给过一条建议，就是如果你要考虑这些股票，一定要问自己：如果交易告吹，我的股票会怎么样。

I think that one of the pieces of advice that Warren Buffett gave out one time was with these things, if you're going to consider these stocks, always ask yourself what happens to my stock if the deal falls through.

我可以亲身作证，亚马逊宣布要收购iRobot的时候我买了它的股票，结果那笔投资在我账户里变成了零。

I can personally attest to buying iRobot when Amazon announced it was going to acquire it, that wound me up with a zero in my portfolio for that.

所以一定要搞清楚这笔交易成功的可能性有多大，以及如果失败会有什么后果。

So make sure you know how likely the deal is to go through and what happens if it doesn't.

我当时没有充分评估这些风险。

And I didn't fully assess those risks at the time.

所以谢谢你们两位提出这一点，并指出了其中的差异。

So thank you to both of you for bringing that to the table and pointing out the differences here.

和往常一样，节目嘉宾可能对他们谈论的股票有兴趣，而The Motley Fool也可能有正式的推荐或反对意见。

As always, people on the program may have interest in the stocks they talk about, and The Motley Fool may have formal recommendations for or against.

所以不要仅仅根据你听到的内容来买卖股票。

So don't buy or sell stocks based solely on what you hear.

所有个人理财内容都遵循Motley Fool的编辑标准，并未经过广告商的批准。

All personal finance content follows Motley Fool editorial standards and is not approved by advertisers.

广告是赞助内容，仅供参考。

Advertisements are sponsored content and provided for informational purposes only.

要查看完整的广告披露，请查看我们的节目说明。

To see our full advertising disclosure, please check out our show notes.

感谢我们的制作人Dan Boyd，以及幕后Motley Fool团队的其他成员。

Thanks to our producer, Dan Boyd, and the rest of The Motley Fool team behind the glass.

代表Matt、Rachel和我自己，非常感谢大家今天收听我们的节目，我们下期再见。

For Matt, Rachel, and myself, thank you so much for listening to our show today, and we will see you again in the next episode.