Coreweave：AI泡沫的代言人还是下一个科技巨头？——对话Michael Intrator与Brian Venturo

人工智能是泡沫还是我们这一代人最大的机遇？

Is AI a bubble or the biggest boom of our lifetimes?

一家名为CoreWeave的公司的命运，或许能告诉我们一切我们需要知道的信息。

The fate of one company, CoreWeave, may tell us everything we need to know.

我们将在稍后回到公司创始人那里。

We'll be back with the company's founders right after this.

欢迎收听《大型科技》播客，这是一档致力于对科技世界及其更广泛领域进行冷静而细致对话的节目。

Welcome to Big Technology Podcast, a show for cool headed and nuanced conversation of the tech world and beyond.

今天我们为大家准备了一场精彩的节目，因为与我们同在演播室的是CoreWeave的创始人。

We have a great show for you today because in studio with us are the founders of CoreWeave.

CoreWeave的首席执行官迈克尔·恩特拉格与我们同在。

CoreWeave's CEO, Michael Entrager, is here with us.

迈克尔，欢迎你。

Michael, welcome.

非常感谢。

Thank you very much.

很高兴来到这里。

Great to be here.

还有CoreWeave的首席战略官布莱恩·文图拉也在这里。

And Coraweave's chief strategy officer, Brian Ventura, is also here.

布莱恩，很高兴见到你。

Brian, great to see you.

你们两人正在运营AI热潮中最引人注目的公司之一。

You are you both are running one of the most fascinating companies in the AI boom.

每个人都在用你们作为试金石，来验证自己对这一AI时刻的信念或不安。

Everyone has used you effectively as a Roshar test to read in their beliefs or insecurities about what's gonna happen in this AI moment.

有些人认为你们是AI泡沫的典型代表。

Some people think that you're the poster child for the AI bubble.

另一些人则认为你们正完美地把握住机遇，应对需求激增带来的建设热潮。

Others think that you're perfectly positioned to take advantage of the boom in building that is occurring as demand goes through the roof.

关于你们，有几个数据。

A couple stats about you.

截至今天，该公司在今年早些时候完成IPO后，估值已达420亿美元。

As of today, the company is worth $42,000,000,000 after an IPO earlier this year.

仅在第三季度，你们就在美国建成了八个新的数据中心。

You've built eight new data centers across The US in the third quarter alone.

最新的数据显示，你们拥有大约25万台NVIDIA的GPU，这些芯片是公司用来运行、扩展或训练AI模型的关键硬件。

And the latest reported numbers have you in possession of something like 250,000 of NVIDIA's GPUs, which are the chips that companies use to run AI models and grow them or train them, as they like to say.

让我们先从这一点说起，因为过去几年对你们来说真是一段非凡的旅程。

Let's just start off with this because it's been heck of a ride for you over the past couple years.

作为这场AI建设浪潮的前线，你们的感受如何？

What has it been like being on the front lines of this AI build out?

请稍微谈谈，帮助大家感受一下这场爆发的速度，以及在一个季度内建成八个数据中心究竟需要付出多少努力。

Talk a little bit, help people feel it, the speed at which it's boomed and what it's taken to do something like build eight data centers in a quarter.

这让人精疲力尽。

It's exhausting.

好吧，那我们就从这一点开始。

All right, so let's start with that.

这真的让人筋疲力尽。

It's been exhausting.

是啊，你说到点子上了，对吧？

Yeah, it's you you hit it that out, right?

这真的非常令人兴奋。

Like it has been incredibly exciting.

这真是难以置信的一年。

It has been an unbelievable year.

我的意思是，我们八个月前才上市，但感觉已经过了两辈子。

I mean, we IPO'd really eight months ago and it feels like it's been two lifetimes.

公司正在以惊人的速度发展。

The company is moving at incredible speed.

我们正在建设全球人工智能基础设施的很大一部分，这些基础设施是让人工智能成为今天样子所必需的。

We are building a massive percentage of the global AI infrastructure that's required to allow artificial intelligence to be what it is.

当我说到‘很大一部分’时，意思是，你知道的，一个有意义的比例。

And when I say massive, it's, you know, like a meaningful percentage.

你对这个百分比有什么估计？

What's your estimate about the percentage?

哦，这很难说。

Oh, that's tough.

你知道，你看

You know, look

很多。

A lot.

很多。

A lot.

我们并不认为自己只是提供足够的计算能力，而是有能力参与关于人工智能如何构建以及未来如何运行的讨论。

You know, we don't We think of ourselves as providing enough of the compute that we have the ability to be relevant in the debate of how AI is going to be built and how it's going to run into the future.

所以我们也不知道具体数字，你知道，有很多不同的技术提供商，真的很难准确把握数据，但可以说是具有重要意义的，对吧？

And so, we don't know what the numbers are, know, there's lots of different providers of technology that are being used, there's no real good way to kind of put your fingers on the data, but, you know, meaningful, right?

这是一个令人兴奋的处境，老实说，我们公司里经常谈论这个话题，能够每天投入精力和创造力，去构建这种智能的组成部分，是一种荣幸，因为这在很多方面是我们这个时代的核心问题。

And that's an exciting place to be, and it's honestly, we talk about this in the company all the time, it's a privilege to come into work and focus your energy, your creativity every day on building a component of this intelligence, which is the issue of our time in many ways.

我们每天都能真正面对这些问题，这很棒。

And we get to really sit there every day and pit ourselves against those issues, which is great.

我的意思是，我对这个充满热情。

I mean, I have a ball of it.

我正在尝试这个。

I'm taking a shot at this.

等一下。

Hold on.

在我们继续之前，你觉得这其实主要涉及实际层面，对吧？

Before we move on, think that that's really around, let's call it the practical side of it, right?

当一家公司像我们这样快速增长时——三年前我们可能只有100名员工，现在已经有大约2500名员工——这其中也伴随着情感层面的问题，对吧？

And when you're a company growing as fast as we have, where we had maybe a 100 employees three years ago, now we have 2,500 employees or so, there's an emotional side of this too, right?

而且，你知道，自从上市以来，我们一直处在世界的聚光灯下，人们都在问：他们在做什么？

And, you know, sometimes, since the IPO, we've been under this spotlight in the world of like, what are they doing?

他们是怎么做到的？

How are they doing it?

他们在执行吗？

Are they executing?

他们是这样吗？

Are they this?

而且，我们内部总是为自己设定最高的标准，那就是我们能多快完成某件事？

And, know, internally, we always set the highest bar for how how fast can we do something?

我们能做到多高的质量？

How high of a quality can we do it at?

而且，你知道，随着这个行业如此迅速地扩张，总会发生一些事情，对吧？

And, you know, as this industry has expanded so rapidly, like there are things that happen, right?

你知道，天气会影响项目的施工。

And you know, you have weather that impacts construction on a project.

有一辆卡车撞上了桥梁。

You have a truck that hits a bridge.

就像供应链中会遇到各种随机的外部或独特事件，然后这些问题反馈到我们身上，世界就会说：你们失败了，对吧？

Like you have all of these random exogenous or idiosyncratic things that happen in a supply chain and then it comes back to us and it's like the world is like, you failed, Right?

从文化角度来看，对公司内部而言，管理这一点一直至关重要。

And inside the company from a culture perspective, it's been so important for us to manage.

听着，我们正在以前所未有的规模和速度做着一些事情。

Like, listen, we're doing something at a scale no one has no one's ever done before at a speed no one's ever seen before.

当然，总会出问题的。

Of course, are gonna go wrong.

但要保持大局观，看看我们已经完成了多少。

But take perspective, like, see how much we've done.

对吧？

Right?

对于我们的员工来说，如果你以每小时一百万英里的速度前进，遇到一个减速带，也没关系。

And for our employees, it's if you're moving at a million miles an hour and you hit a speed bump, it's okay.

对吧？

Right?

这并不会改变你正在做的事情的轨迹。

It doesn't change the trajectory of what you're doing.

它只是留下了战斗的伤痕，这样下次就不会再发生同样的问题。

It just like it just provides the battle scars so it doesn't happen next time.

是的。

Yeah.

对。

Right.

不，我能想象，要为如此苛刻的客户部署如此重要的技术，同时还要以疯狂的速度推进，这一定是个充满挑战的世界。

No, I can imagine it's a rough and tumble world trying to build this with very demanding customers, very important technology that you're deploying, and the speed is crazy.

我的意思是，看看你们的创业故事，你们最初确实是在为加密货币提供基础设施。

I mean, is interesting looking at your founding story, you really started working on providing infrastructure for crypto.

是像以太坊挖矿那样的东西吗？

Was it like Ethereum mining or something like that?

然后你们非常聪明地转向了人工智能这一机遇，并与英伟达建立了合作关系，我们稍后会谈到这一点。

And then pivoted in a very smart way to this AI moment establishing a relationship with NVIDIA that we'll talk about that.

这对你们，可能对英伟达来说，都证明是非常有益和有帮助的。

That's proven to be very useful and helpful for you and probably for NVIDIA as well.

而现在你们再次加速建设数据中心，据我理解，这些数据中心的容量主要通过许可或租赁的方式提供给科技巨头。

And now you're again hyperdrive building data centers and the data centers are, if I have it right, largely licensed or the capacity is rented out mostly the tech giants.

我的意思是，核心客户是微软，根据你们的公开文件，大约三分之二的需求来自微软，但也有其他客户。

I mean, the core customer is Microsoft, something like two thirds of the demand according to your public filings is Microsoft, but there are others as well.

实际上，我们在上一次财报中讨论过客户集中度的问题，目前没有任何一家客户占我们未完成订单的30%以上。

So we actually spoke to company customer concentration in our last earnings, we can kind of there's no customer that represents more than 30% of our backlog.

因此，我们做得非常出色。

And so, we've done an incredible job.

这一直是公司的重点，从销售到建设周期的每一个环节，我们都致力于扩大我们的解决方案在人工智能领域的覆盖范围。

It's been a focus of the company, everything from sales all the way through build cycle, to really begin to broaden the reach with which our solution touches artificial intelligence.

微软是一个重要的客户，也是人工智能生态系统中强大且信誉卓著的组成部分，但我们也成功地吸引了许多其他优秀的客户，他们将在构建产品并推向市场时继续使用我们的解决方案。

Microsoft is an important customer and a large credit worthy and formidable part of the AI ecosystem at large, but they are, you know, we've done a really good job bringing on other wonderful clients, wonderful customers that are going to continue to kind of use our solution as they build their products and deliver them to market.

好的。

Okay.

我确实想深入聊聊客户集中度的问题。

And I definitely want to get into customer concentration a little bit.

但这为我们将要讨论的内容提供了一个很好的铺垫，而且有些新信息对我来说是第一次听到。

So but that's a good preface to what we'll touch on and already some new data to me.

听到这些真好。

So good to hear that.

但我想再次深入了解建造这些数据中心需要什么。

But I wanted to, again, like just get into what it takes to build these things, these data centers.

你们以惊人的速度组装它们。

You're assembling them with incredible speed.

所以我只想听听实地操作中，建造这些数据中心需要哪些条件？

So I just wanna hear a little bit about like on the ground, what does it take to put together these data centers?

历史上，

So historically,

比如说两年前，我们能够出去购买或租赁已经进入开发后期阶段的容量。

you know, let's say two years ago, we were able to go out and buy capacity or lease capacity that was much further through the development cycle.

对吧？

Right?

它们基本上已经建好了外壳。

They were basically the shell already existed.

这只是一个内部装修施工过程，意味着要进去安装冷却基础设施、机柜、所有电缆的布线系统，以及我们这些设施中数百英里的电缆。

It was a fit out construction process, which means going in and installing like the last last pieces of the cooling infrastructure, cabinets, conveyance for all the cabling, all the hundreds of miles of cabling we have in these things.

但过去一年发生了变化，我们现在更多地进行定制化的内部设计，对吧？

But it shifted over the past year, is that now we're doing much more bespoke in house design, right?

以确保我们满足客户部署的需求，对吧？

To make sure that we're meeting the needs of what our customers' deployment is gonna be, right?

所以现在一切都从如何设计冷却和电力分配开始，

So it's everything now from, okay, how is the cooling and electrical distribution designed?

我们如何确保电力的冗余和可靠性？

How are we ensuring electrical redundancy and reliability?

你知道，我们如何为这些设备的风冷部分降温？

You know, how are we cooling the air cooled side of these things?

因为虽然有液冷，但仍有一部分需要通过空气冷却。

Because you have liquid cooling, there's still a component of it that has to be cooled with air.

我们能暂停一下吗？

Can we pause on that?

当然。

Sure.

这些芯片运行时会非常热，对吧？

These chips run extremely hot, right?

它们确实很热。

They're hot.

关于冷却，对于刚接触这个话题的人，我们来谈谈冷却。

Cooling, people talk about cooling for those people who are coming to this for the first time.

要想成功长期运行AI数据中心，你必须能够为芯片降温。

Being able to run an AI data center, you gotta be able to cool the chips if you wanna be able to be successful this long

我认为这是市场误解的一个方面。

is one of the things that I think the market misunderstands.

对吧？

Right?

每个人都认为液冷数据中心的管道系统存在某种差异。

Is that everybody believes that this that there's some differentiation in the plumbing of the liquid cooled data center.

对吧？

Right?

真正的差异并不在这里。

That's not where the differentiation lies.

所有的管道、阀门和接头都是一样的。

It's all the same pipe and valves and fittings.

大家用的都是同样的东西。

Like, everyone's using the same things there.

差异在于你开启系统后，如何控制这些系统。

The differentiation comes after you turn it on and how you control those systems.

好的。

Okay.

对吧？

Right?

我们公司在这方面做得非常出色，但过去几年我们一直刻意对外保密，因为这是我们独有的核心优势——从电力和冷却基础设施到GPU、服务器，我们全程负责数据中心的部署、验证和管理。这也是为什么世界上最有价值的公司、最大的AI实验室都选择我们来运行他们最关键的训练任务。

And that's what we've done incredibly well as a company that we've very consciously not spoken about externally for the past couple years, because it is our secret sauce, is how we provision, validate, and manage those data centers all the way from the power cooling infrastructure up through the GPUs, the servers, and it's why the like the most valuable companies in the world, the biggest AI labs actually use us to run their most critical training jobs.

对。

Right.

我的意思是，这是一项极其艰巨的任务，对吧？

I mean, it's a Herculean task, right?

理解生态系统时很重要的一点是，当你在思考不同的Neo Cloud时，

It's important to understand that when when you're when you're thinking about the ecosystem, right, and you're thinking about the different Neo Clouds that populate the

什么是Neo Cloud？

And what's a Neo Cloud?

所以，Neo Cloud？

So, Neo Cloud?

这是最糟糕的术语了。

The worst term ever.

我讨厌这个词。

I hate it.

可以把它想象成，你知道，在通用术语中，大家都知道AWS是谁，知道亚马逊，知道微软，也知道谷歌。

Think of it as, like, you know, in the common vernacular, you know, everybody knows who AWS is, you know, Amazon, they know who Microsoft is, they know who Google is.

这些就是超大规模云提供商，对吧？

Those are the hyperscalers, right?

如果你愿意的话，也可以把甲骨文算进去。

You can throw Oracle in there if you'd like.

但还有一类供应商能够提供这种基础设施，而你知道，我们是其中的领军者。

But then there's a class of providers that can deliver this infrastructure, and you know, we are the leader among that.

需要理解的重要一点是，即使将所有其他NeoCloud的GPU集群加起来，我们在实际运行并交付给客户的GPU数量上仍然是它们总和的好几倍。

And what is important to understand, that if you took all of the other NeoClouds and added their GPU fleets up, we would still be a multiple of all of them combined in terms of the number of GPUs that are up and running and delivered to clients.

所以

And so

很大的倍数。

Large multiple.

当Brian谈到市场难以理解的事情时，关键是要明白我们的差异化优势——我们构建的软件套件能够将商用GPU转化为去商品化的高端服务，让人们从这套基础设施中榨取尽可能多的价值。

What when Brian is talking about, you know, things that the market is struggling to understand, is it's important to understand that what differentiates us, what allows us to be as successful as that we have, is that the software suite that we have built allows us to take the commodity GPU and deliver a decommoditized premium service that allows people to extract as much value from this infrastructure as possibly can be extracted.

这正是CoreWeave正在做的事情，也解释了为什么当Brian说，世界上领先的公司和顶尖实验室都在依赖我们提供服务时，情况就是这样。

And that's really what CoreWeave is doing, and it's why when Brian says, Hey, you know, the leading companies in the world and the leading labs in the world are relying upon us to deliver our service.

这就是原因。

That is why.

因为最终他们所获得的产品，能够最大程度地提高他们使用GPU成功开发公司产品的可能性。

It's because the product that ultimately they receive is the product that will allow them the greatest probability of being successful at using the GPUs to deliver products that their company is building.

对。

Right.

所以用通俗的话说，这非常有帮助：当像微软这样的公司与你们合作构建人工智能基础设施时，你们已经构建了一些专有组件，比如冷却系统、运行数据中心的软件，这些都让他们能从芯片中获得比以往更多的价值。

So just to put it in plain English Always helpful When for a company like Microsoft will work with you on building infrastructure for artificial intelligence, you've built some proprietary pieces of the puzzle, like your cooling system, like the software that runs the data center, and that allows them to get more out of the chips than they would have typically.

是的，这里的关键在于，当你建造一个拥有3000英里光纤电缆、百万级光模块连接到交换机的数据中心时，这些组件都会出故障，对吧？

Yeah, and nuance here is that when you build one of these data centers and it has 3,000 miles of fiber optic cabling, and it has a million optics that connect into the switches, like these things all fail, right?

一旦发生故障，如今训练任务的运行方式是：如果某个组件失效或性能受限，整个训练过程的性能将由表现最差的组件决定。

And when they fail, the way that training jobs are run today is if one component fails or one component limits the performance, the balance of the training run is going to be governed by the worst performing component.

哦。

Oh.

对吧？

Right?

我们的全部工作就是构建自动化、预测性分析以及机器学习模型，用来识别：好吧，我们这里发现问题了。

And our entire job is to build the automation, the predictive analytics, the you know, the machine learning models around saying, okay, we're seeing a problem here.

我们该如何优雅地处理这些问题，以最小化对客户任务的影响，对吧？

How do we gracefully handle these things so it has the least impact on our customers jobs, right?

这就是我们的核心秘密武器。

And that's the core we've secret sauce.

好的。

Okay.

我们拥有全球最大的数据集，记录了这些系统如何运行、如何故障，我们已经构建了所有恢复机制和软件智能，以帮助客户运行这些系统。

Is that we have the world's largest dataset of how these things run, how they fail and we've built all the recovery mechanisms and the software intelligence to help our customers run these things.

你们客户的需求，你提到的训练，主要是用于训练AI模型吗？

Is the demand that you're getting from your customers, you mentioned you know training very well, is it mostly training the AI models?

因为，嗯，大部分基础设施之前都是用来扩展这些模型，投入更多计算资源、更多数据，让模型变得更大，然后模型就会变得更好。

Because, well, that's what a lot of the infrastructure has been used for, scaling these models, throwing more compute at them, throwing more data, making the models bigger, and then the idea is that the models get better.

所以，你们看到的需求主要集中在训练方面，还是已经转向了推理阶段，即公司实际在使用这些模型并将其部署到生产环境中？

So are you seeing most of your demand in the training side of things or has it gone to inference where like companies are actually using the models and deploying them into production?

这是个很好的问题，它反映了过去三年市场的发展方向以及未来趋势的分化。

So it's a great question and I think talks to the split or this kind of delineation of where the market's been for the last three years and where it's going.

过去三年，我们的客户群体主要是最大的AI实验室和正在构建AI能力的企业，对吧？

You know, our customer base for the last three years has primarily been the largest AI labs and enterprises that are building the capabilities of AI, right?

而现在，焦点已经从构建这些能力的人，转向了那些希望利用这些能力来改变业务成果的人。

And it's now shifted from the people building those capabilities to the people that wanna use those capabilities to change business outcomes.

所有企业级采用都来自这个领域。

And this is where all the enterprise adoption's coming from.

我最喜欢的服务之一就是Lovable，对吧？

You know, it's one of my favorite services out there is Lovable, right?

你访问Lovable，就可以构建任何你想要的应用。

You go to Lovable, you can build any app you want.

有一个聊天机器人会引导你完成整个过程。

There's a chat bot that helps you go through it.

你知道，我们终于开始看到人们将这些能力串联起来，构建真正解决问题的产品。

You know, we're finally starting to see people chain together these capabilities to build real products that solve problems.

过去三年，我们的业务主要围绕这些能力的创建，但很快就转向了不仅包括创建，还涵盖它们的部署和在商业实践中的应用。

And our business for the last three years has really been around the creation of those capabilities and has very quickly shifted to include not just the creation of them, but the deployment of them and use in business practices.

好的。

Alright.

所以，有一件事是我没想到的：两年前看起来像是训练的东西，如今却成了推理的样子，对吧？

So, one of the things that I didn't expect was that what looked like training two years ago, is how inference was gonna look today, right?

这意味着你仍然依赖于高度互联的存储，你知道，后端网络变得至关重要，因为模型太大了。

Is that you're still dependent upon highly connected storage, you know, your back end networks become critical to this because the models are so large.

因此，我们为构建这些能力所部署的训练基础设施，与客户最终用于提供服务的基础设施，实际上没有区别。

So, there's really no difference between training infrastructure we deploy to build those capabilities, and what our customers are ultimately using to serve them.

那么，推理对你来说已经超越训练了吗？

So has has inference overtaken training for you?

我们处理着海量的推理需求。

We serve a tremendous amount of inference.

但你知道吗？

But you know what?

不。

No.

我其实不知道这个问题的答案。

I actually don't know the answer to that.

真的吗？

Really?

六个月前，我会说训练占三分之二，推理占三分之一。

Six months ago, I would have said it was two thirds training and one third inference.

现在可能差不多各占一半了。

It's probably close to fifty fifty now.

好的。

Okay.

但我们的某些大客户会从使用校园进行训练，到推出新产品后，不得不将推理需求外溢出去。

But there's also some of our big customers that they go from they'll use a campus for training, they'll launch a new product, they'll have to spill over for inference.

你知道，很多这些东西都是动态的，而且设计之初就是如此。

You know, a lot of this is very dynamic and it's been built to be so.

是的。

Yeah.

这可能为你们在播客中最终会谈到的其他话题提供一个过渡，但对我来说，观察推理过程，理解推理是人工智能投资的变现方式，是人工智能领域最令人兴奋的趋势之一。

I this may provide a segue to some of the other subjects that you'll ultimately get to in this podcast, but, you know, for me, watching inference, understanding that inference is the monetization of the investment in artificial intelligence, is one of the most exciting trends that exists within AI.

我们有机会全面观察几乎所有大型重要实验室构建这些技术的过程，看到它们越来越多地从大约三分之一的推理占比逐渐上升到50%，有时甚至超过50%的算力资源用于推理，这充分说明了市场对使用人工智能服务客户需求的巨大规模。

And we have a front row seat across the entire cross section of almost every large, important lab that's building this stuff and watching them increasingly, you know, move from, let's say, know, one third inference climbing towards, you know, 50%, and at times it's even over 50% of the fleet being used for inference, you know, is just an amazing indication of the scale of the demand to use artificial intelligence to serve customer inquiry.

这意味着一切。

And that means everything.

好吧，再问一个问题。

Alright, one more question about this.

嗯。

Yep.

为什么需要CoreWeave这家公司？

Why does CoreWeave need to exist?

我的意思是，我们在谈论像微软这样的大公司。

Why I mean, we're talking about these big companies like Microsoft.

那他们为什么不自己建造数据中心呢？

Like, why wouldn't they just build their own data centers?

他们为什么要从第三方授权呢？

Why are they licensing it from a third party?

这是个很好的问题。

So, it's a great question.

这个市场曾经存在一个空白，对吧？

There was a void in this market, right?

这里有几方面的原因。

And how there's a couple pieces here.

当今世界上最大的云服务商都是建立在周边业务的现金引擎之上的，对吧？

The biggest clouds in the world today are built off the cash engines of peripheral businesses, right?

谷歌建立在搜索之上，亚马逊建立在零售之上，微软则是建立在企业软件之上。

Google's built on search, Amazon's built on retail, Microsoft was Microsoft was built on enterprise software.

我们几乎是凭空出现的，对吧？

We came pretty much out of nowhere, right?

让我们能够进入这个位置的关键时刻，是由加密货币推动的，对吧？

And our the the moment in time for us to be able to get ourselves into this position was driven by crypto, right?

你之前提到，我们是从以太坊挖矿起步的。

You mentioned earlier that we came out of, you know, Ethereum mining.

我们利用以太坊挖矿的收入，去建设并部署了更大的规模，这样当加密货币热潮退去时，我们的基础设施已经就位，并且 hopefully 拥有了足够多的客户，实现了逃逸速度，对吧？

We were able to leverage the revenue from Ethereum mining to go out and build and deploy additional scale, so that when crypto went away, we had the infrastructure in place and we hopefully had enough clients that we became, like we were an escape velocity, right?

所以我们意识到，计算能力将会变得非常有价值。

So, you know, we recognize that compute was gonna be valuable.

但当时我们并不确定它究竟会在哪些方面变得有价值。

We didn't necessarily know at the time what it was going be valuable for.

比如，我和迈克从未想过，未来每年会有数千亿美元的资本支出投入到人工智能领域。

Like, I don't think Mike and I ever had this idea of like, there's going to be this hundreds of billions of dollars a year in CapEx for AI.

但我们有一个观点：计算能力将变得极其宝贵，我们希望掌控其中很大一部分。

But you know, we had the thesis that compute is going to be incredibly valuable and we wanted to own a lot of it.

我们把这种计算资源视为一种选择，然后想：好吧，我们能用它来做些什么最好的事情？

And we looked at that compute resource as an option, like, and we said, okay, what are the best things that we can do with this?

这正是我们一直以来应对不同业务问题的方式，对吧？

And that's how we've always approached different business problems, right?

我们的资产是什么？

It's like, what is our asset?

我们如何最有效地将其变现？

How do we monetize it the most effectively?

用它来实现最有价值的方式是什么？

What's the most valuable way to use this?

但我要插一句，我想回到我们刚开始讨论时提到的一件事，那就是，我们从零开始构建了一套软件栈，专门优化并行计算的使用场景。

But then So, so, I'm going to jump in here on this, but I want to go back to something that we kind of talked through as we started this, right, is that, like, we've built a software stack from the ground up to optimize for the use cases associated with parallelized computing.

我们在这一点上做得比任何人都好。

We do it better than anyone else.

我们存在的原因是我们提供了一款极受欢迎的出色产品。

The reason we exist is because we deliver a fantastic product that is highly in demand.

而且极具差异化。

And incredibly differentiated.

而且极具差异化。

And incredibly differentiated.

所以我们服务的不仅是最大的玩家，还有很多其他正在构建应用程序的AI公司，它们可以选择使用我们，也可以选择使用各大云服务商。

And so, you know, we serve the largest players, but we also serve, you know, a ton of other AI companies that are building applications where they have the choice to go and use us or to go and use one of the hyperscalers.

其中许多公司选择使用我们的解决方案，因为这能让他们更高效地提供计算能力。

And many, many, many of them choose to use our solution because it allows them to more effectively deliver compute.

但这一点常常被忽视：人们并不理解从云计算1.0转向云计算2.0、从顺序计算转向并行计算的根本性变革。

And one of the things that's really just lost on this is that there's not an understanding of how fundamental the change from Cloud one point zero into Cloud two point zero as you move from sequential computing into parallelized computing.

当你从托管网站和数据湖跃升到为人工智能驱动并行计算时，很自然地，计算使用方式的根本性变化也要求云架构本身进行根本性的变革。

And when you made that leap, right, from know, hosting websites and data lakes into driving parallelized computing for artificial intelligence, it stands to reason that a fundamental change in how compute is used will also require a fundamental change in how you build the cloud to serve it.

我们抓住了这一转型机遇，打造了业界领先的产品。

And we took advantage of that transition to build best in class solutions.

没错。

Right.

这就是我们存在的原因。

And that's why we exist.

所以我听到一种说法，认为大型科技公司为了建设这些数据中心，必须提前数年预测需求。

So I've heard an argument made that basically the big tech companies, you know, to build these data centers, they have to forecast demand out years in advance.

这是一项巨大的资本投入。

It's a massive capital commitment.

他们不确定是否能获得回报。

They are not sure whether it will pay off.

而CoreWeave对他们来说很有用，因为你们承担了风险，他们则可以使用你们的算力，相当于租用，而不必自己进行这些大规模投资。

And CoreWeave is useful to them because you're taking the risk and then they will be able to use your capacity and sort of rent it out as opposed to having to make these big investments on their own.

如果出问题，责任在他们。

And it's their asses if things go wrong.

是的。

Yeah.

听我说，这只是一个说法。

Look, you know, that is a narrative.

我不认为这与实际情况相符。

I don't think that actually tracks with the reality of the situation.

我认为实际情况是，大型超大规模云服务商正在尽最大努力快速建设。

I think the reality of the situation is the large hyperscalers are building as fast as they can.

谷歌直接发布了一份新闻稿，称他们正在建设价值500亿美元的基础设施，同时仍在向其他所有供应商采购。

Google went out and just, you know, released a press release where they're building $50,000,000,000 worth of infrastructure while they're still buying from everyone else they can.

微软正在内部建设，同时也从许多其他供应商那里采购。

Microsoft is building internally and they're buying from lots of other players.

我觉得这个论点更像是模型拟合，对吧？

What I feel like that argument is model fitting, right?

这是有人带着先入为主的观念，然后重新构建现实中的事实来迎合这个模型，以便可以说：看，我没错。

It is somebody's got a preconceived notion of what this is going to look like, and now they're reconstructing the the facts on the ground to fit that model so that they can say, Look, I'm right.

但实际情况是，我对它的看法非常不同，对吧？

But the reality is, is that I look at it very differently, right?

我看待我们如何建立起相对于超大规模云服务商以及其他新兴云服务商的竞争优势的方式。

I look at the way that we built our competitive advantage over, you know, the hyperscalers, the way that we built our competitive advantage over other neo clouds.

我们做到这一点的方式是，我们意识到这种计算方式将会变得重要，因此提前构建了基础设施和软件，以便在需求出现时能够提供服务。

And the way that we did that is we understood that this type of computing was going to be important, and we built the infrastructure and the software to be able to serve it when the demand emerged.

而且我们是以一种风险可控的方式完成的。

And we did it in a very risk managed way.

当我展望未来，思考投入到构建AI工厂的资金时，我会比较投入数据中心的资金与投入数据中心内部计算资源的资金，我认为数据中心本质上是一种期权，用于在未来提供并保持在计算交付中的相关性，对吧？

When I look at the future, when I think about investments that go into building an AI factory, and I think about how much money is being put into the data center versus how much money is being put into the compute that goes inside of the data center, I think about the data centers as being basically an option on being able to provide and be relevant for the delivery of compute into the future, right?

我们公司把风险资金投入到那些最关键的环节上。

We take our risk dollars as a company and we invest in the long poles.

而这些关键环节主要有两个方面。

And the long poles are really twofold.

一个是打造世界上最好的软件，另一个是确保能够获得数据中心的容量，以便在需求激增时能够及时交付计算能力。

One is building the best software in the world, and the second one is having access to the data center capacity to be able to deliver compute when a wave of demand hits this market that requires you to deliver it.

你不能一觉醒来就说：嘿，我想交付一吉瓦的基础设施。

You can't just wake up and say, Hey, I want to deliver a gigawatt worth of infrastructure.

你必须提前数年就开始建设这一吉瓦的基础设施，这样当你的客户说：嘿，我刚刚开发出一种新的AI使用方式，需要一吉瓦的基础设施时，你才能及时满足需求。

What you'd have to do is you have to start years in advance building that gigawatt of infrastructure so that you're in a position that when your customers say, Hey, I just produced a new way of using AI that's going to require a gigawatt worth of infrastructure, you're able to serve it.

我们将拥有一个庞大的基础设施组合，能够在未来部署，对此我们感到非常兴奋。

We're going to have a tremendous portfolio of infrastructure that is going to be able to be deployed into the future, and we're really excited about that.

我们认为这是发展业务的绝佳方式。

We think it's a wonderful way to go about building our business.

对。

Right.

这就是关于押注的问题，对吧？你押注于人工智能将继续以惊人的速度被采用。

And that's the question about the bet, right, is that you're betting that AI is going to continue to be adopted at a wild rate.

这种说法并不完全准确。

That's not entirely accurate.

好吧，让我们

Okay, let's

我们目前的做法是，通过与信用良好的实体签订长期合同，利用这些合同作为融资手段来建设基础设施，从而确保需求、信用和资本都已到位，对吧？

What hear we are doing is we are making the majority of our investments by taking long term contracts from creditworthy entities, using those contracts as a way of raising money to build the infrastructure where the demand and the credit and the capital has already been secured, right?

所以，比如说，我们85%的投入都是用于向投资级机构、AI实验室或其他大型算力消费者提供算力，对吧？

So, let's say 85% of our exposure is to deliver compute to investment grade or AI labs or other large consumers of compute, right?

其余的15%是我们对未来实现同样目标的长期合同敞口。

The other 15% is our exposure to long term contracts to be able to do that exact thing in the future.

这就是我的看法，我认为这是一种更好地理解我们如何承担风险、如何处理杠杆以及如何定位自己的方式。

And that's the way I look at it, and I think it's a much better way to think about how we're taking on risk, how we're dealing with leverage, and how we're positioning ourselves.

如果市场继续增长，我们的位置非常有利。

If the market continues to grow, we're in a great position.

如果市场稳定在这一水平附近，我们也完全没问题。

If the market stabilizes in and around this, we're fine.

如果市场收缩，出现了新技术，那么我们那15%中的一部分可能会被迫等待数年，直到市场重新增长到那个水平。

If the market contracts, there's some new technology, then we will be left with some portion of that 15% that we may be in a position where it has to wait for a few years before the market grows back into it.

对此，我们完全能够接受。

And we are fine with that.

我们看待问题的角度是，人们常说，这家公司的创始人用一种不同的视角来看待世界，因为我们并非来自硅谷，而是来自大宗商品领域和华尔街。

We think of it from and, know, people have talked about how the founders of this company kind of look at the world with a different lens, because we don't come from Silicon Valley, you know, we come from the commodity space, we come from Wall Street.

我们考虑的是期权价值，对吧？

We think about option value, right?

当我们考虑计算能力时，我们思考的是与之相关的期权价值。

When we think about compute, we think about what is the option value associated with it?

当我们考虑数据中心时，我们思考的是未来保持相关性的期权价值。

When we think about the data centers, we think about what is the option value to be able to build to be relevant in the future.

这就是我们分配风险并确保当前合同的方式。

And that's the way we kind of go about allocating our risks and securing the contracts that we have in place right now.

是的。

Yep.

说到这一点，你提到如果市场收缩，我认为我们反而会很高兴，因为这为我们带来了巨大的机会。

And, you know, to speak to one thing here, you talked about if the market contracts, I think that we would love for that because it presents tremendous opportunity for us.

为什么？

How?

对吧？

Right?

我的意思是，你处于一个拥有困境资产、并购机会的位置，这才是真正的机遇所在。很多时候，我们会说，好吧，我们在寻找并购机会，想投资一些东西，但估值并不合理。

I mean, you're in a position where there's going be distressed assets, there's going to be consolidation possibilities, like that's when opportunity really comes in and you know, there's a lot of times where we sit there and say, Okay, we're looking for M and A, we're looking to invest in things, but the valuations don't make sense.

对于迈克和我来说，我们职业生涯就是等待这些机会，当别人遇到困难时，我们就说：好吧，这些就是我想收购的东西，对吧？

And for Mike and I, you know, we've made our careers on waiting for those opportunities and saying, okay, these are the things that I want to buy when things don't necessarily go right for them, right?

而这正是让我们兴奋的地方。

And, you know, that's really what excites us.

上周，我们另一位创始人给我打了电话。

Know, one of our other founders last week, he got on the phone with me.

他说：我太喜欢这个布赖恩了。

He's like, I love this Brian.

我问：哪个布赖恩？

I'm like, what Brian?

他说：就是那个你一上来就特别关注机会在哪儿的布赖恩。

He's like, this is the one where you start like, you're so focused on like where are the opportunities?

我该怎么去接管这些事情？

How do I go take things over?

有时候我会跟一些人说，我觉得当市场出现逆风时，反而更容易做这份工作。

And, you know, it's I say it to some people every once in a while is that I feel like when there's headwinds in the market, it's actually easier to do this job.

对。

Right.

对吧？

Right?

而不是当顺风以每小时一千英里的速度吹拂时。

Than when the tailwinds are kind of blowing at a thousand miles an hour.

但我可以问一下，你们是如何设立公司的，以确保在合同出现问题时，你们不会成为困境资产？我的意思是，当合同

But can I ask, how have you set up the company to make sure that you're not the distressed asset when the contract if the contract I mean, at

看看我们客户合同组合的结构，对吧？

our look at our construction of customer of our customer contract portfolio, right?

去年每个人都谈论客户集中度和对微软的依赖是坏事，但他们的资产负债表比美国政府还要好，对吧？

Is everybody last year talked about how customer concentration and exposure to Microsoft was a bad thing, but they have a better balance sheet than the US government, right?

我不担心他们无法履行对我们的长期义务。

Like, I'm not worried about them performing in their long term obligations to us.

这基本上是我们能处于的最好位置。

Like, that's basically the best possible position we can be in.

我们非常谨慎地选择合作的客户，并管理信用风险，以确保我们所做的投资都能得到偿还。

And we've been super thoughtful about the way that we choose which customers to work with and how we manage the credit exposure, so that we're, like, we're certain that the investments we make will be paid back.

如果你看看为这些项目提供债务资金的人，比如黑石，他们是全球最精明的人之一，他们的承销委员会竟然会进来表示：‘是的，我想做这个项目，而且想尽可能激进地扩大规模。’

And if you look at the people that are providing us the the debt to do those projects, like Blackstone, right, they're the some of the most sophisticated people in the world and for their underwriting committee committees to come in and say, yes, I wanna do this and I wanna scale it up as aggressively as possible.

你是在告诉我，要让某个金融分析师去和约翰·格雷对抗吗？

Like, you're telling me you're gonna pit some financial analyst against John Gray?

我会选择约翰·格雷。

I'm gonna go with John Gray.

是的。

Yeah.

好吧，我想稍微多说一点，关于我们如何扩展业务以及如何使用债务的一个基本要素，因为我认为这是外界对如何构建或我们如何构建这家公司的一个误解点。

Well, you know, I mean, may maybe a second on just, like, kind of one of the fundamental building blocks of how we expanded the way we have and how we use debt, because I think that's one of the misunderstood components of how you build or how we have built this company.

因此，非常重要的是要理解，我们构建这些组件的方式是：进入市场，以微软为例——因为我们确实用过他们，但还有很多其他客户，从结构角度来看，它们完全可互换。

And so, it is really important to understand that we the way that we build the components is, we go into the market let's use Microsoft because we've used them, but there's lots of other clients you could use and they're totally interchangeable from the perspective of the structure is still the same.

我们去找他们，说：‘嘿，我们有这个数据中心的访问权限’，他们说：‘我们需要算力。’

We go to them and we say, Hey, you know, we've got access to this data center, they say we need compute.

我们说：好的，我们要签订合同。

We say, Okay, we're going to sign a contract.

他们签订了一份五年期的合同。

They sign a contract for five years.

我们将这份合同结构化，以便能够回到像黑石这样的机构，向他们借款来建设基础设施，为微软提供服务。

We structure that contract in a way that we can go back out to the Blackstones of the world and we can borrow money from them to go ahead and build the infrastructure to deliver to Microsoft.

在与微软合同期的五年内，我们会支付基础设施费用、运营支出和利息，并从基础设施中获得巨额利润。

Within the five years of the contracted period with Microsoft, we pay for the infrastructure, we pay for the OpEx, we pay for the interest, and we earn an enormous margin on the infrastructure.

是的，确实有债务。

So yes, there is debt.

我们并不否认这一点。

We're not arguing that.

我们坚信，当你在这一规模上建设任何类型的基础设施时，债务是最合适的方式。

We believe fundamentally, when you build any type of infrastructure at this scale, debt is the correct way to go about doing it.

纵观历史，无论是建设发电厂、电力配送网络，还是电话系统、蒸汽机或铁路，你都会发现，这一直都是你所使用的工具，对吧？

The examples run through history, whether you're talking about building a power plant, building a distribution grid for electricity, whether you're talking about the telephone, whether you're talking about the steam engine or railroads, like, you go throughout history, this is the tool that you use, right?

我们并没有发明任何新东西，只是采用了经过验证的方法，并将其应用于与该资产相关的折旧特性以及该资产的过时曲线，调整了参数，使其能够以严密的方式运作，让约翰·格雷、黑石、黑岩或任何大型贷款方都能一看就明白他们是如何对这笔交易进行承销的。

We didn't invent anything new here, we just took a tried and true method and applied it to the specifics of depreciation associated with this asset, of the obsolescence curve associated with this asset, and made the contours so that it worked in an airtight manner, so that guys like John Gray or, you know, Blackstone or any or BlackRock or any of the big lenders could look at it and say, I understand how they're going to underwrite this.

我理解这其中的风险。

I understand the risk in this.

我明白这些人会向资产负债表提供算力。

I understand that these guys are going to deliver compute to that balance sheet.

他们会得到偿还，而当他们得到偿还时，我们也会得到偿还。

They're going get paid back, and when they get paid back, we're going to get paid back.

所以，让我们借钱给他们。

So let's lend them the money.

没错。

Yeah.

这一点在市场上被忽视了。

And that's lost on the market.

他们以为我们到处冒险，拥有不可思议的风险承担能力，但实际上这是一种风险极低的做法。

They think we're running around with this, like, you know, incredible capacity to take on risk, but that's a really low risk approach.

事实上，这比说我们要用股权来做要低风险得多，因为我们把股权留给了那些必须投资的长期项目。

Matter of fact, it's way more low risk than saying, hey, we're going to do it on equity because we're saving our equity for the long poles that you've got to invest in.

这才是你应该集中火力的地方。

That's where you want to put your bullets.

你应该利用债务市场来处理折旧资产。

You want to use the debt markets to deal with a depreciating asset.

这就是标准做法。

It's the way it's done.

这自古以来就是这么做的。

It's the way it's been done throughout history.

是的。

Yeah.

顺便说一句，我们能进行这场对话真是太好了。

By the way, it's great that we're able to have this conversation.

这正是我们节目想做的：把复杂的东西讲清楚，谈谈公众的反应，与核心人物对话，真正讲出这个故事。

This is what we want to do on the show, is take this complex stuff, talk about what the reactions have been in public, speak with the principals, and actually get the story.

谢谢你和我一起梳理这个问题。

So thank you for talking it through with me.

就此而言，我们继续吧。

And on that note, let's continue.

我认为人们会提出的论点并不是微软对资金不利。

The argument I think that would be made is not that Microsoft isn't good for the money.

人们会认为生成式人工智能仍然是一个正在发展的领域。

The argument would be made that generative AI is still a developing category.

它尚未展现出持续盈利的能力。

It hasn't really shown the ability to turn consistent profit.

因此，那些大力投资于该领域的企业，有一天可能会醒来意识到：我们其实并不想继续这种扩张。

And so the companies that are investing in a big way in it may one day wake up and say, you know, we can't we don't really want to do that build out.

以OpenAI为例，我们就拿它来说。

OpenAI, for instance, let's just use them as an example.

它们承诺投入约1.4万亿美元用于基础设施建设。

They have something like 1,400,000,000,000.0 committed to spend on infrastructure.

我认为只有OpenAI相信他们真的会花掉这1.4万亿美元，也许他们的投资者也这么想。

I think OpenAI might be the only ones that believe that they'll actually spend that $1,400,000,000,000 and maybe they're investors.

那你对这种风险怎么看？

So what do you think about that risk?

因为AI是新兴领域，不像其他类别那样可预测，如果用债务融资，即使像微软这样信用评级极佳的公司，风险也更高。

That AI is because AI is new and not as predictable as you would have in a different category, financed by debt, that therefore it is riskier even if the credit rating of a company like Microsoft is golden.

你

You're

关于OpenAI有几点要说，因为他们在人工智能的许多方面都处于前沿。

a couple of things on OpenAI because they're they, you know, they are the tip of the spear in in in many ways for artificial intelligence.

他们的产品拥有8亿月活跃用户，这是一个庞大的用户群体。

They have a franchise that has 800,000,000 monthly users of their product, which is A lot of users.

全球每10个人中就有1个人会登录OpenAI。

Fully one tenth, one out of every 10 human beings on the planet Yeah.

登录OpenAI

Logs on to OpenAI