$CRWV

CRWV 因巨额资本支出(140亿美元)相对于较低的经营现金流(36亿美元)以及疲弱的资产负债表(短期债务与现金比率为3:1)而面临高债务风险。

看跌
Sam Altman JUST Accidentally Exposed the AI Collapse.
Meet Kevin发布 2026-09-02 · 18 段内容
05:56/ 打开原视频

逐段跳播

18 段内容
5:5616:24

Sam Altman 关于 Neo 云不可持续的观点是否正确?分析这一点的最佳方式是查看 CoreWeave 的财务数据,我们稍后将进行。

Is Sam Altman right about the unsustainability of Neo Clouds? The best way to analyze this is to look at CoreWeave's financial data, which we will do in a few moments.

然而,这将是最后一次提供终身订阅。那么,这种循环性质是如何运作的?我们在 CoreWeave 实际看到了什么?像 Cudo 这样的公司发生了什么?

However, this will be the last time a lifetime subscription is available. So, how does this circular nature work ? What do we actually see at CoreWeave? What's happening with companies like Cudo?

好的。这个想法很简单,像 CoreWeave 这样的公司会说:“我们将建设基础设施,客户自然会来。”他们正在建设大量的处理能力。如果他们只能租出一部分,而无法租出全部产能,他们就不需要购买更多的英伟达芯片。

Good. The idea is simply for a company like CoreWeave to say: "We will build the infrastructure, and the customers will come." They are building a huge amount of processing capacity. If they can only lease a portion and cannot lease their full capacity, they will not need to buy more Nvidia chips.

然后他们最终会去找英伟达说:“看,我们需要你处理这件事。我们需要你对此做点什么。”

Then they will end up going to Nvidia and saying, "Look, we need you to take care of this." "We need you to do something about this."

他们认为 CoreWeave 是目前现代云领域的最佳买入标的。他们提出了一些合乎逻辑的观点。他们指出,其股价仅为营收的 2.2 倍。对于寻求押注 AI 热潮的风险承受型投资者来说,这是一个机会。

我们预计今年的营收为 129 亿美元。这相当于 Nebius 营收的四倍,也是 iRent 营收的四倍。我们拥有 1.5 吉瓦的活跃电力,另有 3.7 吉瓦已签约。

They say Coreweave is the best buy in the modern cloud space right now. They are making some logical points here. They say it is trading at only 2.2 times its revenue. There is an opportunity for risk-tolerant investors looking to bet on the AI boom. We have $12.9 billion in revenue projected for this year. This is equivalent to four times the revenue of Nebius, and four times what iRent provides. We have 1.5 GW of active power and another 3.7 GW under contract.

好吧,这没问题。所以,我们对 CoreWeave 抱有极大的热情,但他们认为,由于客户集中度风险,尤其是微软——它已开始购买自己的计算能力——CoreWeave 的股价被折价了。

Okay , that's fine. So, we have a lot of enthusiasm directed towards Coreweave, but they argue that it is discounted because of the risks of customer concentration, particularly Microsoft, which has now begun to purchase its own computing capabilities.

那么,是否存在这样的风险:如果 Meta 和 OpenAI 生产自己的“Jalapeno”芯片,并且 Meta 和微软购买自己的硬件,他们可能就不再需要 CoreWeave 了?这是否是真正的危险?

So, is there a risk that if Meta and OpenAI produce their own "Jalapeno" chips , and Meta and Microsoft buy their own hardware, they might not need CoreWeave? Is this the real danger?

他们认为 CoreWeave 正在努力实现客户多元化。他们有一家名为“Gane Street”的公司可以销售。嘿,有很多人会租用这种容量。这可能是真的。

They believe that CoreWeave is working on diversifying its clientele. They have a company called "Gane Street" to sell to. Hey, there are lots of people who will rent this capacity. And that may be true .

但我实际上认为这里有一个更大的故事。因为我们必须考虑到,这些数据中心可能需要被英伟达回购。

But I actually think there's a bigger story going on here. Because we somehow have to take into account that those data centers may need to be repurchased by Nvidia.

但如果计算能力过剩,为什么 CoreWeave 或 OpenAI 还在购买 Mac Ultra,而与此同时 CoreWeave 的股价却处于折价状态?

But if there is a surplus of computing, why are CoreWeave or OpenAI buying Mac Ultras , while at the same time CoreWeave is trading at a discount?

我认为正在发生的是,人工智能的分化正在将其分裂为两种不同的芯片组件。具体来说,有企业推理芯片,然后有高级模型推理芯片。

I think what's happening is that the divergence of artificial intelligence is splitting it into two different chip components. You specifically have enterprise inference chips, and then you have advanced model inference chips.

这或许可以解释我们今天可能看到的股市走势。

This may explain the likely stock market movement we are witnessing today.

当然,CoreWeave——据 The Information 称这是一笔收购交易——但你会发现像 Sam Altman 这样的人批评它,说:“我不知道,老兄。”他们在扩张中可能会变得有些投机。

And of course, CoreWeave, which the information claims is a purchase deal, but you find people like Sam Altman criticizing it, saying, "I don't know, man." They may become a bit speculative in their expansion."

所以,当你审视 CoreWeave 时,他们的大部分芯片——至少基于公开文件和相关报告——预计将是高级训练型芯片。是的,他们正在转向推理。他们正在转向 RTX 6000。他们去年就转向了。预测表明这种敞口非常低。机构的推理是以下限制。

So , when you look at CoreWeave , the majority of their chips, at least based on public files and reports on the subject, are expected to be advanced training-type chips. Yes, they are moving on to reasoning. They are moving to the RTX 6000. They moved to that last year. The projections indicate that this exposure is very low . The reasoning of institutions is the following limit.

但如果你被困在大量先进数据中心芯片中,是否存在债务风险对像 CoreWeave 这样的公司造成压力?就我个人而言,当我审视 CoreWeave 的财务数据时,我认为答案是肯定的。

But if you're stuck with a lot of advanced data center chips , is there a debt risk putting pressure on a company like CoreWeave? Personally, when I look at CoreWeave's financial data, I think the answer is yes.

我们不需要知道他们在技术上亏损。这已经是旧闻了。如果再加上折旧,他们是赚钱的。但这是他们现金流的投资回报,因为他们将巨额资金用于资本支出。他们花费 140 亿美元,而运营现金流为 36 亿美元。

所以他们显然需要借款或筹集资金,这正是他们正在做的。自便吧。他们在资本支出上花费了 140 亿美元。他们筹集了 30 亿美元的股权和 110 亿美元的净融资。

We don't need to know that they are technically losing money. This is already old news. If you add depreciation again, they make money. But this is a return on investment from their cash flows, as they spend huge sums on capital expenditures. They are spending $14 billion against $3.6 billion in operating cash flow . So they clearly need to borrow or raise money, which is exactly what they are doing. Help yourself. They spent $14 billion on capital expenditures. They raised $3 billion in equity and $11 billion net in financing.

好的。公司都会这样做。他们试图扩张。他们试图投资。但 CoreWeave 的问题在于其资产负债表不太好。与 Enphase 这样的公司相比,他们的资产负债表确实很弱。

Good. Companies do this. They are trying to expand. They are trying to invest. But the problem with Coreweave is that its balance sheet is not very good. Their balance sheet, compared to a company like Enphase, is really weak.

另一方面,CoreWeave 拥有 55 亿美元现金,而应付账款为 180 亿美元。他们每持有 1 美元现金,就有 3 美元的短期债务。这并不好。这会减缓扩张并使融资更加困难,尤其是如果 Sam Altman 暗示:“嘿,也许我们在领先计算水平上开始有点过度了。”

Coreweave, on the other hand, has $5.5 billion in cash against $18 billion in invoices. They have $3 of short-term debt for every $1 in cash they have . This is not good. This slows expansion and makes funding more difficult, especially if Sam Altman suggests: "Hey, maybe we're starting to overdo it a bit with the level of leading-edge computing."

这个频道对 $CRWV 的历次判断

Meet Kevin 在这只股票上共 3 条判断,这里只列相邻的几条。

2026-09-02看跌本条
Sam Altman 关于 Neo 云不可持续的观点是否正确?分析这一点的最佳方式是查看 CoreWeave 的财务数据,我们稍后将进行。
方向翻转
2026-08-27看涨
现在 Cerebrris 一直严重高估。我在 IPO 时就称它严重高估。它已经跌下来了,不久前跌到了 200 多美元的低到中位区间,然后又跌得更低了。
原话 2:51:54 ›
方向翻转
看完整历史 ›
美股听风