NVDA的1500亿美元回购规模相对于其估值与自由现金流而言并不充分，表明公司缺乏能够打动投资者的新催化剂。
逐段跳播
他们今天发布了两份声明，一份关于人工智能的安全性，而对投资者来说，或许最重要的是董事会宣布了一项1500亿美元的股票回购协议。
They issued two statements today, one about the safety of artificial intelligence, and for investors, perhaps the most important was the board's announcement of a $150 billion share buyback agreement.
正如新闻稿所述，这是历史上规模最大的一次，使董事会目前拥有约2500亿美元用于回购股票。我认为这太小了。我的意思是，你看。他们终于取消了每股一美分的股息，这一点我也一直不理解。
但没错，这里的“大数定律”似乎脱离了背景，没有什么看起来足够大。没错，1500亿，无聊。
As the press release states, it is the largest in history, giving the board approximately $250 billion to repurchase shares currently. I think it's too small. I mean , look. They finally got rid of the one-cent-per-share dividend , which I also didn't understand.
But yes, the law of big numbers here seems out of context, and nothing seems big enough. Yes, 150 billion, boring.
是的，我的意思是这是历史上规模最大的一次，但它只占其市值的约2%。因此，你需要根据我们所谈论的这家公司来调整。我的意思是，换个角度看，Nvidia正在通过建设人工智能基础设施产生巨大的现金流。
他们拥有的资本超过了他们在研发上能够从物流上再投资的规模。所以我认为这就是我们现在看到的情况。
Yes, I mean it's the largest in history, but it only represents about 2% of its market value. Therefore, you need to adjust that to suit the company we are talking about. I mean, another way to look at it is that Nvidia is generating huge cash flows from building AI infrastructure.
They have more capital than they can logistically reinvest in research and development. So I think this is what we are seeing happening now.
我想说，这个公告的规模至少应该是这个数额的三倍。我知道这听起来很疯狂，可能会让我上CNBC，但你知道，我并不是这里的技术专家。
I would say that the advertisement should have been at least three times this amount. I know this sounds crazy and might put me on CNBC, but you know, I'm not the tech expert here.
好，好，你看，这么说吧。我们可以像“CNBC”分析那样给出一个观点，但我想说：首先，在它的估值和当前的债务状况之间——正如我们所说，债务仅占其市值的2%——这里有机会达到5%的比例。
我甚至都跟不上，它每天变动上千亿美元，所以我已经跟不上这个点了，但作为一家估值超过5万亿美元的公司，这个数额至少应该达到其市值的5%的水平。如果你想打动投资者，就不要害怕使用一点杠杆。
这是一家已经产生了1340亿美元自由现金流的公司。所以，你相当于增加了一年的自由现金流。来吧。让我开开眼，Nvidia。
Okay, okay, look , let's put it this way. We could offer an opinion like a "CNBC" analysis, but I would say; First, between its valuation and its current debt position , which, as we said, is only 2% of its market value, there is an opportunity here for a 5% ratio.
I can't even keep track of it, it's changing by a hundred billion dollars a day, so I can't follow the point anymore, but as a company valued at more than 5 trillion, this amount should be at least at the level of 5% of its market value.
If you're trying to impress investors, don't be afraid to use a little leverage. This is a company that has generated $134 billion in free cash flow. So, you have added the equivalent of one year's free cash flow . Let's go. Impress me, Nvidia.
所以，我在这里提出一个大胆的观点，但我认为他们宣布这件事的原因之一——我想知道你对这个具体观点怎么看——是除了股票回购之外，还有一份与人工智能安全相关的新闻稿，或许是为了建立一些屏障或安全空间，这样我们就不会看到诸如“天哪，人工智能可能入侵了Hugging Face”这样的新闻。
天哪，人工智能可能入侵了澳大利亚政府的健康网站。所有这些新闻开始出现。也许这1500亿美元是一种让投资者更容易接受的方式。
So, I'm making a bold point here, but one of the reasons I think they announced this—and I want to know what you think about this point specifically—is that in addition to the share buyback, there was a press release related to AI security and perhaps building some barriers or safe spaces so that we don't have news like, "Oh my god, AI may have hacked Hugging Face."
Oh my God, artificial intelligence may have hacked into the Australian government's health websites. All these news stories began to appear. Perhaps this $150 billion was a way to make it easier for investors to accept.
是的，我认为这肯定是其中一部分。我认为这个时机绝非巧合。我认为这是由高层管理团队精心协调的。但这是这个行业的一个真实问题。正如我提到的，我们看到来自各个实验室的复杂人工智能模型，以及定期突破安全防护环境的大公司。
显然有著名的Hugging Face事件。但我们也正处于这样一个时刻：CEO黄仁勋非常明确地表达了他想在这里采取的方式。他不想等待政府监管。事实上，他反对监管。他准确地将人工智能安全界定为一个可以在硬件和平台层面管理的工程问题。
Yes, I think that's definitely part of it. I don't think the timing was a coincidence in any way . I think it was very well coordinated by senior management. But this is a real problem in the industry.
We see sophisticated AI models, as I mentioned, from various labs, as well as large companies that regularly breach secure protection environments. There is obviously the famous Hugging Face incident.
But we are also at a time when CEO Jensen Huang has been very clear about the approach he wants to take here. He does not want to wait for government regulations. In fact, he opposed it.
He accurately framed AI security as an engineering problem that can be managed at the hardware and platform levels.
稍微谈谈他们做了什么，Nvidia提供了一个开源框架。它叫Open Shell。他们还推出了一款名为“Centry”的设备监控工具，其基本思路是这些工具将帮助在几分之一秒内隔离恶意软件。
Open Shell框架很有意思。正如其名，它是开源的。它旨在支持来自ARM和Intel的竞争平台，所以你可以看到Nvidia正试图为安全设定一个全球行业标准。他们想在我们看到更多企业安全团队或监管机构叫停独立人工智能代理的部署之前，降低机构采用的风险门槛。所以我认为他们确实在试图抢占先机。
To talk a little about what they do, Nvidia offers an open-source framework. It is called Open Shell. They also introduced a device monitoring tool called " Centry," the basic idea of which is that these tools will help isolate malware in fractions of a second.
The Open Shell framework is interesting. It is open source, as its name suggests. It is designed to support competing platforms from ARM and Intel, so you can see how Nvidia is trying to set a global industry standard for security.
They want to lower the risk barrier to institutional adoption before we see more corporate security teams or regulators halting the deployment of independent AI agents. So I think they're really trying to get ahead of things here.
回到投资的问题。是的，我认为他们把坏消息包装在好消息里，但你看，这是当前趋势的一部分。Nvidia已经没有能打动市场的办法了。黄仁勋越来越像最伟大的表演家。你知道他现在多频繁地上台只是为了发表大胆的声明。
自春季以来，这只股票几乎毫无表现。令人印象深刻的财报被相当冷淡地接受。我认为这是一家很棒的企业，但当前的主流看法——无论是由于对人工智能的悲观情绪，还是仅仅因为投资者厌倦了——正是他们试图对抗的。
Nvidia已经没有能打动或激发投资者的办法了，我认为这个大规模回购计划就是证明。这可能只是又一次失败的尝试，想说：“你好，你还记得我们吗？我们仍然真的很棒。”
Returning to the question of investment. Yes, I think they wrap bad news in good news, but look, this is part of a prevailing trend. Nvidia has run out of ways to impress the market.
Jensen increasingly looks like the greatest showman. You know how often he goes on stage now just to make bold statements. The stock has achieved virtually nothing since the spring.
The reports of impressive earnings are received rather coldly. I think this is a great business , but the prevailing perception, whether due to pessimism about artificial intelligence or simply investor boredom, is what they are trying to counter.
Nvidia has run out of ways to impress or excite investors, and I think the massive buyback program is proof of that. This might just be another failed attempt to say, "Hello, do you remember us?" We're still really good.
是的，好吧，你猜怎么着？如果你是首席财务官，而且你真的想打动投资者，那就回购价值5000亿美元的股票。这对他们来说是个很好的卖点。
Yes, okay, guess what? If you're the chief financial officer and you really want to impress investors, buy back $500 billion worth of shares. This is a good selling point for them.
好，但此时此刻，Nvidia的股价约为其明年盈利预期的18倍。这是十年来最便宜的水平。而我说，或者这也许是我喜欢的一个小评论，市场是非理性的，但并不愚蠢。所以这里正在积累一些我们可能看不到的风险。
Okay, but at this point, Nvidia's stock is trading at about 18 times its earnings estimate for next year. It's the cheapest it's been in a decade. And I say, or perhaps it's a small comment I like, that the market is irrational but not stupid .
So there are some risks accumulating here that we may not see.
由于这一过程的周期性，投资者可能想考虑哪些隐藏的、大多数人没有想过的风险？当然，这个行业存在周期。但有哪些事情是人们可能不太关注的？
What risks might investors want to consider that are hidden and not thought about by most people due to the cyclical nature of the process, of course? There is a cycle in this industry.
But what are some things that people probably don't pay much attention to?
我认为有一些事情需要考虑。我的意思是，其中之一是一种循环融资。这不是秘密。这是已经被大量讨论过的事情。我的意思是，我们看看Nvidia今天刚刚公布的那个数字。他们披露，在Nvidia支持的云计算实体中，与Anthropic的合同总价值仅这一项就已超过1800亿美元。
这仅仅是他们与“Anthropic”的安排。Nvidia的资产负债表包括13项公开股权、200多项私募股权，以及200亿美元的数据中心租赁负债。它们主要发挥金融中心的作用，而这是必要性的产物。
其思路显然是要回到Nvidia的增长故事。但存在那种循环融资机制，我认为这令人担忧，取决于扩张是否会放缓或何时放缓。
I think there are some things that need to be taken into consideration. I mean, one of them is a type of circular financing. This is not a secret. This is something that has been talked about a lot.
I mean, we're looking at that number that Nvidia just announced today. They revealed that the total value of Anthropic contracts across Nvidia-backed cloud computing entities has exceeded $180 billion alone.
This is only in their arrangements with "Anthropic". Nvidia’s balance sheet includes 13 public equity stakes, more than 200 private equity stakes, and $20 billion in data center lease liabilities.
They primarily function as a financial center , and that is a function of necessity. The idea is clearly that this goes back to Nvidia's growth story. But there is that circular financing mechanism which I find worrying depending on whether or when the expansion will slow down.
另一点需要注意的是，大公司正以3到5年的快速生命周期消耗这些人工智能图形处理器（GPU）。它们基于这样一种假设运作：将有持续的价值数十亿美元的升级。但如果你有一块用了几年、最终完全足以应对你80%或更多企业推理工作负载的图形处理器呢？
这个升级周期可能会慢得多。如果发生这种情况，像Nvidia这样的公司可能会经历急剧下滑，或至少是增长放缓，即使那是在遥远的未来。我的意思是，基本上，这个想法并不意味着这项技术会失败。
如果当前的设备运行效率非常高，Nvidia可能会发现自己处于自身升级周期之外。这是一个极端情况，但我认为这是Nvidia投资者现在评估投资时应该考虑的事情。
Another thing to note is that large companies are wearing out these AI graphics processing units (GPUs) at rapid lifecycles of 3 to 5 years. They are operating on the assumption that there will be ongoing upgrades worth billions of dollars.
But what if you have a graphics processing unit that is a few years old and ends up being perfectly adequate for 80% or more of your enterprise inference workload? This upgrade cycle can be much slower.
If that happens, companies like Nvidia could experience a sharp decline or at least a slowdown in growth, even if that is in the distant future. I mean , basically, this idea doesn't mean that the technology will fail.
If current devices are operating at very high efficiency, Nvidia may find itself outside its own upgrade cycle. This is an extreme case, but I think it's something Nvidia investors should take into account when evaluating their investments right now.
我认为存在一种合乎逻辑、并不极端的观点，即人工智能周期最好的日子已经过去。这并不是说它在一段时间内不会表现良好。但Google、Amazon、Microsoft、Meta和OpenAI都至少在尝试制造自己的芯片。
没有人会停止与Nvidia做生意，但华尔街关心的是增长。如果竞争加剧，更不用说还有其他芯片公司，这很可能会导致增长触及天花板。另外，如果先进模型不再那么重要——而且这方面的证据越来越多——那么对最新、最昂贵芯片的需求是否还会以人们所期望的规模存在？这听起来可能很无聊，但实际上很重要。
I think there is a logical, not extreme, viewpoint that says the best days of the artificial intelligence cycle are over . Not in the sense that it won't be good for a period of time.
But Google, Amazon, Microsoft, Meta, and OpenAI are all at least trying to manufacture their own chips. No one will stop dealing with Nvidia, but Wall Street is interested in growth.
And if there is increasing competition, not to mention other chip companies , this will probably lead to a ceiling on upward growth. Also, if advanced models are no longer of great importance, and there is increasing evidence of this, will the need for the latest and most expensive chips still exist at the same hoped-for scale?
It may seem boring, but it is actually important.
估值是不可避免的现实。2023年，Nvidia的每股收益每年不到一美元。而现在已达到八美元。所以，按市盈率来看它可能显得便宜，但问题是：这样的每股收益水平能维持多久？如果这种情况不能持续，市盈率看起来就会大不相同。
Patrol is an unavoidable reality . In 2023, Nvidia's earnings per share were less than one dollar per year. And now it's reached eight dollars. So, it may look cheap based on the price-to-earnings ratio, but the question is: how long can this level of earnings per share be sustained?
If this does not continue , the price-to-earnings ratio will look very different.
我想知道，是什么在推动那样的每股收益？回购价值5000亿美元的股票。我会试着自己解开这个难题。
I wonder, what drives that earnings per share? Repurchasing shares worth $500 billion. I will try to solve this dilemma myself.
这个频道对 $NVDA 的历次判断
The Motley Fool 在这只股票上只有这一条判断。