$NVDA · Value Investing with Sven Carlin, Ph.D. · 2026-11-13
说话者认为,NVDA 的低市盈率具有欺骗性(前瞻市盈率 18;按 FY2029 每股收益 21 美元计算仅 7.8):涨价与竞价抢购推高了营收和利润率,而在利润率见顶时利润率会回落(约 5000 亿美元营收、2000 亿美元利润、市盈率 20)。当 AI 热情消退,利润和股价应当下跌;在需求毁灭情形下,2029-2030 年每股收益可能为 10,750 亿乘以 10 等于 7500 亿美元,比今天的 4 万亿美元低 80%。现在买入如同 1999 年/加密货币式的赌博;应等待真正的毁灭、破产和血流成河。Jensen Huang 已卖出 19 亿美元,所以要关注他的实际行动。
播放时,YouTube 可能接收设备与播放信息并使用 Cookie。 隐私说明 · Google 隐私政策
逐段跳播
Nvidia 的股价正在下跌,而你们中的许多人昨天买入了它。这一点在开盘后个人投资者掀起的买入潮中清晰可见,股价从150涨到了166。
Nvidia's stock is falling, and many of you bought it yesterday. This is clearly evident in the wave of buying by individual investors that began at the market opening, with the stock rising from 150 to 166.
有人报告说个人买入非常狂热,在开盘后的前40分钟就创下了纪录水平,我们见证了追踪 Nvidia 的 ETF 和杠杆基金——包括正向和反向基金——历史上第五大的三日现金流入。
Some reported that the individual buying was frenzied, as it set record levels in the first 40 minutes of the market opening, and we witnessed the fifth largest three-day cash inflow in the history of ETFs and leveraged funds that follow Nvidia, in both positive and inverse funds.
所以没有净卖出,也没有人想做空,因为随着 Nvidia 的远期市盈率降至历史低点,大家都在买入。评论区里也有很多问题:Sven,分析一下 Nvidia,它非常便宜。即使它的增长放缓,它仍然非常便宜。
So there is no net selling, and no one wants to short sell, as everyone is buying with Nvidia's forward price-to-earnings multiple falling to a record low. Also in the comments, there are many questions: Sven, analyze Nvidia, it's very cheap.
Even if its growth slows down, it is still very cheap.
是的,如果你看市盈率,远期市盈率是18,对于一家预计盈利将增长两倍的公司来说,这简直难以置信,我不知道,是在未来两个财年。
Yes, if you look at the price-to-earnings ratio, the forward ratio is 18, which is incredible for a company that is expected to triple its earnings, I don't know , in the next two fiscal years.
这是华尔街的共识,从当前水平到2029财年每股收益21美元。如果把它算成市盈率,我们得到7.8倍。大多数华尔街分析师都极度乐观。我们有57个强烈买入评级,只有四个持有评级,还有一个可怜的家伙给出了强烈卖出评级,他是唯一一个说对了的人。
This is the Wall Street consensus from the current level to earnings per share of $21 for fiscal year 2029. If you calculate that as a price-to-earnings ratio, we get a ratio of 7.8.
Most Wall Street analysts are extremely optimistic. We have 57 strong buy ratings, only four hold ratings, and one poor soul with a strong sell rating, who was the only one who was right .
有趣的是,Seaport 的 Jay Goldberg 对该公司给出了强烈卖出评级,他本想在 Jensen 那件事发生前几天上调评级。当然,他没有。
The funny thing is that Jay Goldberg of Seaport, who has a strong sell rating for the company, wanted to raise the rating just a few days before everything that happened with Jensen happened. Of course, he didn't .
所以,我们现在需要讨论 Nvidia 是否是一笔收购交易。如果你看华尔街分析师的目标价,他们仍然保持得很高。现在下调评级将开始出现,但华尔街一直如此。评级下调只在股价下跌时才会发生。
So, we now need to discuss whether Nvidia is a buyout deal. If you look at the targets of Wall Street analysts, they are still keeping them very high. Now the downgrades will start to appear, but that's always been the case on Wall Street.
Rating downgrades only occur when the stock price falls.
但你说,Sven,是的,下调华尔街评级,这总是很疯狂。如果我们能以七倍市盈率买入 Nvidia,那可是一笔很划算的买卖。
But you say, Sven, yes, downgrading Wall Street ratings, that's always crazy. If we can buy Nvidia at a price-to-earnings ratio of seven, that's a great buy.
我们七个月前讨论过 Nvidia,当时我听了几个小时 Jensen Huang 的演讲。当时预计到2028财年营收将达到1万亿美元。所有数字都是目标:50%的净利润率和5000亿美元的净利润。
鉴于当前市值,就是这样,市盈率七到八倍。这与当前华尔街的预期一致。
We discussed Nvidia seven months ago when I listened for hours to Jensen Huang's presentation . At the time, revenues of $1 trillion were expected by fiscal year 2028. All numbers were targeted: 50% net profit and $500 billion net income.
Given the current market value, that's it, a price- to-earnings ratio of seven or eight. This is in line with current Wall Street estimates.
在这里你可以看到 Jensen 只是在预测 Nvidia、Vera Rubin 等将会发生什么。但说到市盈率,七倍市盈率包含许多组成部分。有价格,有利润,还有其他东西。首先,我们应该讨论的是价格,售价,而不是股价。我们甚至也会谈到这一点。
Here you see Jensen simply predicting what will happen with Nvidia, Vera Rubin, and so on. But when it comes to price-to-earnings ratios, the price-to-earnings ratio seven has many components.
There's the price , there's the profit, and there are other things. First , what we should be discussing is the price, the selling price, not the share price. We will even touch on that as well .
如果你看过去几年,许多半导体的价格从200涨到了1000。我从 Gamers Nexus 得到这个信息。如果价格上涨五倍,营收也会以同样的幅度爆发。但如果发生经济衰退,这对半导体也适用,所有这些价格都会大幅下跌。
If you look at the last few years , the prices of many semiconductors have increased from 200 to 1000. I got this information from Gamers Nexus. If prices increase fivefold, revenues will explode by the same amount.
But in the event of a recession, and this applies to semiconductors, all those prices fall sharply.
当然,存储商品受到了严重损害,这解释了存储半导体的四倍市盈率。而其他价格的下降幅度稍小一些。然后我们有 CPU 和 GPU,它们可以在减少产量的同时维持价格。但我们不知道其人工智能组件是否会保持同样的韧性。
Of course, the memory of goods has been severely damaged, and this explains the four price-to-earnings ratios for memory semiconductors. While other prices decrease to a slightly lesser degree.
Then we have CPUs and GPUs that can maintain prices while reducing production. But we don't know if its artificial intelligence component will remain as resilient.
因为 Nvidia 在过去几年大幅提高了价格。在这里你可以更清楚地看到这一点,因为公司之间的竞价战。如果那些竞价战消退,那些战争本会在22到26年间将 Nvidia 的年营收从260亿美元推高到2150亿美元。这非常重要。
Because Nvidia has significantly raised its prices over the past few years. You can see this more clearly here because of the bidding wars between companies. If those bidding wars were to subside, those wars would have pushed Nvidia's annual revenue from $26 billion to $215 billion between $22 and $26. That's very important.
然后你看毛利润从60%到75%,净利润从30%到60%。这解释了 Nvidia 营收的巨大增长。但在某个时候,这会达到顶峰。当没有增长时,假设营收达到5000亿,利润率会下降,因为价格上不会有竞争。
这留下2000亿的利润,这意味着20倍的市盈率。
Then you look at the gross profit from 60% to 75 %, and the net profit from 30% to 60%. This explains the huge increase in Nvidia's revenues. But at a certain point, this will reach its peak.
When there is no growth, let's say revenues reach 500 billion, margins will decrease because there will be no competition on prices. This leaves 200 billion in profits, which means a price-to-earnings ratio of 20.
所以,等到这股热情在某个时候消退,你就可以预期会出现下跌。当利润大幅下滑时,股价也会跟着下跌,此外还有一大堆其他因素。
So, at some point when this enthusiasm wanes, you can expect a decline. When profits fall significantly, the price follows, and there are a host of other things.
我们现在已经看到了 Nvidia 的这桩丑闻,还有新任 CFO 接任 CEO,但这是可以接受的,也许没那么重要。
We have now seen the scandal with Nvidia, and the new CFO taking over as CEO, but this is acceptable, and perhaps less important.
最重要的是,自2020年以来,数据中心开发商至少筹集了1.8万亿美元,尤其是最近几年。尤其是最近几年,这与 Nvidia 的销售额存在相关性。然而,这不可能永远持续下去。
Most importantly, data center developers have raised at least $1.8 trillion since 2020, especially in the last few years. Especially in the last few years, there is a correlation with Nvidia's sales. However, this cannot continue forever.
目前,这没有成本。利息成本很低,折旧和摊销很低,再融资要到2030年才会到来。但市场会在2028-2029年开始考虑这件事。市场通常就是这样思考的。
Currently , there is no cost. Interest costs are low, depreciation and amortization are low, and refinancing will not come until 2030. But the market will start thinking about it in 2028-2029. This is usually how the market thinks.
然后 Nvidia 会迫使供应商将产量翻倍、再翻倍,增长、再增长。这意味着,恰好在芯片需求不再那么高的时候,会有大量芯片涌入市场。
Then Nvidia would force suppliers to double production, double that, double growth, and more growth. This means there will be a flood of chips entering the market exactly when they are not in such high demand.
TMC 将不得不在未来十年内将产能翻倍,诸如此类的说法。但这是最近才发生的。Google 的 AI 芯片、中国在做这做那,消息不断传来,但所有人都无视它。但华尔街分析师没有更新。尽管市场已经下跌了40%。
TMC will have to double its capacity over the next decade, stories like this. But this happened very recently. Google's AI chips , China doing this and that, news that keeps coming in but everyone ignores it.
But Wall Street analysts did not update. Although the market fell by 40%.
是的,这里面有一点黄仁勋的操纵,但也许他也能看到市场,而不是市场。市场看不到聪明资金所看到的东西。好吧,Nvidia 快要见顶了,我们开始卖吧。那么是谁在买入、让大笔资金得以卖出?
是小投资者吗?因为小投资者喜欢赌博,看它多便宜,他们就开始买入,正如我们所见。所以有人卖出,小投资者进来买入。
Yes, there is a little bit of Jensen Huang's manipulation, but perhaps he also sees the market, not the market. The market doesn't see what smart money sees. Well, Nvidia is about to peak, let's start selling.
And who is buying and allowing large sums of money to sell? Small investors? Because small investors love to gamble, look how cheap it is, and they start buying, as we have seen.
So there are those who sell and small investors come in and buy.
不管这与黄仁勋有没有关系,黄仁勋建立起来的这轮全球大规模融资很可能会减少,这意味着 Nvidia 的收入会减少。1500亿美元用于支付 OpenAI 的租赁费用,诸如此类。
Whether or not there is a connection to Huang, this massive round-the-world funding that Jensen Huang built is likely to decline, meaning less revenue for Nvidia. $150 billion to cover OpenAI's lease payments, and things like that .
所以是的,Nvidia 的每股收益在2028年可能达到20,但在2029-2030年也可能只有10。10乘以10,市盈率等于100股。那仍然是两万亿。这仍然非常正面。
So yes, Nvidia’s earnings per share could reach 20 in 2028, but it could also be only 10 in 2029-2030. 10 times 10 is a price-to-earnings ratio equals 100 shares . It's still two trillion. This is still very positive.
但假设它会演变成真正的需求毁灭,因为当一切都过剩时,一切都会崩塌。价格在下跌,利润率在下降。什么都没有了。什么都不剩了。半导体周期就是这样运作的。假设什么都没有进来,750亿乘以10,等于7500亿。这仍比大家急于买入的当前四万亿低80%。
But let's say it will come down to a real destruction of demand, because when there is too much of everything, everything collapses. Prices are falling , and margins are falling. nothing.
There is nothing left . This is how the semiconductor cycle works . Let's assume nothing comes in, 75 billion times 10, that equals 750 billion. This is still 80% lower than the current four trillion that everyone is eager to buy.
如果你现在买入 Nvidia,我对你有一个小小的请求。当这一切结束时,当真的血流成河时,那还不算什么。这还不是痛苦。你需要看到市盈率被摧毁、利润被摧毁、其他东西被摧毁、需求被摧毁。
你需要看到破产案例。然后,当所有人都亏光了钱,再考虑,好吧,我该怎么办?那才是投资的正确时机。
I have one small request for you if you buy Nvidia now. When this is over , when there is really blood in the streets, then that is nothing yet . This is not pain. You need to see destruction in the price-to-earnings ratio, destruction in profits, destruction in other things, and destruction in demand.
You need to see bankruptcy cases. Then when everyone loses their money, consider , well, what should I do? That's the right time to invest.
因为如果你现在买入 Nvidia 股票,你就是在赌博,就像1999年那样,就像加密货币所经历的那样。是的,她卖的是淘金设备。是的,市盈率很低。是的,它看起来很便宜,但这只意味着它是最后一张倒下的多米诺骨牌。
Because if you buy Nvidia shares now, you are gambling just like in 1999, and just like what happened with cryptocurrencies . Yes, she sells gold prospecting equipment . Yes, the price-to-earnings ratio is low.
Yes, it looks cheap, but that just means it's the last domino to fall.
总之,一只市盈率为七倍的股票,可能成为你买过的最贵的股票,因为如果有人了解它的经济周期,那就是这个人:他在2016年到2020年间没有卖出任何 Nvidia 股票。然后在2021年科技股拆股后的一段高速增长期,他卖出了两百万股,而在2022年低迷期间什么都没卖。
很好。接着他在2023年卖了一些,然后在 AI 炒作的高峰期卖出了1200万股。他总共卖出了价值19亿美元的 Nvidia 股票。他是谁?黄仁勋。
In conclusion, a stock with a price-to- earnings ratio of seven might be the most expensive stock you ever buy, because if anyone knows its economic cycles, this guy didn't sell any Nvidia shares between 2016 and 2020.
Then he sold two million shares during a period of intense growth following the tech stock split in 2021, and didn't sell anything in 2022 during the downturn. Good. He then sold a few in 2023, and then sold 12 million shares at the height of the AI hype.
In total, he sold $1.9 billion worth of Nvidia shares. Who is he? Jensen Huang.
我总是说,不要听沃伦·巴菲特说什么,要看沃伦·巴菲特做什么。不要听黄仁勋说什么,要看黄仁勋做什么。你需要知道的就这些。
I always say don't listen to what Warren Buffett says, watch what Warren Buffett does. Don't listen to what Jensen Huang says, watch what Jensen Huang does . That's all you need to know.
这个频道对 $NVDA 的历次判断
Value Investing with Sven Carlin, Ph.D. 在这只股票上共 4 条判断,这里只列相邻的几条。