NVDA 是首选买入标的；2350 亿美元回购缩减股本，一体化 AI 技术栈与供应主导地位驱动增长，14.6 倍远期市盈率相较标普 500 平均水平具备上行空间。
逐段跳播
我原以为自己对比亚迪的股票已经过于乐观了。我把它评为今年最值得买入的股票。我的投资组合中持有英伟达股票。它是我最大的金融中心之一，我正准备买入更多。
I thought I was being overly optimistic about Nvidia's stock. I have ranked it as the best stock you can buy this year. I own Nvidia stock in my investment portfolio. It is one of my largest financial centers, and I am preparing to buy more of it.
所以，我原以为自己是英伟达股票最乐观的投资者之一。但今天，9月28日，英伟达管理层宣布了额外1500亿美元的股票回购授权，这意味着他们将在此金额之外，再加上剩余的850亿美元授权，购买更多英伟达股票。
So, I thought I was one of the most optimistic investors in Nvidia stock. But today, on September 28, Nvidia's management team announced an additional $150 billion share buyback mandate, meaning they will purchase additional Nvidia shares for that amount, in addition to their remaining $85 billion mandate.
因此，从现在到下一财年结束，英伟达计划回购超过2300亿美元的自家股票。
Thus, from now until the end of the next fiscal year, Nvidia plans to buy back more than $230 billion of its own shares.
所以，我道歉，英伟达管理层显然是对英伟达股票最乐观的人，我不怪他们。在本视频中，我将重点介绍一些让我对英伟达股票非常乐观的原因，为什么我给它很高的评级，以及为什么我认为该股仍有很大的上涨空间，尽管它是一家市值超过5万亿美元的公司。
So, I apologize, the Nvidia management team is clearly the most optimistic about Nvidia's stock, and I don't blame them for that. In this video, I will highlight some of the reasons that made me very optimistic about Nvidia stock, why I ranked it highly, and why I believe the stock still has a lot of room to rise, even though it is a company with a market value exceeding $5 trillion.
你可以在这里看到英伟达9月28日的这项授权公告。这项授权，也就是这个回购计划，比他们之前所做的要大好几倍。例如，在2026财年，他们回购了价值400亿美元的股票。在2027财年至今，他们已经回购了价值530亿美元的股票，比上一年多了130亿美元，而且他们还有时间买入更多。
You can see this authorization announcement here on September 28 from Nvidia. This mandate, i.e., this buyback program, is several times larger than what they have done previously.
For example, in fiscal year 2026, they repurchased $40 billion worth of shares. So far in fiscal year 2027, they have repurchased $53 billion worth of shares, $13 billion more than the previous year, and they still have time to buy more.
他们新的1500亿美元授权将加到现有的850亿美元之上，从现在到明年总计达到2350亿美元。
Their new mandate of $150 billion will be added to the existing $85 billion, bringing the total to $235 billion from now until next year.
如果你一直在关注我的视频，这是我期待并预测了好几个月的事情。因为我看到的许多负面方面，或者说，我在评论区、邮件以及与投资者的讨论中看到的许多担忧是，该公司的市值已经达到5.4万亿美元。
该股还能涨多少？它没有多少上涨空间，因为它的市值已经非常高了。
This is something I have been expecting and predicting for many months, if you have been following my videos. Because many of the negative aspects that I see, or rather, many of the concerns that I see from investors in the comments section, emails, and discussions I have with them, are that the company’s market value has already reached $5.4 trillion.
How much higher can the stock rise? It doesn't have much room to rise because its market value is already very high.
针对这一论点，我提出，如果你想想市值的公式，它等于市场价格乘以流通股数量。市场价格乘以流通股数量。如果英伟达每年花费数千亿美元回购股票，这就减少了流通股数量。因此，市场价格可以上涨，而流通股数量下降，整体市值可以继续上升，因为流通股数量在未来五年或十年内将继续下降，如果我的估计接近我预期会发生的情况，也就是从现在到未来五年或十年间，为这家公司产生数千亿美元的自由现金流。
In response to this argument, I suggested that if you think about the equation for market value, it is equal to the market price multiplied by the number of outstanding shares.
Market price multiplied by the number of outstanding shares. If Nvidia spends hundreds of billions of dollars each year to buy back shares, this reduces the number of outstanding shares.
Therefore, the market price could rise while the number of outstanding shares falls, and the overall market capitalization could continue to rise because the number of outstanding shares will continue to fall over the next five or ten years, if my estimates are close to what I expect to happen, which is generating hundreds of billions of dollars in free cash flow for this company between now and the next five or ten years.
所以，推动英伟达在数据中心市场前进的因素之一，是其AI工厂的一体化产品。它不只是销售图形处理器。这是它的旗舰产品。这是被宣传的主要产品，几乎每个投资者都知道。但它还有“CODA”编程平台，这是伴随图形处理器的软件系统。
它还销售网络设备，将所有这些图形处理器连接成一个大集群。它还有中央处理器。它还有集成电路，所有这些协同工作，提供一个买家可以即用的一体化系统。
So, one of the things that has propelled Nvidia forward in the data center market is its integrated offering of AI factories. It doesn't just sell graphics processing units. This is its flagship product.
This is the main product that is advertised and that almost every investor knows about. But it also has the "CODA" programming platform, which is the software system that accompanies the graphics processing unit.
It also sells networking equipment that connects all these graphics processing units into one large cluster. It also has central processing units. It also has integrated circuits, and all of that works together to deliver a fully integrated system that the buyer can get ready to use.
他们不需要从博通寻找连接器，或从Marvell或其他公司寻找组件，并试图把所有这些拼凑起来建设数据中心。他们可以直接去找英伟达说：“把一切都给我们。”“给我们每样东西10个或50个。”
“他们在这个日期把它交给了我们。”而英伟达有能力处理所有这些。这对这项技术的许多买家都是有利的。
They don't need to search for connectors from Broadcom or components from Marvell or other companies and try to put all of that together to build a data center. They could simply go to Nvidia and say, "Give us everything."
"Give us 10 of everything or 50 of everything." "They handed it over to us on this date." And Nvidia is capable of handling all of that. This is beneficial for many buyers of this technology.
作为一名企业家，我知道你们中很多人是关注这个频道的企业主，你可以想象，你要与许多不同的供应商打交道，还有许多项目和属于你职责范围的事情。如果你能至少整合业务的一部分，那就是一个有益的激励，尤其是如果那个供应商非常擅长他们所做的。
他们以足够合理的价格为你提供性能优于竞争对手的产品，并带来强大的总拥有成本，比你亲自连接、构建和定制所有这些不同组件更好。
As an entrepreneur, and I know many of you are business owners who follow this channel, you can imagine that you are dealing with many different suppliers and many projects and things that fall under your responsibility.
If you can consolidate at least one part of your business, that is a helpful incentive to do so, especially if that provider is very good at what they do. They offer you a product that outperforms its competitors at a reasonable enough price, and delivers a strong total cost of ownership that is better than if you were to connect, build, and customize all these different components yourself.
现在，不要误解我。有些公司仍然那样做，对吧？尤其是像Alphabet、Amazon和Microsoft这样的大公司，因为这构成了他们业务的很大一部分。他们花费数千亿美元来建设这些数据中心。
这是他们想要擅长的事情。如果他们能找到一种方法，让数据中心的性能提升3%，而他们必须花更多时间去做，他们也愿意，因为这是他们业务的一大部分。这是他们竞争优势的一大部分。
Now, don't misunderstand me. Some companies still do that, right? Especially large companies like Alphabet, Amazon and Microsoft, because this makes up a large part of their business.
They are spending hundreds of billions of dollars to build these data centers. This is something they want to excel at. If they can find a way to deliver 3% better performance in their data center and they have to spend more time doing it, they are willing to do it because it is a big part of their business. It's a big part of their competitive advantage.
然而，并非所有公司都想那样做。这主要集中在宽带公司，如Amazon、Microsoft、Meta和Alphabet。大多数其他构建AI的公司，甚至大多数构建AI的新兴云公司，都不想卷入所有这些工作。
他们想专注于他们认为更具战略重要性的其他业务领域。因此，他们更偏好英伟达的这种一体化产品。
However, not all companies want to do that. This is mainly concentrated among broadband companies, such as Amazon, Microsoft, Meta, and Alphabet. Most other companies that build AI, and even most emerging clouds that build AI, do not want to get involved in all of this work.
They want to focus on other areas of their business that they see as more strategically important. Therefore, they prefer this integrated offering from Nvidia.
近年来，还有一个额外的关键因素变得对英伟达有利。因为英伟达的业务在过去12个月的收入方面，以及在供应商、供应协议和购买组件与服务上花费的金额方面都如此巨大，它现在拥有更强的议价能力。
One additional key factor has changed in Nvidia's favor in recent years. Because Nvidia’s business is so huge in terms of revenue over the past 12 months, and in terms of the amount of money it spends on suppliers, supply agreements, and the purchase of components and services, it now has greater bargaining power.
由于英伟达的影响力、规模和重要性的提升，它现在在与台积电、SK海力士、三星和美光等供应商的关系中处于更有利的地位。由于这种规模因素，它可以获得更多的供应、更多的产能和更好的条款，这一点自2023年以来确实发生了变化，也许2024年就是随着其业务增长，形势将变得对英伟达有利的一年。
Because of Nvidia’s reach, increased size and importance, it is now in a stronger position with suppliers such as Taiwan Semiconductor Manufacturing Company, SK Hynix, Samsung and Micron.
It can get more supplies, more capacity and better terms because of this size factor, which has really changed since 2023, and perhaps 2024 is when the equation will change in Nvidia’s favor as its business grows.
因此，他们与几家供应商建立了合作，锁定了大量库存以满足这一需求。他们的负债在上个季度从1190亿上升到2790亿。所以，他们锁定了充足的供应。
Therefore, they have partnered with several of their suppliers and secured large quantities of stock in order to meet this demand. Their liabilities rose from 119 billion to 279 billion in the last quarter. So, they secured plenty of supplies.
要记住，这个行业受制于供应。需求超过供应，而这些公司中许多公司的近期收入将取决于它们能锁定多少供应。你会开始从许多公司那里听到这一点，而且你已经听到许多公司说我们正面临供应约束。
如果我们有更多供应，本可以创造更多收入。随着Nvidia拿下这些供应，它巩固了自己的地位，而其他所有人都必须等到Nvidia“吃饱”之后才能分到自己的那一份。这是我们在展望未来两年时的一个重要因素。
Remember that this industry is constrained by supply. There is demand that exceeds supply, and near-term revenue generation for many of these companies will depend on how much supply they can secure.
You will start hearing this from many companies, and you have already started hearing many of them say that we are facing supply constraints. We would have been able to generate more revenue if we had more supplies.
With Nvidia's acquisition of those supplies, it has strengthened its position, while everyone else has to wait until Nvidia has "had its fill" before they get their share. This is an important factor when we look ahead to the next two years.
Nvidia还向投资者证明了它年复一年强劲创新的能力。其下一代“Vera Rubin”技术在每兆瓦符号数方面比上一代“Grace Blackwell”技术好30倍。这仅仅一年之内就是巨大的提升，这正促使企业升级它们可能在2022年安装的旧技术。
而我们现在已经来到2026年和2027年。仅仅4到5年之后，它们就有巨大的动力升级到最新技术，因为这些技术提供更好的总拥有成本，并且让竞争对手落后Nvidia好几步，因为Nvidia继续以竞争对手跟不上的更快速度创新。
Nvidia has also proven to investors its ability to innovate strongly year after year. Its next-generation "Vera Rubin" technology is 30 times better in terms of symbols per megawatt compared to the previous "Grace Blackwell" technology.
That's a huge improvement in just one year, which is motivating companies to upgrade their older technologies that they may have installed in 2022. And here we are now in 2026 and 2027.
After only 4 or 5 years, they have a huge incentive to upgrade to the latest technologies because they offer a better total cost of ownership, and they keep competitors several steps behind Nvidia as it continues to innovate at a faster pace than competitors can keep up with.
Nvidia最近甚至在它长期没有强大存在的品类中也在创新，特别是在中央处理器（CPU）方面。Nvidia最近在中央处理器方面的进展带来了数十亿美元、甚至数百亿美元的销售额，而Nvidia预计这一数字明年将翻倍，并很可能继续增长。
Nvidia has recently been innovating even in categories where it has not had a strong presence in the long term, specifically in central processing units (CPUs). Nvidia's recent advancements in central processing units have led to sales in the billions, even tens of billions, of dollars, and Nvidia expects this figure to double next year and likely continue to increase.
Nvidia技术的买家获得良好的投资回报，而这些组件的价值保持稳定。事实上，就B200（也称为Blackwell 200）而言，两年后其价值比最初更高，因为二级市场需求在增加，人们正在抬高获取这项技术的价格。
它在一年半后达到初始价格的158%。H100技术在其推出四年后仍保留其价值的58%。Nvidia的A100技术推出已超过6年，仍保留其价值的25%。因此，购买这一产品的人的投资资本回报强劲，并且继续足以证明向这些产品投入巨额资金是合理的。
Buyers of Nvidia technologies get a good return on investment, and the value of these components remains stable. In fact, in the case of the B200, also known as the Blackwell 200, values became higher two years later than they were initially because demand in the secondary market is increasing, and people are raising the prices for access to this technology.
It reached 158% of the initial price after a year and a half. Four years after its launch, the H100 technology still retains 58% of its value. Nvidia's A100 technology, which was launched more than 6 years ago, still retains 25% of its value.
Therefore, the return on invested capital for people who buy this product is strong and continues to be sufficient to justify allocating huge sums to these products.
这甚至还没有算上来自实体人工智能的大规模收入回报，例如太空、机器人技术、自动驾驶出租车、机器人和通信。我认为下一个会带来回报的将是自动驾驶出租车。我们看到许多汽车制造商在这项技术上进行重大投资。
Nvidia宣布其汽车业务显著增长。两年之内，如果自动驾驶出租车的增长率使Nvidia汽车业务的收入翻倍或翻三倍，我不会感到意外。
This is even before we see large payouts of revenue from physical artificial intelligence, such as space, robotics technologies, self-driving taxis, robotics, and communications.
I think the next thing that will pay off will be self-driving taxis. We are seeing significant investments from many car manufacturers in this technology. Nvidia announces significant growth in its automotive sector.
Within two years, I wouldn't be surprised if the growth rate of self-driving taxis caused Nvidia's automotive sector revenues to double or triple.
因此，这些风险目前已经以相当大的程度反映在Nvidia的估值中。我向你们介绍的是一家蓬勃发展的公司，它凭借在研的新创新至少领先竞争对手一步，并且创新速度比竞争对手更快。它在一个价值数万亿美元的市场中运营，而且这个市场正在我们说话之际扩张。
公司预计下一财年收入增长70%，营业利润率接近65%。这不是一家通常会以仅14.66的远期市盈率出售的企业。
Therefore, the risks are currently factored into Nvidia's assessment in a substantial way. I have shared with you a thriving company that is at least one step ahead of the competition with new innovations in the pipeline, and is innovating faster than its competitors.
It operates in a market worth trillions of dollars and is expanding as we speak. The company expects revenue growth of 70% in the next fiscal year with operating profit margins approaching 65%.
This is not a business that would typically sell at a forward price-to-earnings ratio of only 14.66.
Nvidia以相对便宜的估值被出售，其原因在于其股价大约只有Walmart、Costco这类公司或其他增长更慢、或没有Nvidia那样的利润率或机会的公司的股价的一半。
The reason Nvidia is being sold at a relatively cheap valuation is roughly half the stock price of companies like Walmart, Costco, or other companies that are growing more slowly or do not have the margins or opportunities that Nvidia does.
但Nvidia以如此便宜的价格交易的原因是，风险已经非常严重地反映在其估值中。因此，这不是一只让投资者完全满意的股票。这不是一只投资者认为成功不可避免的股票，不是吗？
But the reason Nvidia is trading at such a cheap price is that the risks are very heavily factored into its valuation. Therefore, this is not a stock that investors are completely satisfied with.
This is not a stock where investors believe success is inevitable, is it?
对于Nvidia来说，风险每天都存在。我们在谈论风险。AI数据中心泡沫会破裂吗？OpenAI和Anthropic会怎样？它们在亏损，它们能继续下去吗？它们能持续吗？所有这些循环融资安排又怎么说？
来自AMD、Intel的竞争加剧，以及由Amazon、Alphabet、Microsoft、Meta等公司创建的专用集成电路，又怎么说？因此，风险每天都在被讨论和汇总，而且有些日子，一家人工智能公司的公告会导致Nvidia股价下跌2%、3%、4%或5%。
因此，这些风险是被计算在内的。在评估时已被考虑进去。这正是让我对Nvidia股票充满信心、非常有信心的原因，因为我没有看到这种满足感。我没有看到评级在不考虑这些新兴威胁的情况下直线上涨。
With Nvidia, risks are present every day. We are talking about risks. Will the AI data center bubble burst? What will happen to OpenAI and Anthropic? They are losing money, can they continue?
Will they be sustainable? What about all these circular financing arrangements? What about the growing competition from AMD, Intel, and the application-specific integrated circuits created by Amazon, Alphabet, Microsoft, Meta, and others?
Therefore, risks are discussed and consolidated every day, and there are certain days when an announcement from an AI company causes Nvidia's stock to drop by 2, 3, 4, or 5%. Therefore, these risks are calculated.
It was taken into consideration during the evaluation. This is what gives me confidence, a lot of confidence in Nvidia stock, because I don't see this kind of satisfaction. I don't see the rating rising straight up without taking into account these emerging threats.
恰恰相反，我的看法相当不同。我认为投资者过度敏感。每当有任何进展发生，他们就急于卖出。不是这样吗？他们时刻保持警惕。他们如履薄冰，一看到任何关于人工智能领域让他们对Nvidia股票感到担忧的报道，就准备卖出。
On the contrary, I see it quite differently. I think investors are overly sensitive. Whenever any development occurs, they rush to sell. Isn't that so? They are on constant alert.
They are walking on eggshells, ready to sell as soon as they see a report about anything in the field of artificial intelligence that makes them concerned about Nvidia's stock.
远期市盈率为14.6，而标普500股票的平均远期市盈率约为25，Nvidia相对于标普500股票的平均水平处于折价交易，而事实上，以所有重要业务指标衡量——收入增长、利润率、现金流、经营现金流与销售额之比、资产负债表、竞争优势等——它都是表现最好的公司之一。
因此，竞争加剧和不利的行业经济性等这些负面因素已经被考虑并纳入公司的估值中。
With a forward price-to-earnings ratio of 14.6, while the average S&P 500 stock is trading at a forward price-to-earnings ratio of around 25, Nvidia is trading at a discount to the average S&P 500 stock, while in fact it is one of the best performing companies when measured by all the important business metrics: revenue growth, profit margin, cash flow, cash flow from operations versus sales, balance sheet, competitive advantage, etc. Therefore, these negative factors such as increased competition and unfavorable sector economics have been taken into account and incorporated into the company's valuation.
因此，今天在Nvidia的演示之后，我也更新了我对该公司的公允价值估计，我计算出公允价值为344美元。当前市场价格为230美元。因此，我认为Nvidia股价在未来12到18个月内有约50%的潜在上涨空间。
Therefore, today, after Nvidia’s presentation, I also updated my fair value estimate for the company, and I calculated a fair value of $344. The current market price is $230. Therefore, I see a potential rise in Nvidia's stock by about 50% over the next 12 to 18 months.
这大概就是为什么我持有Nvidia股票，为什么我有兴趣增持更多，以及为什么我把Nvidia列为你现在能买入的最佳股票。
This is probably why I own Nvidia shares, why I'm interested in adding more of them, and why I rank Nvidia as the best stock you can buy right now.
这个频道对 $NVDA 的历次判断
Parkev Tatevosian, CFA 在这只股票上共 14 条判断，这里只列相邻的几条。