Is Nvidia entering the “big trend” era?

美股研究社 · May 31, 2023 15:26

来源：美股研究社作者：Vlad Deshkovich

$英伟达 (NVDA.US)$最新的财报可能是今年股市上最大的事件。英伟达的业绩与指引均大幅超出预期，使其股价在一个交易日内上涨了27%，市值增加了2020亿美元。

虽然距离财报发布已经有一段时间，但英伟达继续成为牛市和熊市的热门话题。许多投资者称这是一次“前所未有”或“一生一次”的业绩发布。

英伟达的股票今年以来一直保持着强劲的势头，而最近的这一表现进一步推动了这一势头。截至上周收盘，英伟达今年迄今的纳斯达克综合指数回报率为6.9倍，标准普尔500指数的回报率为惊人的16.89倍。

这些最新的研究结果使围绕英伟达的讨论更加激烈。有投资者认为，如果我们正处牛市，英伟达是一个长期投资标的。但就熊市情况来说，目前估值过高。

过去一周的业绩和股价走势为双方的论点增加了论据，但没有从根本上改变任何一方的说法。

那英伟达究竟是否是一个好的投资标的？其估值是否已经超出其本身价值？

01、生成式AI时代的最大受益者

在看英伟达的估值之前，我想先谈谈它的业务。众所周知，英伟达是市场领导者，也是人工智能趋势的核心受益者，抛开宏观利好，更详细地评估其产品和市场定位对投资者来说才是明智的选择。

目前，英伟达的销售额正在迅速增长，因为其在人工智能发展中占据了绝对重要的地位。这是英伟达真正核心竞争力的长期趋势的一部分——加速计算。加速计算是指计算架构使用不同于CPU的另一个处理器来完成某些任务。在加速计算体系结构中，计算机将某些类型的任务从中央处理单元卸载到另一个处理器上。这个协处理器是针对并行计算进行了优化的专用硬件，可以提高整个计算系统的效率。

自2006年以来，这一直是NVIDIA的重点。在此之前，NVIDIA一直专注于优化图形处理的硬件，但在那一年，通过发布CUDA, NVIDIA大大扩展了其产品和整个潜在市场。CUDA软件允许在NVIDIA硬件上进行一般并行处理，允许客户首次利用显卡进行其他类型的并行计算。这使加速计算成为一种特别适合大规模并行计算机工作负载的范例。

事实证明，这种计算是构建人工智能系统所需要的。加速计算架构对于人工智能工作负载来说非常高效，因为人工智能需要大规模的并行处理，就像计算机图形学一样。一段时间以来，这一直是人工智能或机器学习工作负载的标准架构。已经参与真正的人工智能研究的公司需要使用这种架构来运营一个重要的数据中心，以便开发人工智能软件。在人工智能领域刚刚起步的公司最终需要拥有它或有机会获得它。

这一切都在企业GPU市场上演。公司的数据中心需要图形处理单元，以实现加速计算。GPU是并行优化的处理器，支持这种计算范式。在这个市场上，英伟达无疑是王者。截至2021年第四季度，英伟达在企业GPU市场的份额为91.4%。唯一值得注意的竞争对手是AMD，当时的市场份额为8.5%。这是一个市场主导地位，表明英伟达的优越产品，并且看起来会持续下去。

从更投机的角度来看，还有另一个市场值得一提:个人电脑GPU市场。就像企业数据中心一样，GPU和优化的协处理器也越来越多地出现在消费电子产品中。虽然电子游戏可能已经开启并维持了这一趋势，但人工智能似乎也成为了一个因素:自2017年第三季度以来，苹果自iPhone 8以来就在iPhone上安装了专用的人工智能协处理器。随着人工智能计算工作负载变得越来越普遍，越来越多的人工智能计算工作负载可能开始出现在终端用户设备上——这可能也会增加这个市场的需求。

也许令人惊讶的是，英伟达不是PC GPU市场的领导者，英特尔（INTC）才是。

看看这个市场的需求和市场份额构成如何演变将是一件有趣的事情。虽然目前还没有像企业GPU市场那样直接发挥作用，但随着人工智能变得越来越普遍，它是一个值得关注的市场。

总的来说，我们可以看到为什么英伟达具有独特的优势，可以从人工智能趋势中受益并能继续这样做。

02、估值

在概述了英伟达的差异化之后，我们现在可以看看它的预期收益和相对估值。推断英伟达和类似实体的当前增长率和估值，以了解当前的市场价格意味着什么。这样做的效果是更详细地构建熊市的情况，以确定它是否合理，以及在什么时间间隔内。

我认为，首先应该将英伟达与其在GPU领域的主要竞争对手——英特尔(Intel)和AMD——进行比较。

此外，我认为将英伟达与其他超大型(市值超过5000亿美元)科技股进行比较也是有道理的。这是因为这一篮子股票作为具有显著增长溢价的前瞻性工具进行交易，其波动可以告知市场如何在相对基础上为英伟达的未来收益定价。

下表比较了英伟达与同行的预期(共识)预期收益：

稀释的每股收益。由于各公司的财政报告期略有差异，因此利用2022年日历年度的四个季度数据进行计算。

由于亚马逊（AMZN）上一财年的EPS为负数，因此其估计值的计算方式与众不同。为了更好地反映远期盈利复合增长率，假设它将在2023年恢复到其10年追踪年度EPS的平均值（0.78美元），然后再应用预期长期盈利增长率。这样可以在没有负/正数之间交叉的数值扭曲情况下进行一个前瞻性估计。

按照当前预期，英伟达将在2023财年的每股盈利上轻松超过英特尔。然后，它将继续在其他半导体竞争者中表现出色，但整体而言仍落后于大型科技公司的平均水平，只有到2028年才能与这些公司的平均水平相交。此时，英伟达的每股盈利将超过除微软（MSFT）和META（META）之外的所有竞争对手。这表明，它有潜力实现其估值增长。

接下来，我们可以看一下英伟达及其竞争对手在此期间相对市盈率估值的演变情况。该图表不考虑流通股数量的变化。

英伟达在过去一年、2023年和2024年的市盈率确实显得特别昂贵。同时还要注意的是，在这段时间里，谷歌(Google)的股价仍然非常便宜。

从2025年开始，英伟达的情况开始变得更加合理，其市盈率大幅接近平均水平。英伟达的市盈率在2026年接近AMD，在2027年实际上低于AMD，在2028年甚至变得相对更便宜。

到2026年，英伟达的市盈率溢价将在平均水平的50%左右，此后将会更低，到2028年将低于25%。对于目前和历史来看，这种股票的估值溢价还不算是特别高。

考虑到英伟达的相对增长率，从长远来看，英伟达的股价并不算太高。它每年都会有相对便宜的趋势出现，甚至从2027年开始，它的价格不如AMD那样昂贵。

03、风险

这种情况发生的第一个风险是相对增长率的变化。英伟达的半导体竞争对手可以开始吸收它的市场份额，或者大大加快他们自己的增长率，它的远期价格在相对基础上将不那么有吸引力，在目前的价格下，它可能不再被视为一个合适的投资。

另一个风险是，无论相对估值如何，该公司的成长溢价都会降低。如果英伟达最终表现不如其目前显著提高的增长预期，投资者对未来回报的预期将会降低，并可能抛售股票。这将降低该股的增长溢价，并使该公司的交易市盈率更接近行业中位数。

英伟达的最后一个也可以说是最重要的风险是供应链风险。英伟达的业务是围绕着设计而不是生产自己的芯片而展开的，如果其供应链出现中断，其业务将受到严重阻碍。虽然台积电（TSM）层面正在进行供应链多样化的努力，但它们仍处于早期阶段。因此，我认为这种风险是实质性的，拥有英特尔股票将是对这种性质的任何潜在风险的绝佳对冲。

04、总结

按照目前的价格和预期增长率，预计英伟达将在不到十年的时间内达到目前的估值水平，并在更短的时间里变得相对便宜。持续的高增长预期可能会使该股的估值保持健康的增长溢价，使其接近当前/历史的市盈率。在其他条件不变的情况下，期望这种增长溢价能够持续存在，只要公司在目前的增长阶段能够交付高于指导水平或有更好的业绩。

与此同时，与其他大型科技公司相比，它的相对市盈率也在不断提高。这意味着该股在未来5年或更长时间内有上涨潜力。

需要明确的是，预计在接下来的四个季度中，英伟达的盈利不仅会增长，还会保持或超过目前的市盈率。以目前的水平增长收益，将使英伟达能够维持投资者对其未来强劲增长的预期，并保持其股票目前的增长溢价水平。

编辑/lambor

Source: US Stock Research Institute Author: Vlad Deshkovich

$NVIDIA (NVDA.US)$The latest earnings report is probably the biggest event in the stock market this year. Nvidia's performance and guidance greatly exceeded expectations, causing its stock price to rise 27% and market capitalization to increase by $2020 billion in a single trading day.

Although it has been a while since the earnings report was released, Nvidia continues to be a hot topic in bull and bear markets. Many investors called this an “unprecedented” or “once-in-a-lifetime” earnings release.

Nvidia's stock has maintained a strong momentum since this year, and this recent performance has further fueled that momentum. As of last week's closing, Nvidia's return on the Nasdaq Composite Index so far this year was 6.9 times, and the return on the S&P 500 was an astonishing 16.89 times.

These latest findings have fueled discussions surrounding Nvidia. Some investors think that if we are in a bull market, Nvidia is a long-term investment target. However, as far as the bear market situation is concerned, the current valuation is too high.

The past week's performance and stock price trends have added arguments to both parties' arguments, but have not fundamentally changed either side's statement.

So is Nvidia a good investment target? Has its valuation exceeded its own value?

01. The biggest beneficiaries of the generative AI era

Before I look at Nvidia's valuation, I'd like to talk about its business. As we all know, Nvidia is a market leader and a core beneficiary of artificial intelligence trends. Leaving aside macro-benefits, evaluating its products and market positioning in more detail is a wise choice for investors.

Currently, Nvidia's sales are growing rapidly as it occupies an absolutely important position in the development of artificial intelligence. This is part of a long-term trend of Nvidia's true core competency — accelerated computing. Accelerated computing means that the computing architecture uses a different processor than the CPU to complete certain tasks. In an accelerated computing architecture, computers offload certain types of tasks from the central processing unit to another processor. This coprocessor is dedicated hardware optimized for parallel computing and can improve the efficiency of the entire computing system.

This has been NVIDIA's focus since 2006. Until then, NVIDIA had been focusing on hardware optimized for graphics processing, but in that year, by releasing CUDA, NVIDIA greatly expanded its products and the entire potential market. CUDA software allows general parallel processing on NVIDIA hardware, allowing customers to use video cards for other types of parallel computing for the first time. This makes accelerated computing an example particularly suited to massively parallel computer workloads.

It turns out that this kind of calculation is needed to build artificial intelligence systems. Accelerated computing architectures are highly efficient for artificial intelligence workloads because artificial intelligence requires large-scale parallel processing, just like computer graphics. This has been the standard architecture for artificial intelligence or machine learning workloads for some time. Companies already involved in real artificial intelligence research need to use this architecture to operate a critical data center in order to develop artificial intelligence software. Companies that are just starting out in the field of artificial intelligence will eventually need to own it or have a chance to get it.

All of this is happening in the enterprise GPU market. The company's data center needed graphics processing units to enable accelerated computing. GPUs are parallel optimized processors that support this computing paradigm. In this market, Nvidia is unquestionably king. As of the fourth quarter of 2021, Nvidia's share of the enterprise GPU market was 91.4%. The only notable competitor was AMD, which had an 8.5% market share at the time. It's a market dominance that shows Nvidia's superior product, and it looks like it's going to last.

From a more speculative perspective, there is another market worth mentioning: the PC GPU market. Just like enterprise data centers, GPUs and optimized coprocessors are increasingly appearing in consumer electronics. While video games may have started and maintained this trend, artificial intelligence also appears to be a factor: since the third quarter of 2017, Apple has installed dedicated artificial intelligence coprocessors on iPhones since the iPhone 8. As AI computing workloads become more common, more AI computing workloads may begin to appear on end user devices — which may also increase demand in this market.

Perhaps surprisingly, Nvidia isn't the leader in the PC GPU market; Intel (INTC) is.

It will be interesting to see how the demand and market share composition of this market evolves. Although it doesn't have the same direct role as the enterprise GPU market, it's a market worth watching as artificial intelligence becomes more common.

Overall, we can see why Nvidia is uniquely positioned to benefit from AI trends and continue to do so.

02. Valuation

After outlining Nvidia's differentiation, we can now take a look at its expected earnings and relative valuation. Extrapolate the current growth rates and valuations of Nvidia and similar entities to understand what current market prices mean. The effect of this is to structure the bear market situation in more detail to determine if it is reasonable and within what time interval.

In my opinion, Nvidia should first be compared to its main competitors in the GPU field — Intel (Intel) and AMD.

Furthermore, I think it makes sense to compare Nvidia with other very large (market capitalization over 500 billion US dollars) technology stocks. This is because this basket of stocks is traded as a forward-looking instrument with a significant growth premium, and its fluctuations can tell the market how Nvidia's future earnings are priced on a relative basis.

The following table compares Nvidia's anticipated (consensus) expected earnings with peers:

Diluted earnings per share. Due to slight differences in the financial reporting periods of each company, the calculation was made using data for the four quarters of calendar year 2022.

Since Amazon (AMZN) had a negative EPS for the previous fiscal year, its estimate was calculated in a different way. To better reflect the compound growth rate of forward earnings, let's say it will return to its 10-year average tracking annual EPS in 2023 ($0.78) before applying the expected long-term earnings growth rate. This makes it possible to make a forward-looking estimate without numerical distortions crossing between negative/positive numbers.

According to current expectations, Nvidia will easily surpass Intel in earnings per share for fiscal year 2023. It will then continue to excel among other semiconductor competitors, but overall it still lags behind the average of big tech companies, and will only intersect the average of these companies until 2028. At this point, Nvidia's earnings per share will surpass all of its competitors except Microsoft (MSFT) and META (META). This shows that it has the potential to grow its valuation.

Next, we can look at the evolution of Nvidia and its competitors' relative price-earnings ratio valuations over this period. The chart does not take into account changes in the number of outstanding shares.

Nvidia's price-earnings ratios for the past year, 2023, and 2024 have indeed been particularly expensive. At the same time, it is important to note that during this period, Google (Google)'s stock price was still very cheap.

Beginning in 2025, Nvidia's situation began to become more reasonable, and its price-earnings ratio was significantly close to average. Nvidia's price-earnings ratio was close to AMD in 2026, actually lower than AMD in 2027, and even became relatively cheaper in 2028.

By 2026, Nvidia's price-earnings ratio premium will be around 50% of the average, then lower, and below 25% by 2028. Looking at the present and history, the valuation premium for this kind of stock is not particularly high.

Considering Nvidia's relative growth rate, Nvidia's stock price isn't too high in the long run. It has a trend of being relatively cheap every year, and even starting 2027, it's not as expensive as AMD.

03. Risks

The first risk of this happening is a change in the relative growth rate. Nvidia's semiconductor rivals can begin to absorb its market share, or significantly speed up their own growth rate. Its forward price will be less attractive on a relative basis, and it may no longer be seen as a suitable investment at current prices.

Another risk is that the company's growth premium will decrease regardless of the relative valuation. If Nvidia's final performance falls short of its current significantly higher growth expectations, investors' expectations for future returns will decline, and stocks may be sold off. This would lower the stock's growth premium and bring the company's price-earnings ratio closer to the industry median.

Nvidia's last and arguably most important risk is supply chain risk. Nvidia's business revolves around designing rather than producing its own chips, and if its supply chain is disrupted, its business will be severely hampered. While supply chain diversification efforts are ongoing at the TSMC (TSM) level, they are still in the early stages. So I think this risk is substantial, and owning Intel stock would be an excellent hedge against any potential risk of this nature.

04. Summary

Based on current prices and projected growth rates, Nvidia is expected to reach its current valuation level in less than a decade and become relatively cheap in a shorter period of time. Continued high growth expectations are likely to keep the stock's valuation at a healthy growth premium, bringing it close to current/historical price-earnings ratios. With other conditions unchanged, it is expected that this growth premium will continue as long as the company can deliver higher than the guidance level or perform better during the current growth phase.

At the same time, its relative price-earnings ratio continues to rise compared to other big tech companies. This means the stock has the potential to rise over the next 5 years or more.

What needs to be clear is that in the next four quarters, Nvidia's profit is expected not only to grow, but also to maintain or exceed the current price-earnings ratio. Growing earnings at current levels will enable Nvidia to maintain investors' expectations for strong future growth and maintain its stock's current growth premium level.

edit/lambor

The translation is provided by third-party software.

The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.

英伟达迈进“大势”时代？