ChatGPT has sparked an AI frenzy, sweeping away the &#34;tech winter&#34; in the US stock market at the end of last year. Tech stocks in the US have risen one after another, becoming the main driving force behind the rise of the US stock market. Among them, the most eye-catching is NVIDIA, known as the &#34;sole arms dealer in the AI era&#34;.
Since the beginning of the year, NVIDIA has outperformed with a staggering 180% increase, making it the best-performing company in the S&amp;P 500 index. The company's market value has surpassed $1 trillion, keeping pace with the four major tech giants: Apple, Alphabet, Microsoft, and Amazon.
Investors are always willing to bet on an imaginative future. However, when it comes to rational valuation, those who delve into NVIDIA's current valuation are likely to be somewhat apprehensive. Based on the company's earnings over the past 12 months, NVIDIA's price-to-earnings ratio has reached an exaggerated 212. By comparison, in the tech industry where valuations are relatively high, Amazon's price-to-earnings ratio is 110, while Tesla's is &#34;only&#34; 70.
High valuations often come with the most tense market sentiment. Any slight movement can cause the company's stock price to quickly turn around. This week, without any substantial &#34;bad news,&#34; NVIDIA's stock price fell by 9.43%, marking the largest weekly decline since September last year. In comparison, the Nasdaq fell by 2.34% during the same period.
On Tuesday, Morgan Stanley analysts went so far as to directly &#34;short&#34; NVIDIA, warning that the &#34;bull market of AI stocks in the US&#34; represented by NVIDIA may be nearing its peak.
However, if NVIDIA's future performance can achieve solid high growth as expected by analysts, then compared to other tech stocks, although NVIDIA's stock price may appear high, it is actually more reasonable. According to FactSet data, based on the expected earnings for the next 12 months, NVIDIA's price-to-earnings ratio is 42. By comparison, Amazon's is 51, and Tesla's is 58.
This implies a series of assumptions about growth, including doubling net profit this year, being able to dominate the AI wave for a considerable period of time in the future, standing tall in the increasingly fierce competition with Google, AMD, and others, and avoiding any major supply issues.
On Thursday, Christopher Gannatti, an analyst at asset management company WisdomTree, wrote:
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/07716abd-c1a3-4008-a82e-f4655ef6130b.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/07716abd-c1a3-4008-a82e-f4655ef6130b.png"/>
<h2>Strengthening the Moat</h2>
While NVIDIA's valuation continues to rise, the company is taking various measures to ensure its leading position and meet investors' expectations.
For example, in June of this year, Huang Renxun flew to Taiwan, China to have dinner with the Chairman of TSMC, Zhang Zhongqian. After the meal, Huang Renxun expressed that he &#34;feels very secure relying on this foundry,&#34; implying that NVIDIA has ensured the necessary supply. It is worth mentioning that currently, TSMC is the sole producer of NVIDIA's popular GPU H100.
In addition, in recent years, NVIDIA has gradually developed into a heavyweight venture capital company in the field of investment, specifically focusing on supporting companies that collaborate with AI models.
According to Pitchbook data, NVIDIA has already invested in at least 12 startups this year, including Runway, Inflection AI, CoreWeave, and other AI unicorns that have recently gained attention.
These investments may create a growing customer base for the company, which can not only increase NVIDIA's sales in the future but also bring a more diverse customer group.
<h2>GPU Supply Shortage</h2>
Since the rise of pre-trained large models, the computational challenges people face have become increasingly significant. As a result, many training and inference solutions have been proposed for large language models (LLMs). Clearly, most high-performance inference solutions are based on CUDA and optimized for NVIDIA GPUs.
Recently, companies such as Microsoft and OpenAI have stated that they are taking necessary measures to alleviate the shortage of H100 and A100 dedicated GPUs used for AI tasks. For example, Microsoft is limiting employee access to GPUs, and the CEO of Quora stated that hardware shortages are masking the true potential of AI applications. Elon Musk even joked that it is harder to get enterprise-grade GPUs than to buy &#34;drugs.&#34;
As Harsh Kumar, an analyst at investment firm Piper Sandler, put it:
<blockquote>
&#34;In short, they have the best GPUs.&#34;
</blockquote>
Previously, the well-known tech blog GPU Utils conducted an in-depth investigation into the GPU shortage issue. According to their findings, OpenAI may need 50,000 units of H100, Meta may need 25,000 units, large cloud providers may each need 30,000 units, and combined with other startups, the total demand for H100 could be approximately 432,000 units.
Based on the calculation of approximately $35,000 per unit, the value of the required GPUs is about $15 billion. This does not include Chinese companies like ByteDance (TikTok), Baidu, and Tencent, which require a large number of H800s.
Blogger Pascal wrote:
<blockquote>
Therefore, the supply shortage may take some time to disappear.
But my estimates may be greatly exaggerated, as many companies today will not immediately purchase H100s; they will upgrade over time.
In addition, NVIDIA is actively increasing production capacity.
</blockquote>
Pascal also pointed out that chip giant TSMC is still unable to produce enough high-end GPUs. NVIDIA may also collaborate with chip manufacturers Intel and Samsung in the future, but the supply shortage will not be resolved in the short term.
Analysts also believe that currently, when AI companies and programmers use CUDA and NVIDIA's GPUs to build models, they rarely turn to competitors such as AMD's chips or Google's Tensor Processing Units (TPUs). Patrick Moorhead, a semiconductor analyst at Moor Insights, said:
<blockquote>
&#34;NVIDIA currently has a double moat. They not only have the highest-performance training hardware, but they also have libraries and CUDA on the software input side in the AI field.&#34;
</blockquote>
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6590965a-b3ff-4f62-bcd7-885daecfa805.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6590965a-b3ff-4f62-bcd7-885daecfa805.png"/>
<h2>NVIDIA's &#34;Road to Greatness&#34;</h2>
Who would have thought that NVIDIA, which is now shining brightly, was just a &#34;small company&#34; with a market value of only $8.4 billion ten years ago.
Before the rise of artificial intelligence, NVIDIA was known for producing high-performance graphics cards for gaming enthusiasts.
In 2006, AMD fell into a severe financial crisis due to its aggressive acquisition of ATI and entered a dark decade. Meanwhile, Intel, due to its graphics processor partner ATI being acquired by rival AMD, failed to make any progress in the GPU field for several years.
However, the acquisition that caused both AMD and Intel to suffer losses actually gave NVIDIA an opportunity to grow and expand. Under the leadership of &#34;Leather Jacket Guy&#34; Jensen Huang, NVIDIA took the lead in commercializing the GPU business that was initially vertically applied to the gaming industry.
Later, in the early stages of AI industry development, when the market had little confidence in it, the &#34;Green Factory&#34; foresaw the application of GPUs in the AI market and decisively invested in research and development. These visionary technological choices not only allowed NVIDIA to survive in the squeeze between semiconductor giants like Intel and AMD but also laid a solid foundation for NVIDIA's subsequent explosive growth.
On the stock front, NVIDIA, which dominates the graphics processor field, has experienced a surge in stock price since 2015.
Subsequently, driven by waves of trends such as gaming graphics cards, cryptocurrency mining, autonomous driving, and the AI boom, NVIDIA's stock price has grown exponentially. This year, the company's market value has even surpassed one trillion, putting it on par with tech giants like Apple, Alphabet, Microsoft, and Amazon.
However, when it comes to stock prices, as demonstrated by the volatile market conditions this week, investors are quick to hit the sell button without hesitation whenever there is any sign of movement, given the high PE multiples.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c9fbee7e-545e-43b0-a0e3-ae7678991a54.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c9fbee7e-545e-43b0-a0e3-ae7678991a54.png"/>

NVDA

Accompanying high valuations is often the most tense market sentiment - any slight movement can cause the company's stock price to quickly reverse.

NVIDIA, with a staggering 180% surge in value this year and a price-to-earnings ratio exceeding 210, is it too expensive?