Is NVIDIA's "triple jump" to the top of the global market cap driven by irrational enthusiasm?

Despite repeated executive sell-offs, NVIDIA (NVDA.US) stock recently hit a new high, with a market value of $3.34 trillion at the close of the U.S. stock market on June 18, surpassing Microsoft to become the world's most valuable listed company.

Since reaching a market value of $1 trillion in May 2023, NVIDIA's stock price has "triple-jumped" in just about a year. Earlier this year, the market questioned whether NVIDIA's high valuation was a bubble, but analysts have now set even grander targets, predicting that the company's market value could climb to nearly $5 trillion next year.

Despite ongoing stock sales by insiders at NVIDIA this year, the sell-offs have not put downward pressure on the company's stock price. According to statistics compiled by Washington Service, NVIDIA executives and directors have sold approximately 770,000 shares of NVIDIA stock this year, worth over $700 million (excluding the impact of the stock split on June 10 at a ratio of 1 to 10). Since the release of the Q1 financial report in May, more than a third of NVIDIA insiders have chosen to sell their shares.

As of now, NVIDIA's price-earnings ratio is 78 times, far exceeding Microsoft (38.5 times) and Apple (32.6 times) which follow closely behind. After setting a new record once again, is NVIDIA's current valuation already "overheated"?

Q1 Performance Exceeds Expectations, Revenue Surges Over 200% for Three Consecutive Quarters

Benefiting from the strong demand for data center server AI accelerator chips brought by the AI "arms race," NVIDIA's Q1 performance continued to significantly exceed market expectations.

According to the Zhitong Finance APP, in the first quarter of 2024, NVIDIA achieved revenue of $26 billion, a year-on-year increase of 262%, with a net profit of $14.81 billion, a year-on-year increase of 628%. This is the third consecutive quarter that NVIDIA's revenue growth has exceeded 200% year-on-year, with data center business revenue reaching $22.6 billion, a year-on-year increase of 427% and a quarter-on-quarter increase of 23%, making it the main contributing part to this growth.

Regarding the first-quarter performance, NVIDIA's CFO stated that the demand from customers in various segments was very strong, especially from enterprises and consumer internet companies, as well as the revenue increase brought by large cloud service providers deploying and improving their infrastructure.

Among the products currently on the market, the Hopper architecture GPU (such as the H100) is the main product driving the company's first-quarter performance growth. For example, Meta's release of the Llama 3 open-source large model used nearly 24,000 H100 GPUs; the H200 has been upgraded from the H100, with memory bandwidth increased by 1.4 times, memory capacity increased by 1.8 times, and an enhanced ability to process generative AI tasks According to company sources, the supply situation of the H100 GPU has improved somewhat, but there is still a shortage; the H200 released in November last year is being shipped at full capacity and is expected to be launched in Q2 2024.

In terms of the AI industry ecosystem, NVIDIA recently released the NVIDIA NIM cloud-native microservices, allowing developers to download NVIDIA NIM to deploy AI models in the cloud, data centers, or workstations, and build various generative AI applications. NVIDIA also revealed that nearly 200 technology partners, including Cadence, Cloudera, Cohesity, DataStax, NetApp, Scale AI, and Newsight Technology, are integrating NIM into their platforms to accelerate the deployment of generative AI.

It is worth mentioning that NVIDIA seems to particularly value the medical and biological industry in downstream AI applications. Not only can NIM support a variety of AI applications including surgical planning, digital assistants, drug discovery, and clinical trial optimization, but in recent years, it has also invested in several companies focusing on AI pharmaceuticals, gene editing, precision medicine, and other fields.

Looking ahead, NVIDIA is planning to launch the new generation AI chip and supercomputing platform Blackwell series. It is reported that NVIDIA's first Blackwell chip, named GB200, is currently the "world's most powerful chip," has been fully produced, is expected to ship in the second quarter of this year, increase production in the third quarter, and officially deploy in data centers in the fourth quarter, with significant revenue growth expected this year.

According to the Zhitong Finance APP, NVIDIA's next-generation architecture is named Rubin, and the next-generation CPU is named "Vera." Products based on the Rubin architecture are expected to arrive between the fourth quarter of 2025 and the first quarter of 2026, rumored to be manufactured using TSMC's 3nm process, with CoWoS-L packaging like B100, likely one of the first products to use HBM4. Unlike the industry's usual "two-year update" cycle, NVIDIA will follow a "one-year update" fast-paced product iteration to keep up with market demand changes.

Despite the optimistic market outlook, are there still risks in the Chinese market?

While NVIDIA's performance and stock price continue to soar, its market share also continues to rise. According to the latest report from market research firm Jon Peddie Research, the global discrete graphics card (AIB) shipments in the first quarter of 2024 were 8.7 million units, a decrease of 7.9% from the previous quarter's 9.5 million units, but a year-on-year increase of 39.2%. NVIDIA's share further increased to 88%, AMD's share decreased by 7 percentage points to 12%, and Intel's share was less than 1%.

![image.png](https://img.zhitongcaijing.com/image/20240620/1718885003526705.png? According to the Zhitong Finance APP, despite NVIDIA's stock price having risen continuously for over a year and surged over 200% in the past year, most research institutions still believe that the company's upward trend is not over. For instance, Rosenblatt Securities analyst Hans Mosesmann raised the target price of this chip manufacturer from $140 to a high of $200. Bank of America reiterated its "buy" rating on NVIDIA stock in a report released on June 5, stating that the company led by Huang Renxun remains the top choice in the IT industry and set a target price of up to $150 within 12 months. Susquehanna also raised the company's target price from $145 to $160, indicating that although the company's valuation is much higher than its peers, it is still reasonable because NVIDIA can seize significant opportunities in the booming AI terminal market.

In addition, former world's richest person Elon Musk also highly praises NVIDIA's AI chips. At Tesla's Q1 2024 earnings call, Musk announced plans to increase the active quantity of the company's H100 (NVIDIA's flagship AI chip) from 35,000 to 85,000 within the year and invest up to $10 billion this year for joint training and inference work in AI. Not only does Tesla need a large number of NVIDIA GPUs for AI training and workloads for autonomous driving, X company also needs to deploy H100 GPU clusters to build a "computing power super factory." Musk previously revealed that training the Grok 2 model requires approximately 20,000 NVIDIA H100 GPUs, while training the next-generation models and higher versions will require 100,000 NVIDIA H100 chips.

However, there are also voices in the market warning against the AI bubble. Howard Marks, co-founder of Oaktree Capital, recently stated in an interview that the market currently exhibits a very fierce wildness (artificial intelligence AI), which dominates and disconnects the market from economic fundamentals. If emotions dominate and positive changes in the real world do not meet investors' expectations, the stock market may even decline. For example, in the case of artificial intelligence, if there is excessive hype, overly high expectations for NVIDIA, and failure to meet expectations, it may lead to disillusionment and disappointment.

As for the previously concerned decline in NVIDIA's market share in China, NVIDIA's response is to provide several new types of AI chip samples to the Chinese market, which contain most of the latest AI functions but with reduced computing power to comply with export control regulations. Whether this product can be accepted by Chinese buyers is still unknown. Previously, NVIDIA executives admitted at an earnings call that the company's sales in China "substantially" declined in Q1 this year Behind the record-breaking financial performance, NVIDIA not only faces competition from rivals such as Google and Microsoft, but also needs to be wary of the risk of slowing growth in generative AI. Some analysts point out that the cost of the "arms race" in computing power is growing exponentially, but the number of cases that can truly achieve commercialization is not optimistic. If NVIDIA wants to continue to deepen its value in the process of integrating AI into its business, it will also face many challenges