As the AI fervor stirred up by ChatGPT dispels the "tech winter" that plagued U.S. stocks at the end of last year, tech shares in the U.S. stock market have ridden the wave, emerging as the main force propelling U.S. stock indices. Among them, Nvidia, often hailed as the "sole arms dealer of the AI era," stands out.
From the start of the year, Nvidia has surged by over 180%, making it the top performer in the S&P 500 index. The company's market capitalization has crossed the trillion-dollar mark, putting it in league with tech titans like Apple, Alphabet, Microsoft, and Amazon.
Investors always seem eager to invest in a promising future. However, when turning a rational eye to Nvidia's current valuation, there might be some jitters. Based on the company's earnings over the past 12 months, Nvidia's Price-to-Earnings (P/E) ratio has skyrocketed to a staggering 212. For perspective, Amazon's P/E ratio stands at 110, while Tesla's is "merely" at 70.
With such lofty valuations, market sentiments often remain on edge. Any minor disruptions can cause rapid stock downturns. Just this week, without any substantial "bad news," Nvidia's stock dipped by 9.43%, marking its largest weekly decline since last September. By comparison, the NASDAQ fell by only 2.34% during the same period.
On Tuesday, analysts from Morgan Stanley even sounded the alarm, warning that the booming U.S. AI stock market, epitomized by Nvidia, might be nearing its peak.
Yet, if Nvidia's future performance can genuinely achieve high growth and meet analysts' projections, its stock price, while appearing high, could be seen as more reasonable. Data from FactSet indicates that Nvidia's forward P/E ratio, calculated based on anticipated earnings over the next 12 months, is 42, compared to Amazon's 51 and Tesla's 58.
Such data brings with it a set of assumptions about growth, including doubling net profits this year, leading the AI wave for a considerable time, holding its own amid fierce competition with the likes of Google and AMD, and avoiding any significant supply chain issues.
On Thursday, Christopher Gannatti, an analyst at asset management firm WisdomTree, shared his thoughts. He hinted at growing skepticism among investors, wondering if Nvidia's bright future is already reflected in its current stock price. He mentioned that high performance expectations are among the hardest hurdles for companies to overcome.
Bolstering Defenses
As Nvidia's valuation continues to climb, the company is taking steps to maintain its leadership position and meet investor expectations.
For instance, this June, Jensen Huang, Nvidia's CEO, dined with Taiwan Semiconductor Manufacturing Company's (TSMC) Chairman in Taiwan. Post dinner, Huang expressed his deep confidence in relying on this foundry, suggesting that Nvidia has secured the necessary supplies. Notably, TSMC is the exclusive manufacturer of Nvidia's popular GPU H100.
In recent years, Nvidia has also evolved into a major venture capital player, specifically focusing on supporting companies that collaborate with AI models.
According to Pitchbook data, this year, Nvidia has invested in at least 12 startups, including rising AI unicorns such as Runway, Inflection AI, and CoreWeave. These investments could potentially cultivate a growing customer base for Nvidia, not just boosting sales but also diversifying its clientele.
GPU Supply Shortfalls
With the rise of pretrained models, computational challenges have magnified. Consequently, various training and inference solutions have been proposed for Large Language Models (LLMs). Most high-performance inference solutions are CUDA-based and optimized for Nvidia GPUs.
Recently, companies like Microsoft and OpenAI have indicated that they are taking steps to address the shortage of specialized GPUs like H100 and A100 for AI tasks. For instance, Microsoft is limiting employee access to GPUs. Even Elon Musk humorously commented on how challenging it is to acquire enterprise-level GPUs.
As Harsh Kumar, an analyst at investment firm Piper Sandler, succinctly put it, "In essence, they [Nvidia] have the best GPUs."
A deep dive by the well-known tech blog GPU Utils estimated a demand of about 432,000 H100 GPUs across various entities, equating to a value of approximately $15 billion. This figure doesn't even consider Chinese firms like ByteDance (TikTok), Baidu, and Tencent that need a considerable number of H800s.
The author of the blog, Pascal, noted that it might take a while for this supply shortage to dissipate. He added that Nvidia is aggressively ramping up production, but chip giant TSMC still can't produce enough high-end GPUs. Nvidia might collaborate with Intel and Samsung in the future, but this wouldn't resolve the supply crunch in the short term.
Analysts believe that when AI firms and developers use CUDA and Nvidia GPUs for model construction, they are less likely to switch to alternatives like AMD chips or Google's Tensor Processing Units (TPUs). Patrick Moorhead, a semiconductor analyst at Moor Insights, emphasized that Nvidia currently enjoys a dual advantage: not only do they possess the highest performance training hardware, but in the realm of AI software, they also have libraries and CUDA.
Nvidia's Rise to Greatness
Who could've predicted that Nvidia, valued at just $8.4 billion a decade ago, would blossom into the tech behemoth it is today?
Before the rise of AI, Nvidia was renowned for producing high-performance graphics cards for gaming enthusiasts.
In 2006, while AMD plunged into financial distress following an aggressive acquisition of ATI and Intel struggled in the GPU market after its graphics partner ATI was acquired by AMD, Nvidia seized the opportunity. Under the leadership of Jensen Huang, Nvidia was ahead of the curve in commercializing its initially gaming-centric GPU business.
Later, even when the nascent AI industry wasn't well-received by the market, Nvidia astutely foresaw the potential of GPUs in AI model training and launched CUDA (Compute Unified Device Architecture). This software platform became an essential tool for researchers and developers, thrusting Nvidia into AI dominance.
For a time, Nvidia was also aggressively pursuing ARM's acquisition, a move designed to counteract the emerging challenges from Apple Silicon and RISC-V. However, regulatory hurdles stalled this deal.
Today's Nvidia and Tomorrow's Challenges
It's undeniable that Nvidia currently holds a significant position in the tech industry. But with the rise of competitors like Google's Tensor Processing Units, there's increasing pressure on Nvidia to innovate and remain the leader in this fast-evolving landscape.
However, with its diverse portfolio, including ventures into automotive AI, the NVIDIA DRIVE platform, and investments in various AI startups, Nvidia appears to be laying a solid foundation for its future.
As the AI wave continues to surge, it remains to be seen whether Nvidia can ride the crest or get engulfed by the looming waves of competition. Whatever the outcome, Nvidia's journey is a testament to the transformative power of technology and innovation in the modern era.
