Artificial intelligence (AI) leader Nvidia isn’t perfect, but that may be a painful realization for investors.
Thirty years ago, the internet began to go mainstream, setting in motion a chain of events that would forever change the growth trajectory of American companies.
Wall Street has been waiting, sometimes impatiently, for the next big trend to rival the impact the internet has had on business, and after a long wait, the artificial intelligence (AI) revolution seems poised to answer the call.
AI gives software and systems the autonomy to manage tasks typically handled by humans. The key to AI’s long-term success, and the source of its seemingly limitless potential, is the ability of AI-driven software and systems to learn without human intervention. Machine learning gives AI the ability to become more adept at existing tasks and potentially acquire new skills.
No company has benefited more directly from the rise of AI than semiconductor giant Nvidia. (NVDA -0.21%).
Image source: Getty Images.
Until recently, NVIDIA’s expansion was flawless.
At the start of 2023, Nvidia’s market cap was on the verge of reaching $360 billion and becoming one of the most influential technology companies in the United States. On June 20, 2024, less than two weeks after completing its historic 1-for-10 stock split, Nvidia’s market cap peaked at $3.46 trillion on an intraday basis. Investors had never before seen a market leader gain more than $3 trillion in value in less than 18 months.
The catalyst for this likely once-in-a-lifetime move is the company’s AI graphics processing units (GPUs), which have become the standard in high-computing enterprise data centers. Semiconductor analysts at TechInsights estimate that of the 3.85 million GPUs shipped to enterprise data centers in 2023, Nvidia will be responsible for all but 90,000.
Demand for the company’s chips overwhelmed supply, allowing Nvidia to significantly raise the selling price of its superstar AI acceleration chip, the H100. Over the five quarters, the company’s adjusted gross margin increased by about 13.7 percentage points to 78.4%.
The company’s hardware has become a hot topic in enterprise data centers, which has also fueled continued innovation. In March, Nvidia unveiled its next-generation Blackwell platform, which it says will accelerate computing power in many areas, including generative AI solutions, while consuming less power than the previous generation. In June, CEO Jensen Huang teased the debut of the company’s Rubin GPU architecture, which will feature a new processor called “Vera.” Rubin is expected to debut in 2026.
The final piece of Nvidia’s textbook expansion so far has been suppliers ramping up production capacity to meet strong demand. For example, Taiwan Semiconductor Manufacturing Co. (TSMC), the world’s leading chip maker, has (TSM 1.56%) The company has increased its chip-on-wafer-on-substrate (CoWoS) capacity, which is essential for packaging high-bandwidth memory in AI-accelerated data centers.
Nvidia is no longer perfect
This seemingly textbook recipe for leading the hottest trends on Wall Street helped Nvidia briefly surpass Microsoft and Apple to become the largest publicly traded company. But after a tough few weeks for Nvidia and the stock market as a whole, it’s clear that Nvidia, like any other company, can make mistakes.
To maintain its historic growth spurt, Nvidia needed flawless execution: It had to stay competitive by selling off all of its hardware, charging top prices for its products and software (its CUDA platform, which helps developers build large-scale language models), and bringing its next-generation GPU architecture to market on time.
Unfortunately, reports surfaced last weekend that Nvidia had informed many of its major customers (all members of the “Magnificent Seven”) that it would be delaying shipments of Blackwell chips by at least three months, pushing deliveries out to the first quarter of 2025, from a planned arrival date later this year.
Various reports have blamed the delays on potential design flaws at Blackwell and capacity constraints at Taiwan Semiconductor (TSMC). Even if TSMC were to effectively double CoWoS’s production capacity, it would still be far from enough for Nvidia to meet the demand of its enterprise clients.
Blackwell’s delay is the first domino to fall that shows Nvidia isn’t perfect. It also paves the way for Nvidia’s competitors to thrive.
On July 30, Advanced Micro Devices announced (AMD -1.50%) The company reported second quarter results that were well received by Wall Street and investors. AMD’s Data Center division revenue grew 115% year over year and 21% quarter over quarter (from the March quarter reported). AMD attributed the strong results to its growing AI GPU footprint.
Notably, AMD’s MI300X is significantly cheaper than Nvidia’s H100, and while the H100 has a number of computational advantages over the MI300X, a backlog of H100 supplies, coupled with the now-delayed Blackwell chips, makes AMD’s hardware more attractive.
Moreover, all four of Nvidia’s major customers are developing AI chips for use in their own data centers. Even if Nvidia were to maintain its computing dominance, it would lose valuable data center “real estate” as its major customers choose to deploy its own (more cost-effective) chips.
Image source: Getty Images.
History suggests Nvidia’s sale will be bad
To make matters worse, not a single next-generation big innovation, technology or trend in the past 30 years has escaped an early stage bubble. Since the advent of the internet, investors have seen early bubbles burst in areas such as genome sequencing, business-to-business commerce, housing, Chinese stocks, nanotechnology, 3D printing, blockchain technology, cryptocurrencies, cannabis, the metaverse, and more.
The problem with breakthrough innovations and technologies is that investors overestimate their adoption and usefulness. Regardless of the size of the addressable market, new technologies take time to change the environment for American companies to grow.
For example, even though most of the leading companies are investing heavily in AI-driven data centers, many of them don’t have a clear blueprint for how artificial intelligence can help them grow sales and revenue. Investors saw the same thing a few years ago with the rise of the metaverse, and before that with blockchain technology. Every innovation needs time to mature. There are no exceptions.
Over the past 30 years, the market leaders of every next-generation trend have consistently fallen in value by over 80%. Looking from peak to trough, the major businesses behind internet/networking, 3D printing, genome decoding, cannabis, cryptocurrency, and blockchain technology all plummeted by over 90% before hitting their troughs.
The hope for Nvidia is that the company has established presence outside of its AI GPU business that could give it a better footing than other market leaders have when their respective bubbles burst: Nvidia’s GPUs used for gaming and cryptocurrency mining, coupled with the company’s virtualization software, should prevent total collapse.
However, history has shown that once the excitement around the next big thing fades, a big pullback will occur. Blackwell’s delay was the first domino to fall, strongly suggesting that Nvidia’s selloff will worsen in the coming weeks, months and quarters.
