For more than two years, Wall Street's bull market rally has been virtually unstoppable. While there are a confluence of factors pushing wind into the stock market's sails, including better-than-anticipated corporate earnings, stock-split euphoria, and Donald Trump's Election Night victory, no catalyst has been more pivotal to Wall Street's success than the artificial intelligence (AI) revolution.
Software and systems empowered with AI can become more proficient at their assigned tasks, and more importantly have the capacity to learn new skills without the need for human assistance. This a game-changing leap forward for businesses that can be utilized in most industries around the globe.
Image source: Getty Images.
It's also a technology with a tantalizing addressable market. In Sizing the Prize, the analysts at PwC laid out their case for a $6.6 trillion boost to global gross domestic product (GDP) from productivity gains by 2030, along with another $9.1 trillion in consumption-side effects. In total, PwC anticipates AI will lift global GDP by 26% come 2030.
Though dozens (if not hundreds) of stocks have benefited from the rise of AI, none has been a more direct beneficiary than semiconductor behemoth Nvidia (NVDA 1.00%). But as this week's events have reminded investors, history has a flawless track record of putting next-big-thing technologies in their place.
Nvidia's ascension has been historic
When 2023 began, Nvidia was on the fringe of being a key technology company. It entered the year worth $360 billion and was prominently known for its graphics processing units (GPUs) used in personal computers and for cryptocurrency mining. However, the creation of GPUs specific for high-compute data centers has been a game changer.
Nvidia's Hopper (H100) chip and successor Blackwell GPU architecture are the undisputed top options for businesses wanting to run generative AI solutions and build/train large language models.
NASDAQ: NVDA
Key Data Points
With orders for the Hopper backlogged and demand for Blackwell labeled as "insane" by Nvidia CEO Jensen Huang, the company has had no trouble increasing its prices. This combination of high demand and AI-GPU scarcity has allowed Nvidia to charge $30,000 or more for its Hopper chips, which is more than double what businesses are paying for Advanced Micro Devices' Instinct MI300X AI-accelerating chips. The end result is Nvidia's gross margin pushing as high as 78.4% on a quarterly basis.
In addition to a lengthy backlog for its hardware, Nvidia is snagging orders from America's most-influential businesses. Many of its largest clients by net sales are members of the "Magnificent Seven," such as Microsoft, Meta Platforms, Amazon, and Alphabet.
It took less than two years for Nvidia to add more than $3 trillion in market value and become Wall Street's most-valuable publicly traded company. However, the arrival of DeepSeek reminds investors that next-big-thing technologies have an ominous early stage track record.
Image source: Getty Images.
History is undefeated when it comes to next-big-thing innovations
On Monday, Jan. 27, artificial intelligence-stock investors were given a rude awakening. The innovation-driven Nasdaq Composite shed more than 600 points (a little over 3%), with key AI stocks leading the market lower. The AI golden goose, Nvidia, lost 17% of its value, equating to $593 billion in market cap. It's the largest-ever nominal daily decline in market cap for a publicly traded company.
The culprit behind this selling is China-based DeepSeek, which has created an open-source large language model (LLM) similar to OpenAI's ChatGPT.
The buzz is that DeepSeek allegedly trained its LLM using less-powerful Nvidia GPUs, and did so for a fraction of the cost that Magnificent Seven companies are spending on their AI-accelerated data centers. In other words, investors are concerned that 1) companies won't have to spend as much on AI hardware to train LLMs, and 2) Nvidia's computing advantages aren't as rock-solid as they once appeared -- especially if less-powerful AI chips can be used to train LLMs.
Regardless of whether DeepSeek's claims about the type and number of GPUs used to develop its LLM are accurate, it's a reality check for Wall Street in the sense that competition is an inevitability in the AI space. For instance, even if Nvidia's GPUs were to sustain their computing superiority for years to come, minimizing the AI-GPU scarcity over time that had propelled its pricing power and gross margin would be a decisive negative for a company that was, arguably, priced for perfection.
Nvidia's gross margin should decline further as AI-GPU scarcity wanes. NVDA Gross Profit Margin (Quarterly) data by YCharts.
The more prevailing concern of the DeepSeek saga is that it reinforces the idea of history being undefeated when it comes to next-big-thing technologies, innovations, and trends.
Roughly 30 years ago, the internet began going mainstream and opened new doors for businesses to expand their reach beyond physical storefronts. Eventually, this technology positively altered the growth arc of American businesses. But this wasn't without the dot-com bubble taking place, which sent the Nasdaq Composite tumbling by 78% on a peak-to-trough basis.
Since the advent of the internet, numerous next-big-thing technologies and trends have endured early stage bubble-bursting events, including genome decoding, businesses-to-business commerce, nanotechnology, 3D printing, blockchain technology, cannabis, and the metaverse.
Aside from mouthwatering addressable markets, what each and every one of these former next-big-thing trends had in common was unrealistic investor expectations. For three decades, investors have been consistently overestimating the early innings adoption rate of new technologies/trends, as well as their utility.
The emergence of DeepSeek's LLM from out of left field is a plain-as-day reminder that a lot of businesses are throwing money at AI without having anything close to firm game plan as to how they'll monetize their investments. Although use cases exist on paper, most businesses aren't yet benefiting from their AI investments. Historically, this is indicative of a bubble.
DeepSeek isn't just a threat to Nvidia. It's a reminder to direct and ancillary AI players that there's a lot we still haven't learned about AI, and all game-changing technologies need ample time to mature.
