Roughly 30 years ago, the internet began going mainstream and completely changed the way businesses operate. Although it took some time before companies fully understood the potential of business-to-business e-commerce and the power of virtual storefronts, the internet was nothing short of a game-changer for corporate America.
Since the mid-1990s, Wall Street and investors have been waiting for the next game-changing technology, innovation, or trend to provide the next leap forward for businesses. Artificial intelligence (AI) looks as if it’s answered the call.
Image source: Getty Images.
The lure of AI-driven software and systems is that they can become more proficient at their assigned tasks over time, and learn new skills without human intervention. In theory, this gives the technology utility in almost every sector and industry around the globe. It also likely explains why the analysts at PwC believe AI will add $15.7 trillion to the global economy by the turn of the decade.
While no public company has benefited more directly from the AI revolution than Nvidia(NASDAQ: NVDA), the catalyst most responsible for its gains is liable to disappear in 2025.
Nvidia’s fortunes shifted as a result of its ultra-popular AI graphics processing units (GPUs). These are effectively the brains of enterprise data centers that allow for split-second decision-making. Orders for Nvidia’s H100 GPU (commonly known as the “Hopper”) are backlogged, while CEO Jensen Huang has labeled demand for the successor Blackwell GPU architecture as “insane.”
The laws of supply and demand clearly state that if the demand for a good or service outweighs supply, prices will rise until demand tapers. Nvidia’s Hopper chip has commanded a price of between $30,000 and $40,000, which represents anywhere from a 100% to 300% premium over competing AI-GPUs. Being able to sell its GPUs at a premium has lifted the company’s adjusted gross margin to north of 75%, as of the quarter ended July 28.
Don’t overlook the role Nvidia’s CUDA software platform has played, either. CUDA is the toolkit developers use to build large language models and get as much computing capability as possible out of their Nvidia GPUs. CUDA is keeping Nvidia’s customers contained within its ecosystem of products and services.
The final puzzle piece for Nvidia has been its ability to land big orders from Wall Street’s most influential businesses. When this calendar year began, approximately 40% of Nvidia’s net sales could be traced to Microsoft, Meta Platforms, Amazon, and Alphabet. There’s no better advertising or ringing endorsement than America’s top businesses wanting to use your hardware in their AI-accelerated data centers.
But while Nvidia’s ascent has been textbook, it’s my belief that its biggest competitive edge will go away next year.
In one respect, Nvidia is in no danger of losing is its computing superiority. The company’s Hopper and Blackwell GPUs should have no trouble sustaining their computing edge in AI-focused data centers for the foreseeable future.
However, the advantage I fully expect Nvidia will be saying goodbye to in the new year is AI-GPU scarcity.
As noted, demand for AI-accelerated chips has been off the charts. Unfortunately, the supply of these chips has been constrained. Leading global chip fabrication company Taiwan Semiconductor Manufacturing(NYSE: TSM) has been increasing its chip-on-wafer-on-substrate (CoWoS) capacity, which is a necessity for the packaging of high-bandwidth memory needed in AI-accelerated data centers. Throughout 2023 and most of 2024, this increase in CoWoS capacity simply hasn’t been enough to satiate demand. With orders overwhelming supply, Nvidia has enjoyed otherworldly pricing power.
Image source: Getty Images.
However, 2025 looks as if it could be the perfect storm that derails the AI-GPU scarcity and pricing power that have worked in Nvidia’s favor.
Based on a report from the Taiwanese media, along with a forecast from analyst Zhan Jiahong of Morgan Stanley, Taiwan Semiconductor looks to be ahead of schedule in its efforts to increase CoWoS packaging capacity to 80,000 wafers per month. Rather than achieving this feat in 2026, as planned, it may occur a full year ahead of schedule. This means more high-powered GPUs being available across the board.
Something else to consider is that Nvidia is no longer the only rodeo in town. Advanced Micro Devices is ramping up production of its MI300X GPUs, and it recently unveiled its next-generation MI325X chip, which it intends to begin producing before the end of this year. As external competition picks up, AI-GPU scarcity will wane.
It’s not just external competitors that have the ability to pressure Nvidia’s pricing power. The four “Magnificent Seven” companies that account for roughly 40% of its net sales are all developing AI-GPUs for use in their respective data centers. Even though these internally developed chips can’t match the computing potential of Nvidia’s hardware, they’re considerably cheaper and more accessible. In other words, there’s a value proposition that should, in the coming quarters and years, lead Microsoft, Meta, Amazon, and Alphabet to choose their own AI-GPUs over Nvidia’s.
This perfect storm will steadily increase the availability of AI-GPUs in 2025, thereby reducing the stratospheric pricing power Nvidia has used to pump up its gross margin. When this competitive edge for Nvidia goes away, it could be nearly impossible for its shares to sustain their near-parabolic rise since 2023 began.
Before you buy stock in Nvidia, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Nvidia wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Nvidia made this list on April 15, 2005… if you invested $1,000 at the time of our recommendation, you’d have $867,372!*
Stock Advisor provides investors with an easy-to-follow blueprint for success, including guidance on building a portfolio, regular updates from analysts, and two new stock picks each month. TheStock Advisorservice has more than quadrupled the return of S&P 500 since 2002*.
John Mackey, former CEO of Whole Foods Market, an Amazon subsidiary, is a member of The Motley Fool’s board of directors. Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool’s board of directors. Randi Zuckerberg, a former director of market development and spokeswoman for Facebook and sister to Meta Platforms CEO Mark Zuckerberg, is a member of The Motley Fool’s board of directors. Sean Williams has positions in Alphabet, Amazon, and Meta Platforms. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Amazon, Meta Platforms, Microsoft, Nvidia, and Taiwan Semiconductor Manufacturing. The Motley Fool recommends the following options: long January 2026 $395 calls on Microsoft and short January 2026 $405 calls on Microsoft. The Motley Fool has a disclosure policy.
