NVIDIA 23% Below Record High: 6 Reasons I’m Still Not a Buyer Over the past month, Wall Street has been issuing a stark warning to the investor community that stocks could indeed decline. It took only 16 trading sessions for the benchmark S&P 500 index to enter correction territory—down at least 10% from its recent closing high—with the Nasdaq Composite entering correction territory at an even faster pace.
While stock market corrections tend to affect most sectors and industries, few of the most influential companies on Wall Street have been hit harder than AI giant NVIDIA (NVDA -1.74%), which is now 23% below its record closing high of $149.43 (reached on January 6).
There’s no doubt that NVIDIA’s graphics processing units (GPUs) have played a pivotal role in driving the AI revolution. The Hopper (H100) GPU and the next-generation Blackwell GPU architecture are the foundations for rapid decision-making, training large language models, and powering generative AI solutions.
Most investors, along with the vast majority of Wall Street analysts, see this 23% drop in NVIDIA stock as a surefire buy, but I’m not among them. Here are six reasons why I’m moving away from Wall Street’s leading AI company.
Increasing Internal Competition
First, I’m not convinced NVIDIA can maintain its monopoly market share in AI data centers much longer. Even if the Hopper and Blackwell GPUs easily maintain their compute speed advantage, gaining a prominent position in AI data centers goes beyond speed. Price and accessibility are also important.
Nvidia is starting to lose data center space to its key customers
While most investors focus on Nvidia’s direct competitors ramping up production (such as Advanced Micro Devices), I see internal competition as a far greater threat to its growth potential and financial results. Many of its major customers in terms of net sales are internally developing AI chips for use in their data centers. While their hardware doesn’t quite match Blackwell’s compute capabilities, the internally developed chips are significantly cheaper and not as stacked as Nvidia’s. I fully expect Nvidia to start losing data center space to its key customers in the coming quarters.
This brings me to my next point: the scarcity of AI-powered GPUs has played a crucial role in Nvidia’s success. Demand outstripping supply allowed Nvidia to raise the price of its Hopper chip to over $40,000 early last year. For comparison, AMD priced its Instinct MI300X, an AI-accelerated chip, between $10,000 and $15,000 per unit.
Thanks to this combination of exceptional pricing power and the scarcity of GPUs and AI, NVIDIA’s gross profit margin surged to over 78% in the first fiscal quarter of 2025 (ending April 28, 2024). However, in each subsequent quarter, NVIDIA’s gross profit margin declined. The company reported a GAAP gross profit margin of 73% for the fourth fiscal quarter (ending January 26, 2025) and forecast a GAAP gross profit margin of 70.6% (+/- 50 basis points) for the first fiscal quarter of 2026.
We are witnessing NVIDIA’s greatest competitive advantage—the scarcity of AI-enabled GPUs—disappearing before our eyes. As this happens, NVIDIA’s gross profit margin will further erode.
Tariffs could disrupt business.
Adding fuel to the fire, President Trump’s tariff policy creates uncertainty. The president has imposed a cumulative 20% tariff on select imports from China and has threatened to impose a tariff of approximately 25% on various semiconductor imports. This would raise infrastructure costs associated with building and/or maintaining an AI-accelerated data center.
The biggest potential concern is the impact of deteriorating trade relations between the world’s two largest economies by GDP on NVIDIA’s sales and profits. If President Trump’s tariffs lead to a severe dispute with China, companies in the world’s second-largest economy may choose to bypass NVIDIA altogether and support domestic chipmakers.
Upcoming Major Investments Have a Checkered History
While Wall Street offers no guarantees, I am a student of history. This means I firmly believe in the rhythm of history.
Since the rise of the internet in the mid-1990s, every major technology/innovation has made its way through a boom bubble. This indirectly means that investors consistently overestimate how quickly a new technology will spread and gain widespread adoption by businesses and consumers. Ultimately, these high expectations go unmet, leading to a boom bubble.
While AI has shown every sign of being a disruptive technology in the long run, it is still far from mature at the moment. Most companies lack a clear business plan to generate a positive return on their AI investments and have not optimized their existing AI solutions.
If the AI bubble bursts, no company will feel the impact more directly than NVIDIA, which generated over 88% of its annual sales in fiscal year 2025 from the data center segment.
NVIDIA’s valuation has plenty of room to shrink.
The sixth and final reason I’m not buying NVIDIA stock after its 23% drop from its January closing high is its valuation. I’m sure some will find this statement surprising. NVIDIA’s rapid earnings growth over the past two years has placed its stock at a price-to-earnings multiple of 20 times its expected annual earnings per share. This is roughly in line with the S&P 500’s expected price-to-earnings multiple, and significantly lower than a year ago.
NVIDIA’s price-to-sales (P/S) ratio remains valued at 22 times its sales. While this P/S ratio is notably lower than its all-time high of 42.39 recorded in June 2024, many of the “Magnificent Seven” components have P/S ratios of 12 or lower. Historically, P/S ratios between 30 and 40 have been the highs for companies on a bullish earnings trend. This is another case where historical precedent suggests NVIDIA’s stock is set to fall much further. Nvidia plans to invest hundreds of billions of dollars in US-made chips and electronics over the next four years, the Financial Times reported, citing CEO Jensen Huang.
The AI chip giant expects to spend about half a trillion dollars on electronics over the four-year period, according to the report. “I think we can easily see ourselves making hundreds of billions of them here in the US,” Huang told the Financial Times, adding that the Trump administration could help accelerate the expansion of the US AI industry.
Huang is working to calm investor concerns about demand for Nvidia’s expensive AI chips, which have made the company one of the world’s most valuable companies, after Chinese company DeepSeek launched a competing chatbot with fewer AI chips.
