After the shockwave sent through the artificial intelligence sector Monday following the DeepSeek earthquake, AI stocks recovered some of their lost ground. Nvidia (NASDAQ:NVDA) was the biggest winner, bouncing almost 9% higher to $129 per share just like it led the industry lower.
While the market continues to digest the meaning of the achievements made by the Chinese AI lab, investors shouldn’t rush back into NVDA stock just yet. There is still a lot of air beneath the semiconductor stock.
24/7 Wall St. Insights:
-
Nvidia (NVDA) rebounded sharply higher on Tuesday, leading the chip sector back from the collapse following DeepSeek’s AI model release.
-
There is still too much dust flying around to get a clear picture of where the industry will land as a result of this development.
-
Although this is healthy for the industry as a whole, the pricing reset is likely not finished and investors would do well to wait for a better value.
-
If you’re looking for some stocks with huge potential, make sure to grab a free copy of our brand-new “The Next NVIDIA” report. It features a software stock we’re confident has 10X potential.
Trending Video
The risk remains high
Nvidia is seemingly the most at risk from an AI model that is more streamlined, more efficient, and cheaper to train and run. The chipmaker’s valuation is largely based on its ability to sell larger numbers of gigantic clusters of pricey AI graphic processing units (GPUs). A new paradigm that says inexpensive, underpowered AI chips can do an even better job threatens the growth thesis.
Even though the long-term opportunity Nvidia was chasing was built on a long, wide runway, the valuation reset it and the rest of the AI sector experienced is healthy.
What DeepSeek promises is the democratization of AI. By making its large language model (LLM) open-source, it allows any developer to modify the algorithm, fine-tune it to their needs, and build upon its foundation.
Although its abrupt introduction caused reverberations throughout the AI sector, this is the event horizon for the technology that allows it to continue to grow and thrive.
A different way of modeling
DeepSeek’s R1 LLM took Nvidia H800 AI accelerators that were developed to get around technology export restrictions to China and used workarounds to exploit their inherent capabilities.
Unlike the H100 chips the western world used, the H800s suffered from significantly reduced chip-to-chip data transfer rates. It required the AI lab’s engineers to focus heavily on optimizing the training process that allowed for overlapping computation and communication phases, which significantly reduced training bottlenecks.
Instead of using Nvidia’s proprietary software platform CUDA, DeepSeek used its Parallel Thread Execution (PTX) instructions to fine-tune them for low-level optimizations that enhanced its efficiency. It also reconfigured the GPUs to squeeze performance out of their suboptimal design.
Ultimately, it challenges the industry’s reliance on the brute force of ever-increasing computational power.
Aftershocks will come
This is why Nvidia was shaken. Advanced AI models can be developed with less reliance on Nvidia’s high-end, high-cost flagship GPUs. If other companies follow suit by using fewer or less advanced chips for AI model training, this could significantly decrease the demand for Nvidia’s top-tier GPUs like the H100 or the forthcoming Blackwell accelerators.
The secondary impact is it could lead to lower API pricing for AI services. It indirectly pressures companies that heavily invested in Nvidia’s hardware to achieve similar results. If the market moves towards more efficient, cheaper AI solutions, Nvidia might have to adjust its pricing or see reduced sales volumes, particularly if companies decide to scale back on their hardware investments.
It also opens up the opportunity for rivals like Advanced Micro Devices (NASDAQ:AMD) to steal market share as their GPUs are often available at less cost.
Keep your powder dry
That’s why investors shouldn’t rush in just yet to buy NVDA stock. Despite the recalibration of its price, it still trades at a premium. The valuations implied revenue would grow to the moon. While that was already an unlikely scenario, it is a risk that is more palpable today.
The DeepSeek moment is something that has occurred time and again in the tech sector. For example, PC computing power became commoditized and pricing plummeted, allowing for it to proliferate and expand as a whole.
Even after NVDA stock’s downdraft, it still trades at 50 times earnings, 28 times sales, and 55 times the free cash flow it produces. Nvidia deserves a premium for its leadership position, one that will be eroded but not displaced. Yet its current valuation remains beyond what is reasonable.
The chipmaker will still be a winning investment, but you will be able to buy it later at a much better price.
Thank you for reading! Have some feedback for us?
Contact the 24/7 Wall St. editorial team.
Latest from 24/7
Our expert who first called NVIDIA in 2009 is predicting 2025 will see a historic AI breakthrough.
You can follow him investing $500,000 of his own money on our top AI stocks for free.