Here are the worst scenario for NVIDIA stocks

AI industry has been shaken in recent days by launching The Open Code AI model from Chinese bootable Deepseek This can compete with the best AI models from US companies, although it supposedly costs only $ 6 million for training. Although the allegations of costs may be exaggerated or leveled false, the Depepeek model seems to be a real deal.

Although cheap, powerful AI models are a great thing for companies that want to unfold AI, that’s a potentially terrible news for Nvidia (S)Nvda 8.93%))S The Bull Thesis for NVIDIA, which dominates the market for powerful AI accelerators that are needed to train the most modern AI models, relies on the assumption that any subsequent generation AI models will require more and more computing horsepower and Running.

Deepseek’s breakthrough raises some serious questions. This uncertainty led to an epic immersion for NVIDIA’s shares on Monday, which deleted hundreds of billions of dollars market value.

Another shoe still has to fall

The cheap AI model, capable of trading the best models from Open AI and Anthropic, is not enough to look at the NVIDIA growth history. Cheap and effective AI models can encourage more demand and ultimately stimulate AI sales accelerators in the coming years. Former Intel Executive Director Pat Gelsinger said in an X publication that “the calculation of the Gas Act. Making it dramatically more cheaper, it will expand the market for it.” Cheap AI models that do not require massive clusters of powerful graphic processors (GPU) for training and running may not be negative for NVIDIA if it significantly expands the use of AI.

However, there is another problem that lurks. NVIDIA’s multimillion-dollar valuation is based on another important assumption: AI models will continue to become more capable as more computing resources are thrown on them. This was certainly true in the first days of AI, but it may not be true for a long time.

AI companies have largely exhausted the data used to train AI models, and the new AI models no longer jump forward in performance. The founders of Andreessen Horowitz noted at the end of last year that improvements slow down and AI models can hit the ceiling.

It is important to remember how large language models (Llms) work. In the end, these models predict the next sign in a stream of tokens. That’s all. There are no reflections, only the illusion of reasoning. Although new techniques can unlock improved performance, it makes sense that there will be a limit to the capabilities of this AI model class.

This is the combination of cheap AI models and this potential ceiling in the possibilities that would be death for NVIDIA stocks. If AI models stop improving meaningfully, no matter how many computing power is thrown on them and the highest-level model can be cheaply trained in the second class hardware, this is the ball. The search for expensive AI accelerators of Nvidia is likely to fall off a rock when the AI ​​balloon bursts.

Even with this scenario, AI will still be useful and potentially changing playing technology for companies that find cases of good use. However, the GPU Gravy train for NVIDIA would be over, and the technology giants that pour tens of billions of dollars into AI data centers will probably never recover these investments.

High -risk investment

Nvidia is still valued at more than $ 3 trillion. The company sells very expensive graphics processors for data centers and displays incredible profits, but this assessment depends on two assumptions that begin to look like they may not be true. First, AI models will require ever -increasing quantities of computing power to train and release, and second, that there is no ceiling in their capabilities.

The cheap AI model of Deepseek is a large crack in the history of NVIDIA growth. It is not enough in itself to derail the King of the II. However, when the cheap AI models are combined with the very real perspective that AI models simply will not become much more capable than here, the basis for the stratospheric market value of NVIDIA is falling apart.

I cannot predict the future, as do you, but the uncertainty about the future of the AI ​​industry makes Nvidia an incredible risky stock.