The AI industry was shocked in recent times with the release of the Chinese Startup Dippec’s open source AI model, which could only compete with US companies’ best AI models despite the cost of training only 6 million. Although it is possible that the claims about costs are exaggerated or flat -out, the Deep Sacred model is the actual deal.
Although cheap, powerful AI model AI is a huge thing for companies seeking AI, but it’s probably terrifying news nvidia (Nvda -0.11 %Jes. Bill Thesis for Nvidia, which dominates the market for powerful AI ACCELIPERS which is essential for training the latest AI models, depends on the assumption that more than to train and run every consecutive generation of AI model More computational horsepower will be needed.
The progress of the depressic has raised some serious questions. As a result of this uncertainty, an epic decision was made for the NVIDIA stock on Monday, which cleared hundreds of billions of dollars in market value.
Another shoe must still fall
A cheap AI model that is able to trade with the best models of Open AI and Entropic, which is not enough for NVIDIA growth story. Cheap and efficient AI models can promote more demand and eventually increase the sales of AI ACCELIVERS in the coming years. Former Intel CEO Pete Glysinger said in a post on X that “Computing bans gas law. Making it dramatically cheaper will increase the market for it.” Cheap AI model that does not require large -scale clusters of high -strength graphics processing units (GPU) to train and run. If it increases AI’s use a lot, it may not be negative for Nvidia –
However, there is another problem that causes it. NVIDIA’s multi -trillion dollar diagnosis is based on another important assumption: AI models will continue to be more capable as they are thrown into more computing resources. It was certainly true in the early days of AI, but it could not be true for long.
AI companies have eliminated the data used to train large -scale AI models, and new AI models are no longer moving forward in terms of performance. The founders of Anderson Horwests noted at the end of last year that improvement was diminishing, and AI models were hitting the roof.
It is important to remember how a large language model (LLM) works. Finally, these models predict the next token in the token stream. Just There is no reasoning, just the illusion of reasoning. Although the new technique can unlock better performance, it makes sense that the capabilities of this class of AI model will be a limit.
This is a potential roof combination in cheap AI models and abilities that will be a death sip for NVIDIA stock. If the AI models stop making meaningful improvement, regardless of how much computing power is thrown at them, and an advanced model can be trained cheaply on second -level hardware, this is a ballgium. Nvidia’s expensive AI ACCARS demands potentially fall off the mountain as soon as the AI bubble explodes.
Even under this scenario, AI will still be a useful and potentially game -changing technology for companies that look for good use. However, for NVIDIA, the GPU Gravio train will end, and tech companies are putting tens of billions of dollars in AI data centers.
A high -risk investment
NVIDIA is priced at more than $ 3 trillion. The company is selling a lot of valuable data center GPUs and enlightening incredible profits, but that depends on the two assumptions that appear as if they cannot be true. First of all, that AI models will need the increasing amount of computing power to train and run, and second, that their capabilities have no limit.
The cheapest AI model of Dippic represents a major crack in the NVIDIA growth story. It is not enough to remove the king from the track. However, when cheap AI models are combined with the real possibility that the AI models are not going to be more capable than here, NVIDIA’s Stratsfare market value is different.
I can’t predict the future, nor can you, but the uncertainty around the AI industry’s future makes NVIDIA an incredible stock.
Timothy is positioned in Green Intel. Motley Fool has positions and recommend Intel and NVIDIA. Motley Fool recommends the following options: Short February 2025 Calls on $ 27 Intel. Motley Fool has a diagnosis policy.
