
Deepseek, China “low cost” that puts fear in the body to Silicon Valley
The “boom” of artificial intelligence (AI) has led to the stock market contribution of large technology companies (mostly from the United States) has been put through the clouds during the last two years. However, DeepSeeka Chinese startup, He has put brake to Idilio (Maybe I came out of mother) Of the stock markets with the “Big Tech” of Silicon Valley thanks to a “low cost” that ufana in addition to being extraordinarily efficient.
The entry into the scene of Deepseek, whose AI models has been developed at a price that is just a very small fraction of what Chatgpt and company cost, has triggered an earthquake of many degrees on the Richter scale in the stock market. The parking of the financial resources that Depseek has used to tighten its AI models has left investors in shock. And his surprise (pregnant with joy) has caused for the time Nvidialeader in the manufacture of AI chips, have suffered a prolapse of up to 17% in the stock market parks. This remarkable decrease in the value of the Nvidia titles has been in the dessert in a decrease of more than 500,000 million dollars in the stock market value of the US technology company, a record (in no way positive) in the stock market .
Depseek could change the paradigm for which the development of AI models was governed so far
Until the inflictive irruption of Deepseek, it was assumed that the training of AI models should necessarily rely on huge amounts of computational power (which is the one that provides precisely Nvidia to companies such as OpenAi, Meta or Google). Deepseek has demonstrated, however, that the development of AI models can be substantially cheaper and this has caused Nvidia titles in particular to be bleeding in stock market parks. Chips manufactured by the company led by Jen-Hsun Huang have so far played an absolutely key role in the development of AI. And investors logically assumed that the unstoppable strength of AI was going to continue reporting (virtually infinite) benefits to Nvidia, which today logically contemplates Deepseek as a serious threat.
Only one week OpenAi and several partners announced among large alharacas that were going to invest a whopping 500,000 million dollars in the construction of new AI data centers in the United States. Deepseek is, however, of having trained R1, its new AI model (whose performance is very similar to that of its most direct rivals), with a budget of less than 6 million dollars and less powerful chips to those usually used in the development of this technology. It is unknown, however, to what extent are Deepseek’s assertions, but the truth is that, regardless of their veracity, they have already caused a real tsunami in the bag.
The stock market chaos unleashed by Depseek’s entry has reminded some of the Puntocom Bubble burst in March 2000when suppliers like Cisco suffered a strong varapalo in the bag after it was evident that the business of a good number of startups born to the heat of the Internet was entirely unsustainable.
Thus, 25 years after the Puntocom bubble exploded, few doubt that the applications of AI will transform the economy, the legal system, health and society in general. However, The success of DeepSeek It inevitably puts on the table the question of whether the transition to a society governed by AI could actually be perfectly possible at a much smaller cost than was anticipated a priori.
Is it the face as expensive as we presupposed in the beginning?
If Depseek’s models prove to be efficient and really require less computational power, Perhaps it would be entirely unnecessary to invest 500,000 million dollars in the construction of new data centersas announced last week OpenAi and the partners that are part of the Stargate project (which has the support of President Donald Trump).
Experts have been surprised in any case that Deepseek’s models have obtained a performance comparable to those developed by US companies such as Openai and Meta. A fact that attracts even more powerfully attention if we take into account that the Chinese startup does not have access to the most modern and advanced market chips in the market due to the restrictions imposed by the United States government to China, its country of origin.
And not even the fact of Depseek is clearly at the mercy of the censorship of the Chinese Executive has managed to contain the excite. R1, the Deepseek AI model, does not facilitate, for example, any information when confronted with questions about the massacre in the Tiananmén Square in June 1989.
“Suddenly the high stock quotes of some technologies companies would cease to be justification and the decision to invest 500,000 dollars in AI centers could be totally wrong,” says Jürgen Molnar, expert in Robomarkets markets.
Meanwhile, and surpassed for his sudden fame, DeepSeekwhose AI model is open source, He had to limit the new records in his app yesterday claiming that he had been the victim of several cyber attacks.
R1, Deepseek’s model, saw the light a few days ago and immediately captured the attention of Silicon Valley, particularly the famous Marc Andreessen investor. Last Friday Andreessen called Deepseek as “one of the most impressive advances” of those who had witnessed to date.
The lit that Andreessen dedicated Deepseek spread rapidly in the network of networks. And during the weekend Depseek climbed to number one of the free applications in Apple’s App Storerelegating ChatgPT, the OpenAI chatbot, to the second position.
Discover more from CiptaVisual
Subscribe to get the latest posts sent to your email.