The emergence of Chinese artificial intelligence unleashes panic in the US

Since then, Andreessen – a father of Internet browsers, an early investor in technology giants, a lifetime Democrat – has become a Trump technological squire: Millionaire donor to his campaign and influential advisor to the new president.

Last Friday, Andreessen shared a message on social network X: «The R1 Deepseek is one of The most incredible and impressive advances I have seen in my life «. The achievement, however, goes in the opposite direction to the ambition that Trump presented last summer. Because Deepseek is a Chinese company Dedicated to artificial intelligence and R1, one of its groundbreaking tools, which questions American hegemony in the technological sector that begins to master our lives and that aims to dictate the future.

Deepseek’s thrust has unleashed panic in the US: from Silicon Valley to Wall Street, through the White House. In essence, the Chinese company has developed tools with a capacity comparable to that of the American giants of artificial intelligence, but with A much lower cost.

With fewer chips

The Chinese company launched v3, his ‘chatbot’ -System of user generation system- Based on artificial intelligence, on December 26. The surprising thing about creation was not only that it achieved a level of execution at the height of the tools of the American giants of the sector, as OpenAi or Googlewhich have made great advances with Chatgpt and Gemini, respectively. The relevant thing is that, according to the engineers of the Chinese company, they did it with A much lower number of chips of high capacity that large companies need to work their systems.

According to Depseek, they only needed 5.6 million dollars In computer infrastructure to build its V3, ten times less than what target – the technological giant that encompasses Facebook – has needed to develop its own artificial intelligence tool. Last year, Dario Amodei, CEO of Anthropic, another American large artificial intelligence, said that developing a capacity with a capacity in this sector has a cost of between 100 and 1,000 million dollars. The ‘chatbots’ of the big American companies use up to 16,000 chips for their artificial intelligence systems. Deepseek said it has only required 2,000.

Last week, the Chinese company presented R1, the tool that fascinated Andreessen, and that is dedicated to problem solving. These two models have perched on the top of the Chatbot Arena, a platform developed at the University of California in Berkeley, in which researchers score the performance of this type of artificial intelligence tools. Both R1 and V3 are in the ‘Top 10’. The leader is a Google Gemini model, but Chinese inventions exceed Anthropic and Grok, the artificial intelligence tool developed by Elon Musk.

Deepseek’s success is undeniable and is found at the level of users: its V3 is currently the mobile application more downloaded In the iPhone store.

Lower cost

The finding that China is achieving similar results In artificial intelligence with much lower cost it has meant an earthquake in the US the actions of the technological world power, with multimillion -dollar investments in the sector, they cleared in the start of this Monday’s session. He Nasdaqthe reference index of the technological sector in the US closed with a 3.07%drop. That happened with much more force in central companies in the sector, such as Nvidiawhich develops the high capacity chips with which artificial intelligence systems are created.

The chips giant action sank 17% at the close of the session, erasing about 600,000 million dollars of its market value. The possibility that advanced artificial intelligence systems can be developed with fewer chips cost it very expensive.

The concern, however, went far beyond Silicon Valley and Wall Street. Deepseek’s success question the US strategy to try to hinder the development of Chinese artificial intelligence with limitations to the export of chips.

Battle USA.-China

A central aspect of the battle between the US and China in recent years have been the limitations imposed on the arrival of chips – specially, those developed by Nvidia, the world leader in the Sector- to the Asian giant. Earlier this month, the Joe Biden government issued its last order in this regard, in an attempt to prevent China from accessing chips through third countries. Washington has always justified these export limitations so that China does not use chips for purposes militaryalthough the reason has more to do with winning the technological battle against the thrust of its great rival.

These clippers have been able to be counterproductive: the shortage of chips in their hands has been able to push Chinese engineers to be more creative and make the need virtue. Now they have a capable and cheap technology in their hands.

Panic for Depseek’s success occurs in full landing of Trump in the White House, with great promises to take a “golden era.” The US president feels more comfortable talking about large industrial sectors of the past –petroleum, automotive -but has given signs that he is convinced that the US needs to win the battle of innovation. He has surrounded himself with technologists – from Musk, an inseparable figure, Andreessen, but also others such as David Sacks, his ‘tsar’ for artificial intelligence – and his first great economic announcement was a multimillion -dollar investment plan in this sector: 500,000 million Dollars in four years, an effort led by private sector giants as OpenAI with the objective that the US is ensured in the artificial intelligence of the future.

But that effort began with transpiés: The confrontation Between Musk and Sam Altman, the leader of Openai, fought for a long time, in an episode that sowed doubts about the American project entity. And reinforces fears On the Chinese thrust.