China’s irruption – a big way – in the artificial intelligence market (AI) has caught many by surprise. Only a few hours had passed since the leaders of the Nasdaq technological giants had attended the inauguration of Donald Trump as president of the United States. In that hangover, the masters of the world still believe, handling the threads of a quasi monopoly and with the blessing of the most powerful man on the earth, they received a good handling to their pride. The new Deepseek tool put everything upside down. And on top of that it was free and open source, leaving them even more evidenced.
Some of them surrender to evidence. Satya Nadella, CEO of Microsoft, described the new tool as “super impressive” and advised to take the developments from the Asian giant. Sam Altman, CEO of Openai, creator of Chatgpt also, also described it as “impressive.” Elon Musk was suspicious of the limited resources that Deepseek said to achieve those results. And shortly after he launched a purchase offer by OpenAI and then his own AI system: Grok 3.
The truth is that China had already announced a strategic plan for a long time to become the year 2030 in the IA world leader. And he is doing homework. Not surprisingly, if the greatest driver in this technology until now was the United States, with 30% of investments, China steps on their heels with 28%. And they are collecting the fruits: they already treasure 40% for AI patents, while the United States has to settle for one in ten.
In addition, Depseek could have made those advances and have overcome ChatgPT in number of downloads and power based on much less resources and investments. According to the alleged startup of the same name as the tool, 2048 Nvidia H800 graphics cards were enough to develop it. Only that would be equivalent to an investment of $ 5,600,000 (5,470,000 euros), a ridiculous amount if we compare it with the more than 100 million dollars (97.6 million euros) that Openai needed to train chatgpt4. We have been using in this paragraph the verbal time of the conditional because, in the case of China, all the experts consulted put the truth of these ads very questioned.
No one believes it
“We have to do an act of faith to believe that they have developed it with less budget. There is much to know the truth, or perhaps we will never know. Depseek can be the Wuham of AI, where we will probably never know exactly what happens there,” , explains Rafael Tamames, founder of Vivid Vision and author of the book ‘Artificial intelligence and you’. For Chuck Herrin, Director of Information Security (CISO) of F5, the contributed figure is “quite misleading”, although it does recognize that the efficiency with respect to other models is very substantial. “There is a lot of infrastructure that is not included in that figure. That is a bit like calculating how much it costs to make a car without counting all the investments made in the factory that comes out,” he explains to the elect.
For Alberto Pascual, president of Aslan, “Depseek’s irruption comes at a time when some voices began to show concern because the development of AI could have touched the roof.” “The exponential growth in energy needs and restrictions on data access for models training began to be limiting. The entire ecosystem was looking for alternatives. If Deepseek confirms that greater efficiency is possible, it will mean a great stimulus to the IA’s continuity, and its associated industry, “he explains.
Javier Cuervo, Professor of the Degree of Un University, remembers that China was making great investments such as Government, as a nation, in Ia. “In fact, it is the government that invested the most and continues to invest in this technology. This has been evidence Deepseek, nor do we know your accounts, nor its evolution, to know if it is an emerging company or not. ” For Alberto Pascual, president of Aslan, China has always had a privilege position in the field of AI. And introduces another element to take into account: “being so far the availability of data a determining factor in its development, having the supply, enthusiastic and without restrictions, of a population of more than 1500 million individuals undoubtedly supposes a competitive advantage Obvious, “he says. This reflection clarifies that hegemony in this sector is not measured only by the volume of the investments made. For Omar Benboazza, Dircom de Rootedcon, “it was a matter of time that China hit the table like this.” “It is a country where more than 2.5% of its GDP is intended for research and development, where they currently carry out more than 40% of world research,” he adds.
End of monopoly USA
Can we talk about the end of the United States technological monopoly in this field? How is the overall map of AI with the emergence of these new Chinese models? Benbouazza is convinced that China “can already look at you from you to the United States.” Tamames admits that “we don’t know anything about how China really works” because “there is no business or economic transparency.” But he also recognizes that “they have at least technological capacity to match the Americans.” And remember that we know for years that China is going from copying to innovate, “but we don’t know to what degree or in what dimension.”
Herrin, from F5, admits that everyone is in a race to adopt AI, from small businesses to large companies, the governments of all countries. “Everyone feels that if we don’t do it, they will do it, and everyone has at least a ‘they’ to worry about,” he says. The author of the book “ Artificial intelligence and you ‘remember that “competition is always good and that the one that comes out reinforced is the consumer and companies.” “This does nothing but demonstrate how important the AI and the market it is generating. If it will be proven that Depseek does the same and cheaper, better for everyone.” The Rootedcon DIRCOM provides another interesting point to the debate: “The Chinese tool generates a new paradigm and, having done it Open Source or open source, they are guaranteed that there will be many projects based on Depseek.”
A different model
Apart from having published the code and being free from beginning to end, there is another remarkable feature in Deepseek. “He works with a mixture of experts. If we ask for a melody, let’s say he puts different musicians to compose it and use the specific musicians that are best taken with that melody. Simplifying things, it is a way of working on AI quite novel, not completely novel, but quite novel, “explains Professor Javier Cuervo. About the risks warns the F5 CISO: “One of the most worrisome aspects of Depseek’s releases, and that will probably be maintained in the new releases of these models, is how easy it is to hack them and use them for malicious/destructive purposes.”
At this point, we also ask ourselves if the restrictions to the Asian country from the West have been able to give more wings to China. The boycots to Huawei, zte or more recently to Tiktok perhaps fed those desire for overcoming. Chuck Herrin understands that “perhaps many of the limitations imposed on China were catalysts that led to true advances. The most notable innovations largely refer to efficiency and cost reduction, which is what people undergoing to sanctions “.
In line with this, the president of Aslan also reminds us of what happened in the cold war of the last century. “Great technological disruptions emerged, such as those that led man to the moon, or those that made the internet possible. We now try to simplify thinking again of a bipolar competition, China-Use Be very attentive to the advances of India, the movements of Russia and even of the Arab countries that, in many cases, already have even dedicated ministries exclusively to the development of AI “.
What happened with Depseek and Chatgpt also reminds other discoveries in which the ingenuity could more like gross force. Alberto Pascual recalls what happened to Navteq, a company acquired by Nokia at the time he tried to develop solutions for the optimization of urban traffic through the deployment of millions of sensors. “Then, an Israeli startup thought that we all carry a GPS on our mobile and organizing through a platform to work together in that improvement in urban tours won as a win. That startup was Waze.”
And the EU, spectator
We ask all these experts about how the new geopolitical map of AI is and in what role the European Union is. The answers are quite bleak. “Europe is a stone guest. Continuously spending money from our taxes and time on regulations that make no sense. We are being second citizens, technologically speaking. They expel talent, they condemn us to be poorer and consumers to be less free “Rafael Tamames says sharp. For Omar Benbouazza, “Europe is once again relegated and focusing its efforts to legislate and not promote development and creativity. If you do not act immediately, we will be mere spectators.” Herrin, from F5, is unlikely to leave a project from the institutions themselves. “If we see any great jump in innovation I think it will be at the hands of some intelligent person or a small group of intelligent people with a great idea that uses the existing AI to make a leap forward.”
Javier Cuervo, from university, agrees that he does not see the EU capable of surprising with something at this level and highlights other very visible consequences of that regulation in the old continent. “All this has made Openai, the chatgpt company, has not launched its Sora videos generator in Europe, due to the regulatory risk that it is. Sora can be used throughout the world, except in Europe. The same is happening with What for me is the real surprise, or what will mark more this 2025 in AI, which are the agents, and specifically the operator of OpenAi, “warns. Thus, after the generation of texts with chatgpt, of images with Dall e or of videos with Sora, Operator has the ability to do homework for us. This Professor of the Degree of In will university puts several examples: “We can tell you that it responds to our emails, that orders our conversations into folders, or that draws extracts from our bank and orders them in an Excel table. We can use this Agent to automate tedious tasks. admits.
If something is clear about this debate is that the proliferation of these tools of AI in general will continue to accelerate, especially as improved performance in smaller and cheap models. The lower the cost, the more cases of use are raised. In this regard, Chuck Herrin concludes that “we have to try to prepare for unlimited free intelligence and all its implications.”
#China #controls #patents #front #ten
