Skip to content

What is Deepseek? And how are you going to AI?

    Tech stocks tumbled. Gigantic companies such as Meta and Nvidia were confronted with a barrage of questions about their future. Tech leaders went to social media to proclaim their fears.

    And it was all because of a little-known start-up of Chinese artificial intelligence called Deepseek.

    Deepseek caused waves all over the world on Monday as one of its performance – that it had created a very powerful AI model with much less money than many AI experts who were thought possible – raised a large number of questions, including or American companies Even competing in AI more.

    Deepseek is 'Ai's Sputnik -Moment', Marc Andreessen, a Tech -Venture capitalist, who was posted on social media on Sunday.

    How could a company about which few people had heard such an effect?

    Deepseek is a start-up and owned by the Chinese share trading company High Flyer. The goal is to build AI technologies in the lines of the Chatgpt Chatbot from OpenAi or Google's Gemini. By 2021 Deepseek had taken over thousands of computer chips from the American chip maker Nvidia, who are a fundamental part of any attempt to create powerful AI systems

    In China, the start-up is known for taking young and talented AI researchers from top universities, promised high salaries and an opportunity to work on advanced research projects. Both high flyer and deep chat are run by Liang Wenfeng, a Chinese entrepreneur.

    In recent years, Deepseek has released various large language models, which is the type of technology that supports chatbots such as Chatgpt and Gemini. On January 10, it released its first free chatbot app, which was based on a new model called Deepseek-V3.

    When Deepseek introduced the Day after Christmas its Deepseek-V3 model, it corresponded to the possibilities of the best chatbots of American companies such as OpenAi and Google. That would have been impressive alone.

    But the team behind the new system also revealed a bigger step forward. In a research paper in which it was explained how technology was building, Deepseek said that it only used a fraction of the computer chips on which leading AI companies trusted to train their systems.

    The top companies in the world usually train their chatbots with supercomputers that use no less than 16,000 chips or more. The Deepseek engineers said they only needed about 2,000 Nvidia chips.

    Since the end of 2022, when OpenAi left the AI ​​boema, the prevailing idea was that the most powerful AI systems could not be built without investing billions of dollars in specialized AI chips. That would mean that only the largest technology companies – such as Microsoft, Google and Meta, all of which are located in the United States – can afford to build the leading technologies.

    (The New York Times has sued OpenAi and his partner, Microsoft, and claims infringement of the copyright of news content with regard to AI systems. The two technology companies have denied the claims of the suit.)

    But the Deepseek engineers said they only needed around $ 6 million to train their new system. That was about 10 times less than what Meta spent on building his latest AI technology.

    Top AI -Seniurs in the United States say that the research paper of Deepseek has set out smart and impressive ways to build AI technology with fewer chips.

    In short, the engineers of the startup demonstrated a more efficient way to analyze data using the chips. Leading AI systems learn their skills by locating patterns in enormous amounts of data, including text, images and sounds. Deepseek described a way to distribute this data analysis over different specialized AI models – what researchers call a “mix of expert” method – while time is lost by moving data from place to place.

    Others used comparable methods, but moving information between the models tend to reduce efficiency. Deepseek did this in a way that made it possible to use less computing power.

    “It has become very clear that other companies, not only someone like OpenAI, can build these types of systems,” said Tim Dettmers, a researcher at the Allen Institute for Artificial Intelligence in Seattle and a professor in computer science at Carnegie Mellon University who specialized In building efficient AI systems. “Deepseek used methods that everyone can duplicate.”

    The Deepseek research paper argued whether large American companies can retain an important lead in AI, many experts believe that AI technology will become a merchandise, where many companies sell a lot of the same product.

    DeepSeek-V3 can answer questions, solve logical problems and write its own computer programs as effective as everything that is already on the market, according to Standaard Benchmarkts.

    Just before Deepseek released his technology, OpenAi had unveiled a new system called OpenAI O3, which seemed more powerful than Deepseek-V3. But OpenAi did not release this system to the wider public.

    OpenAI O3 is designed to 'reason' due to problems with mathematics, science and computer programming. Many experts pointed out that Deepseek had not built a reasoning model in these lines, which is seen as the future of AI

    When on January 20, Deepseek released his own reasoning model called Deepseek R1, and it also impressed the experts. That ultimately brought us investors and others in panic at the end of last week and during the weekend when they realized the importance of Deepseek's new technology.

    Yes, it still matters.

    Large numbers of AI chips can still help companies in many ways. With more chips they can perform more experiments while exploring new ways to build AI in other words, more chips can still give companies a technical and competitive advantage.

    More chips will also be needed to serve the new breed of “reasoning” AI models, experts said. These require more computing power when people and companies use them.

    Yes. In order to maintain the American lead in the global AI race, the Biden administration had introduced rules that limit the number of powerful chips that could be sold to China and other rivals.

    But the impressive performance of the Deepseek model raised questions about the unintended consequences of the trade restrictions of the US government. The operating elements have forced researchers in China to become creative with a wide range of tools that are freely available on the internet.

    Some experts continue to argue for American trade restrictions and say that they have only recently been set up and that they will have a greater effect on China's capacities to create AI as the years pass.

    No. The world has not yet seen the O3 model of OpenAi and the performance on standard bench market tests were more impressive than something else on the market. But experts are worried that China jumps ahead to open-source AI systems.

    Like many other companies, Deepseek has “open” its latest AI system, which means that it has shared the underlying computer code with other companies and researchers. This allows others to build and distribute their own products with the same technologies.

    This is part of the reason that Deepseek and others in China have been able to build competitive AI systems so quickly and cheaply.

    In the AI ​​world, Open Source collected steam for the first time in 2023 when Meta freely shared an AI system called Llama. At the time, many assumed that the open-source ecosystem would only flourish if companies such as meta-giant companies with huge data centers full of specialized chips-hun technologies will continue to open.

    But Deepseek and others have shown that this ecosystem can thrive in ways that go beyond the American technical giants.

    Many experts have argued that the major American companies should not open their technologies because they can be used to distribute disinformation or to cause other serious damage. Some American legislators have investigated the possibility of preventing or nipping practice.

    But other experts have argued that if supervisors suffocate the progress of open-source technology in the United States, China will get a considerable advantage. When the best open-source technologies come from China, these experts say that American researchers and companies on top will build their systems that will build technologies.

    In the long term, China could focus on AI research and development, which could further accelerate its efforts to build a wide range of AI technologies, including autonomous weapons and other military systems.