Will DeepSeek Really Beat OpenAI in the Future?

Uzone.id – The world of technology was shocked by DeepSeek which suddenly boomed in the US App Store app store. The popularity of DeepSeek in the US has managed to beat ChatGPT—a chatbot developed by OpenAI—which still dominates the AI world.

It is noted that at the end of January 2025, this chatbot managed to become the most downloaded free application on the Apple App Store in the US.

DeepSeek was first introduced to the public in 2023. This chatbot grew in popularity when they launched their first open-source language model called R1 in November 2024.

Because of its popularity, the presence of DeepSeek managed to make the US technology stock price collapse. One of them, Nvidia which fell 17 percent in a day of trading, Microsoft fell 2.1 percent, to Dell Technologies 8.7 percent. As a result, Donald Trump is again nervous because the US is indeed notoriously sensitive to technology from China.

The idea of DeepSeek itself was born from Liang Wenfeng who dared to pour money into investing in artificial intelligence, betting on AI chips, and putting his time into building a development team.

Unlike other technology companies in China, DeepSeek prefers to focus on fundamental research in the development of AI models. DeepSeek’s attention is devoted to developing AI models that can be used by other companies rather than building end-user applications.

One of the steps of DeepSeek’s strategy is to implement an open-source system on their AI models so that developers around the world can access and modify the underlying code according to their individual needs.

DeepSeek’s decision not only succeeded in getting them noticed, but also created a new standard in AI development. Currently, DeepSeek is one of China’s hopes in facing technological competition with the United States.

However, can DeepSeek really beat OpenAI in the future? For now, there is no definitive answer. However, the breakthrough brought by DeepSeek managed to steal attention, even reaping praise from the ranks of tech bosses, including Tim Cook and Sam Altman. 

The presence of DeepSeek presents a new challenge to established players like OpenAI. On the other hand, the creators of DeepSeek revealed that they have found a way to train their AI by using custom components without wasting resources. 

The innovations that DeepSeek uses have drastically reduced the computing power required to develop and run models, ultimately reducing the costs associated with chips and servers. This reduction has managed to attract the attention of smaller AI developers who are looking for cheaper alternatives.

Additionally, the open approach offered by DeepSeek allows for greater transparency, customization, and accessibility. This allows developers to modify and adapt the model according to their needs.

Another advantage of DeepSeek lies in the cost of its manufacture. Some sources wrote that the chatbot, which was released two years ago, was created at a cost of only $5.6 million – only 10% of Meta’s Llama cost. This then raises the question, is large funding really needed? This factor is what makes Nvidia shares go into free fall.

Unfortunately, while it sounds promising, DeepSeek still shows a lot of flaws. First of all, this chatbot is very lacking in transparency around the training data used, which has led to lawsuits filed by several leading companies.

In addition, security issues also continue to haunt DeepSeek. Australia’s Science Minister, Ed Husic, became the first figure from Western governments to voice these concerns. In an interview with ABC News, Ed Husic said that there are still many questions that DeepSeek has not answered, including about data management and privacy. 

On the other hand, cybersecurity firm KELA reported that they identified a vulnerability in DeepSeek. KELA reported that they managed to break into the DeepSeek model that allows to produce harmful outputs such as ransomware development, to the falsification of sensitive content. 

Previously, DeepSeek itself reported that it had experienced a large-scale cyberattack on their service. To deal with this, DeepSeek then temporarily restricted their user registration.

In fact, several countries including Taiwan, Italy, and Australia have banned their citizens from using DeepSeek—especially for public sector workers and using official devices owned by companies—to use DeepSeek. So, what do you think?

Writer: Aisyah Banowati Editor: Hani Nur Fajrina