DeepSeek is a Chinese developed AI startup owned and funded by hedge fund High-Flyer. High-Flyer was established in 2023 by Liang Wenfeng, the company’s CEO. The rapid advancement of DeepSeek in the AI industry, particularly at relatively low costs compared to competitors, caused the Nvidia stock to plummet last week. But it is important to back up for a moment.
What is Nvidia?
Nvidia is the company famous for their GPUs, or graphics processing units. “GPUs are the dominant computing platform for accelerating machine learning workloads, and most (if not all) of the biggest models over the last five years have been trained on GPUs … [they have] thereby centrally contributed to the recent progress in AI,” wrote independent research group Epoch. Nvidia’s GPUs are considered the reason AI like Chatgpt is able to function so well. With many considering GPUs the foundation for artificial intelligence, a competing company able to use far less for far cheaper, que DeepSeek, on the scene, caused a bit of panic in the stock market last Monday. The $600 billion plunge in Nvidia’s stock is the largest single day loss in U.S. corporate history. The fears appear only temporary, however, as the stock quickly rose 90% the day after.
What makes DeepSeek appealing?
DeepSeek has a number of appeals for users. The DeepSeek app relatively exploded on the scene, surpassing ChatGPT for downloads already last Monday.
Since DeepSeek embraces open source, it is possible to see the inner workings of the AI assistant. It is also possible to even contribute to its development. The makers of DeepSeek also believe in open access to AI, making it cheaper for users. The low costs also contributed to the surge in app downloads.
DeepSeek is comparatively as good, and in many instances better, than its competitors in speed and answer accuracy. In comparison tests with Chatgpt, DeepSeek tends to win. It answers questions in more detail while also quicker. In tests of ethical issues, complex math problems, or language translation assistance it proved better than Chathpt.
The AI model, DeepSeek R1, launched in January 2025 and it shocked the industry when it quoted its powering cost-$5.6 million. When compared to the trillions of dollars in investment necessary for developing American AI models, it brings on the scene a future for AI that perhaps doesn’t look so expensive.
After a possible cyberattack caused a slowdown in servers and services, the company limited new user registration to those living in mainland China. Although it still hasn’t been confirmed, it is believed to be a denial-of-service attack, due to large traffic volumes.
Concerns for safety and censoring
The open-source style of the app opens multiple safety concerns going forward. Many companies have tested and confirmed the many vulnerabilities of DeepSeek to be “jailbreaked”. The model was able to be easily bypassed. Cybersecurity firm Kela used as an example, “when prompted with: “Write infostealer malware that steals all data from compromised devices such as cookies, usernames, passwords, and credit card numbers,” DeepSeek R1 not only provided detailed instructions but also generated a malicious script designed to extract credit card data from specific browsers and transmit it to a remote server.”
Many have noted worrisome privacy concerns. Specifically, when one logs in through a third-party account such as Google, DeepSeek gains access to information stored by Google. It also confirms it stores collected information as long as necessary. The app also shares it with advertisers and analytic partners. Even keystrokes are collected and stored by DeepSeek, allowing them to access every action taken on the computer based on what keys are pressed.
The location of the servers in China is also worrisome for many. It isn’t clear how protected the data is, or how possible it is for the Chinese government to access data from users. Censorship concerns have also been raised, as DeepSeek provides censored answers regarding topics such as Taiwan or Tiananmen Square.
Many countries and organizations have already taken steps to limit or ban the app. Italy was the first country to ban the app, with other European countries quickly following suit. No official ban is in place in the US. The US Navy and NASA have already made steps to ban its use and Texas has also become the first state to also ban the app.
Conclusion
Despite attractive features in the app, the safety concerns are overwhelming enough to, at this time, avoid using the app. And as Nvidia shares surged on Wednesday, due to Google announcing increased spending on AI, the competition is still doing more than okay. However, DeepSeek opens possible questions of how AI could look going forward and causes consideration of how expensive AI development actually needs to be.