The Chinese-built AI app DeepSeek seems to have taken the tech world by storm. It has demolished the market value of a number of tech giants in a single trading day. The speed at which the new Chinese AI app DeepSeek has shaken the technology industry, the markets, and the bullish sense of American superiority in the field of artificial intelligence has been nothing short of stunning.

This Chinese startup recently gained attention with the release of its R1 model, which delivers performance similar to ChatGPT but with the key advantage of being completely free to use. DeepSeek is not only challenging established giants like OpenAI and Meta but is also disrupting global markets, particularly in the AI chipmaking sector.

So What is DeepSeek AI?

DeepSeek AI (or 深度求索) is a Chinese startup lab similar to OpenAI. It was founded in May 2023 by Liang Wenfeng, a millennial hedge fund entrepreneur from South China’s Guangdong province. Wenfeng also co-founded High-Flyer, a China-based quantitative hedge fund that owns DeepSeek. Currently, DeepSeek operates as an independent AI research lab under the umbrella of High-Flyer. The company specializes in the development of advanced AI models and has quickly gained recognition for creating cost-effective, high-performance AI solutions that rival those of industry leaders like OpenAI and Meta.

DeepSeek’s flagship model, known as R1, an open-source AI system trained using an innovative self-reinforced learning method, which eliminates the need for expensive human-labeled datasets allowing it to learn through trial and error. This innovative approach makes the AI more efficient, scalable, and capable of improving itself over time without heavy human intervention.

One of DeepSeek’s standout features is its transparency in reasoning. The R1 model uses a Chain of Thought (CoT) prompting approach, breaking down complex problems into step-by-step explanations. This makes it easier for users to understand how the AI arrives at its answers, fostering trust and usability. Additionally, DeepSeek R1 can work alongside internet search in real-time, a feature that sets it apart from many competitors. It’s also designed to correct its own mistakes, making it more reliable and adaptable.

What’s the Big Deal About DeepSeek AI?

DeepSeek AI is gaining popularity for its powerful AI assistant, which operates similarly to OpenAI’s ChatGPT. According to the app’s description, it is designed to answer user questions and enhance productivity efficiently.

As news of DeepSeek AI’s performance spread, hundreds of billions of dollars were wiped off big tech stocks in the US. This is because DeepSeek was reportedly developed at a fraction of the cost compared to its American rivals, such as OpenAI’s ChatGPT, Meta’s Llama, and Google’s Gemini. The company has quietly impressed the AI world with its low-cost AI and tech innovations, challenging the conventional belief that high budgets and advanced semiconductor chips are necessary for AI breakthroughs.

OpenAI vs. DeepSeek

Undoubtedly, DeepSeek represents the last and the most serious challenge to OpenAI which has established itself as an industry leader with its debut of ChatGPT in 2022. OpenAI has helped push the generation industry forward with its GPT family of models, as well as its o1 class of reasoning models.

While the two companies are both developing generative AI LLMs, they have different approaches. Here’s a comparative table highlighting the key differences and similarities between OpenAI and DeepSeek AI:

Feature	OpenAI	DeepSeek AI
Founding Year	2015	2023
Flagship Model	GPT-4, ChatGPT	DeepSeek R1
Cost to Develop	Billions of dollars	Less than $6 million
Training Method	Supervised fine-tuning, human-labeled data	Self-reinforced learning, no human labels
Pricing for Users	Paid (subscription for advanced features)	Free and open-source
Transparency	Limited reasoning transparency	Chain of Thought (CoT) prompting

A Major Blow to US Tech Giants

The timing of DeepSeek AI’s launch is significant. In recent days, US tech companies have pledged billions of dollars in AI investments, assuming that large-scale funding is required to build the computing infrastructure needed for AI advancements. However, China’s DeepSeek AI has raised serious questions about this assumption.

The company’s emergence has particularly impacted Nvidia, the US-based leader in AI chip manufacturing. Nvidia suffered a historic loss of nearly $600 billion in market value on 27 January, marking the largest single-day drop for any company in US history. Its stock price plunged 17.7% over the course of the day. Once the most valuable company in the world, Nvidia fell to third place behind Apple and Microsoft. However, the company is now slowly recovering.

Additionally, DeepSeek’s R1 model is open-source and free to use, making it accessible to a wide range of users. This has contributed to its rapid rise in popularity, with the model quickly becoming one of the most downloaded apps on the App Store, surpassing ChatGPT within days. The open-source nature of the model also allows users to run it locally, ensuring privacy and security for those who prefer not to rely on cloud-based solutions.

A Game-Changer in AI

DeepSeek’s R1 model has been described as game-changing by users and experts alike. Its ability to reason through problems step-by-step, correct its own mistakes, and provide transparent explanations sets it apart from other AI models. Additionally, R1’s ability to work alongside internet search in real-time gives it an edge over models like GPT-4, which still struggle with this feature.

However, the model is not without its flaws. It performs worse on coding tasks compared to models like GPT-4 and Claude 3.5, and its servers have been overwhelmed by high demand, leading to slower response times. Despite these challenges, DeepSeek R1 remains a powerful tool for users seeking a free and open-source AI solution.

Curiosity

However, like all AI models operating in China, DeepSeek does not answer questions on topics banned by the Chinese government, including politically sensitive events such as the 1989 Tiananmen Square massacre. For instance, when asked simple questions like Who is the president of China? Or who is Xi Jinping? the app will simply reply:

“Sorry, that’s beyond my current scope. Let’s talk about something else”

Conclusion

DeepSeek AI is more than just a startup; it is a disruptor that is redefining the boundaries of what is possible in artificial intelligence. With its cost-effective models, innovative training methods, and open-source approach, DeepSeek is challenging established players and paving the way for a more accessible and transparent AI future. As the company continues to grow and innovate, its impact on the global AI landscape will undoubtedly be profound.

FAQ

1. Who developed DeepSeek AI?

DeepSeek AI was developed by a Hong Kong-based startup founded in 2023. The company is led by its founder, Liang Wenfeng, who is also the co-founder of the quantitative hedge fund HighFlyer.

2. Is DeepSeek owned by the Chinese government?

No, DeepSeek is not owned by the Chinese government. As far as we know, it is an independent startup, though its innovations align with China’s strategic goals for technological self-sufficiency.

3. Is DeepSeek free to use?

Yes, DeepSeek’s R1 model is free to use and open-source, making it accessible to a wide range of users worldwide.

4. Is DeepSeek available in China?

Yes, DeepSeek is available in China and has gained significant attention there, including recognition from top political circles for its contributions to AI development.