Chinese AI Company DeepSeek: A New Power in Global Market and Technology
China-based AI company DeepSeek has begun to attract attention worldwide, and has attracted serious interest, especially from the American public and Wall Street. The company has established itself in the market with high-performance and at the same time low-cost AI technology, competing with giants such as OpenAI, Google and Meta. DeepSeek has brought a new approach to the field of AI by making its innovative technologies open source.
The company's new technology is considered by some experts to be able to compete with OpenAI's ChatGPT. What is particularly interesting is that DeepSeek has spent only $6 million to develop AI applications. For comparison, OpenAI, Google and other US companies plan to invest about $1 trillion in AI in the coming years.
DeepSeek's Early Achievements
DeepSeek's first model, DeepSeek Coder, was introduced in November 2023, followed by DeepSeek-V2 in May 2024. DeepSeek-V2 attracted widespread interest in the market with its high performance and cost optimization. The latest model, DeepSeek-R1, was introduced in January 2025 and has 236 billion parameters. This model competes with OpenAI's leading models with both low costs and high-quality results.
DeepSeek's technologies are based on technologies such as reinforcement learning, mixture-of-experts (MoE) architecture, and multi-head latent attention. In addition, the company uses distillation techniques to transform its large models into smaller and more efficient models.
Impact on the US Technology Sector
DeepSeek's applications had a significant impact on the stocks of US companies in the field of artificial intelligence after their launch. For example, Nvidia shares fell 17%, losing $600 billion. ASML shares fell 6%, and Broadcom fell 17%. The Nasdaq index also fell 3% (612 points) on the same day. This reinforced the idea among analysts that "Silicon Valley is overinvesting in artificial intelligence technologies."
The Power Behind DeepSeek
DeepSeek was founded in July 2023 by Liang Wenfeng, a graduate of Zhejiang University, one of the Chinese universities. Liang financed the company with his own hedge fund and has a fortune of about $8 billion. The company has no external investors, which allows DeepSeek to work long-term and away from commercial pressures.
Liang initially purchased Nvidia A100 chips and used them to develop artificial intelligence models. It is these chips that form the solid technological foundation of DeepSeek. The company has also optimized the DeepSeek-V3 model in collaboration with AMD.
DeepSeek’s Differentiating Aspects
DeepSeek’s open-source large language model is based on the technique of “on-demand computation.” This technology ensures that only the most relevant models are activated, saving both energy and costs. In addition, DeepSeek’s application is available on Apple’s App Store and can be downloaded for free. The company was one of the most downloaded applications on the Apple store in the first days of its launch.
Discussions of the new technology in the US
The presentation of DeepSeek’s applications came a few days after former US President Donald Trump announced the $ 500 billion Stargate project in collaboration with ChatGPT manufacturer OpenAI. This project aimed to secure the technological future of the United States. But DeepSeek’s cheaper and more efficient technology changed the direction of these discussions.
DeepSeek is intensifying competition in the artificial intelligence market by offering an alternative to technologies from Nvidia and other US technology companies. The company's new approaches and low-cost solutions could lead to major changes in the technology field in the future.
