In a shocking turn of events, DeepSeek, a relatively unknown Chinese AI company, has thrown the U.S.-dominated AI landscape into turmoil. On January 27, 2025, DeepSeek’s groundbreaking AI model, R1, not only achieved the top spot in the global AI/GPT app rankings but also sent ripples through the tech market, erasing an unprecedented trillion dollars in market value for giants like Nvidia, OpenAI, Meta, Google’s Alphabet, and others.
DeepSeek’s emergence comes on the back of the release of its R1 model for smartphones, just a couple of weeks after its desktop debut on January 10. Although DeepSeek’s chatbot V3 made waves in December for performing on par with the leading U.S. AI systems, it wasn’t until the end of January that the company truly disrupted the market. On the eve of the Lunar New Year in China, DeepSeek’s success caused massive stock plunges, marking an overnight market disaster for major tech players.
According to Casey Newton from the Hard Fork podcast, the scale of DeepSeek’s achievement is hard to ignore: its model, boasting over 680 billion parameters, is larger than Meta’s Llama and came at a fraction of the cost—just $5.6 million to train, compared to hundreds of millions spent by its American counterparts.
What sets DeepSeek apart from its competitors is its unorthodox approach to developing AI. While major players rely on costly and advanced Nvidia chips, DeepSeek built its model using older, less expensive hardware, significantly lowering its operational costs. Moreover, the model’s training process defied the industry norm. Instead of relying on the expensive supervised fine-tuning methods typically used, DeepSeek employed a unique approach called “self-evolution through a pure reinforcement learning process.” This, according to the company, allows for better reasoning capabilities without the steep costs usually involved in fine-tuning AI systems.
The company’s founder, Liang Wenfung, a hedge fund manager, has turned what began as a side project into a powerful AI challenger. What was once dismissed as an overly ambitious dream is now a reality that has shaken the very foundations of the AI market. “We thought this was only possible from giants like ByteDance and Alibaba,” one of Liang’s business partners shared. “Now, less than two years later, we are seeing Nvidia lose hundreds of billions of dollars.”
DeepSeek’s innovative approach extends beyond its technology to its business model. R1 is open-source, which allows researchers worldwide to access and build upon it—a sharp contrast to the closed systems offered by companies like OpenAI and Google. DeepSeek’s model is available at a fraction of the cost to use, making it a highly attractive alternative for both researchers and tech developers. Despite this, the company has faced challenges, including cyberattacks on its services after the market crash, and controversies surrounding its ties to China, which have led to criticism and censorship concerns.
Nevertheless, DeepSeek’s rapid rise represents a major shift in the global AI landscape. According to Professor Geoff Webb from Monash University, “If DeepSeek’s claims are all true, it means that the US tech sector no longer has exclusive control of the AI technologies, opening them to wider competition and reducing the prices they can charge.” This could ultimately lead to a more diverse and accessible AI landscape, where innovations from other parts of the world, including China, have a more prominent role.
The implications of DeepSeek’s breakthrough are far-reaching. The monopolization of AI by a small group of U.S. tech giants has long been seen as a barrier to global collaboration and innovation. With this new development, the path is now open for greater competition, potentially leading to more affordable and diverse AI technologies that benefit researchers and industries around the world.
DeepSeek may have just marked the beginning of a new chapter in AI, one where the U.S. no longer holds the monopoly on cutting-edge technology.
