The meteoric rise of Chinese artificial intelligence (AI) app DeepSeek has rattled global tech markets, triggering a selloff in shares of major companies like Nvidia, Microsoft, and Meta.
The app recently surpassed rivals such as ChatGPT to become the top-rated free app on Apple’s App Store, shaking confidence in U.S. dominance in AI innovation.
DeepSeek’s rapid ascent has startled investors, with European share prices dropping and tech-related stocks, such as ASML and Siemens Energy, suffering sharp declines of over 10% and 21%, respectively.
Analysts believe the app’s low-cost development model has challenged existing assumptions about the financial barriers to creating cutting-edge AI systems.
“This idea of a low-cost Chinese version hasn’t been at the forefront, so it’s taken the market by surprise,” said Fiona Cincotta, senior market analyst at City Index, who spoke with BBC.
A low-cost AI revolution
DeepSeek’s disruptive potential has also raised concerns about the future profitability of U.S. tech giants that have poured billions into AI infrastructure.
DeepSeek is powered by the open-source DeepSeek-V3 model, which its researchers claim was developed for under $6 million, a fraction of the costs incurred by its competitors.
- While this claim has been met with skepticism in the AI community, the company’s approach has demonstrated that high-performance AI models can be developed with less computing power and lower costs.
- DeepSeek-R1, a recently launched version of the app, has been touted as having performance on par with OpenAI’s advanced models in tasks like math, coding, and natural language processing.
- Silicon Valley venture capitalist Marc Andreessen described DeepSeek’s rise as “AI’s Sputnik moment,” drawing parallels to the Soviet Union’s surprise launch of the first satellite in 1957.
Navigating U.S. export restrictions
DeepSeek’s emergence comes amid heightened U.S. export controls on advanced chips and technology to China.
Despite these restrictions, DeepSeek has reportedly managed to thrive by utilizing Nvidia A100 chips, now banned for export to China, in combination with cheaper, lower-end chips still available for import.
- Founder Liang Wenfeng, a 40-year-old graduate in information and electronic engineering, reportedly stockpiled approximately 50,000 Nvidia A100 chips to power the company’s AI ambitions.
- In response to DeepSeek’s success, U.S. tech firms and foreign investors announced the Stargate Project, a $500 billion investment into AI infrastructure based in Texas.
- While DeepSeek’s rise has sparked concerns about its impact on U.S. AI dominance, Wall Street analysts caution against overstating its implications.
- According to Citi, U.S. firms still benefit from access to more advanced chips and resources, giving them a competitive edge in an increasingly restrictive environment.
The founder behind DeepSeek
DeepSeek was founded in 2023 by Liang Wenfeng in Hangzhou, southeastern China.
Beyond his work in AI, Liang also runs a hedge fund that initially backed DeepSeek’s development.
Known for his pragmatic approach, Liang has expressed surprise at the global reaction to his AI model.
“We didn’t expect pricing to be such a sensitive issue,” Liang said in a 2024 interview. “We simply calculated costs and set prices accordingly.”
What you should know
The rise of DeepSeek comes at a time when Microsoft is pushing the American AI leadership agenda to reduce China’s dominance in the global AI space.
In a January 3, blog post where he announced Microsoft’s plans to invest $80 billion in AI-enabled data centers this year, the company’s Vice Chairman and President, Brad Smith, said the growing competition between U.S. and Chinese AI technologies, particularly in developing nations.
- According to him, the advent of generative AI has intensified competition, particularly with China’s rapidly advancing AI sector.
- He compared this race to the evolution of the telecommunications industry over the last two decades.
- Smith noted that Chinese companies, with substantial government subsidies, overtook Western counterparts in telecommunications, creating dependencies that posed challenges to U.S. national security.
He said China is now replicating this strategy in AI by subsidizing access to critical technologies like chips and promising to build local AI data centers in developing nations. The strategy aims to lock these nations into China’s AI ecosystem for the long term.