Deepseek’s status as a game changer was hinted at when it caused major stock market changes even before its debut. When it became known that Deepseek was entering the battle, Nvidia, the largest chipmaker in America, lost market capitalization of almost $600 billion (£484 billion) as a result of the tech world learning AI development was not as necessarily open and democratized to the public as presumed, causing the Nasdaq—a largely tech-centric index to also fall by more than 3% in response. Two other semiconductor giants who helped Nvidia’s ascension also fell. Broadcom shares dropped nearly 17%, while Taiwan Semiconductor Manufacturing Company dropped nearly 13%. Such drops suggest that Deepseek was more than an announcement for one company, but an announcement for the semiconductor sector overall. Yet despite significant changes in the market, a bright tomorrow for the US with China as an AI competitor is still potentially possible. It is predicted the losses in the market may rebound as it becomes accustomed to the presence of Deepseek, and whether the US is prepared for technological advances will be seen sooner than later. What does this all mean for markets, the government, and more? 

Recent articles generating buzz around Deepseek suggest that training its revolutionary AIs only costs $5.6 million—£4.5 million—paling in comparison to the thousands—if not millions—spent training other Silicon Valley AIs like OpenAI and Google. Should this be reality, investors believe the low training cost would be a detriment to the AI giant’s quality and long-term development; however, it would make Deepseek competitive against them without having to expend so much energy to sustain its existence. Online news articles about Deepseek also record the company behind it unveiling its latest product, the V3 AI model, in December 2024. The V3 is a language model created through training on a paltry 2,048 Nvidia chips—minuscule compared to other vast training endeavors—but an indication of how well this company can train its products. More recently, however, as of last week, Deepseek launched the R1 model. The R1 is a reasoning-advanced model comparable to OpenAI’s o1 model which was released nearly two months ago. One thing is clear–Deepseek is in the game, and has the potential to reshape it.