U.S. Companies Should Sleep Soundly at Night Knowing DeepSeek Isn’t the Threat They Fear
On January 20, 2025, China sent shockwaves through the tech industry when it launched its very own artificial intelligence model—DeepSeek. (Ben Cohen, The Wall Street Journal). DeepSeek is an open-source AI model that its backers claim is more cost-effective than its rivals, including the U.S.’s OpenAI. (Forbes). According to reports, DeepSeek’s founder Liang Wenfreng developed the AI model with just $1.4 million in capital. (Ty Roush, Forbes). Meanwhile, DeepSeek’s largest rival, OpenAI, cost more than $100 million to develop, according to its CEO Sam Altman. (Katharina Buchholz, Forbes). The disparity between these figures and the drop in U.S. tech stocks has sparked fears among leading tech companies in the U.S. that Chinese outfits will soon overtake them. (Brian Cheung et al., NBC News). However, this article explores why the development of DeepSeek may very well be beneficial to America’s dominance in the AI realm, should U.S. tech companies be up for the challenge.
Dan Ives, a technology research analyst, created the analogy that DeepSeek is to OpenAI what Temu is to Amazon. (Adam Spatacco, NASDAQ). Like DeepSeek, Temu is a Chinese company, and “[w]hen it first launched in late 2022, some Wall Street skeptics began declaring the low-cost marketplace an existential threat to Amazon” because a decline in Amazon’s e-commerce sales followed. Id. However, this decline was also attributed to inflation peaks and the Federal Reserve’s interest rate hikes. Id. In fact, in 2023 and 2024, Amazon’s online sales continued to increase from the initial decline. Id. Like current views on DeepSeek, what was perceived to be a foreign market threat to a similar U.S. company turned out to be just another competitor entering the space, and not a very threatening one. As such, it is likely that DeepSeek is another emerging competitor, rather than a major market-shifting force.
One company that has been affected heavily by DeepSeek is Nvidia, a tech company that designs and sells graphics processing units (GPUs) which many U.S. companies utilize in developing their AI models. (Forbes). The creators of DeepSeek also utilized Nvidia’s GPUs in developing their model. Id. However, because the U.S. placed a ban on the export of advanced semiconductor technologies, to allegedly stall China’s progress in AI, significant limits were placed on the sale of advanced Nvidia AI chips to other countries. (Carmen Reinicke, Bloomberg). Yet, the developers of DeepSeek worked around this ban and implemented innovative engineering to continue crafting their model. (Forbes).
DeepSeek operates differently from OpenAI in that it is allegedly less dependent on computational power while still “performing on par with OpenAI’s o1 reasoning model.” (Forbes; Belle Lin, The Wall Street Journal). Essentially, this means that DeepSeek requires less data processing to train its models. (Belle Lin, The Wall Street Journal). DeepSeek was able to achieve this through “[e]ngineering tweaks, such as custom communication schemes to improve data transfer efficiency.” (Forbes). Though, it is important to note that these techniques were not invented by the developers of DeepSeek, but “applying them at the scale they did … was novel.” (Belle Lin, The Wall Street Journal).
The innovation that the DeepSeek developers have shown is said to have caused Nvidia’s market cap loss, reportedly one of the largest market drops of any stock in U.S. history. (Carmen Reinicke, Bloomberg). Following the unveiling of DeepSeek, Nvidia has lost nearly $600 billion in market value, a number that may continue to rise if the AI industry decides to follow DeepSeek’s footsteps and prioritize efficiency over sheer processing power. (Brian Cheung et al., NBC News). The launch of DeepSeek raised significant concerns over the spending on AI in the U.S. considering China’s ability to produce a cheaper and (allegedly) more effective AI model for a fraction of the cost and with less advanced GPUs. Id.
There is one theory of economics which could prove DeepSeek’s model and priorities beneficial to Nvidia, though. Jevons paradox theorizes that efficiency improvements will lead to increases in consumption, not decreases. (Ben Cohen, The Wall Street Journal). If this theory proves true, cheaper AI models will lead to more people using AI and thus, more companies buying Nvidia’s GPUs to continue innovating. Id. This theory has merit as it is unlikely that U.S. tech companies are going to throw in the towel and give up development of their own models. Instead, it is entirely likely that the spirit of competition will inspire and drive U.S. tech companies to reach new heights with their AI innovations and maintain dominance in the AI global market.
Nvidia does not appear too concerned with the market drop as in a recent statement it referred to DeepSeek’s model as an “excellent AI advancement” and noted that the “work of running AI models, ‘requires significant numbers of Nvidia GPUs and high-performance networking.’” (Carmen Reinicke, Bloomberg). Despite its market cap loss, Nvidia seems to remain confident that its GPUs will continue to be utilized in the ongoing race for AI dominance.
If there is one serious takeaway from DeepSeek’s development, though, it should be the Chinese government’s role in the project. The Chinese government implemented several mandates and state funding for technology ambitions to drive DeepSeek’s development. (Amrith Ramkumar, The Wall Street Journal). While President Trump has signed an executive order directing his team to create an AI action plan and has said that he would like to support companies by relaxing regulations and boosting energy production, this remains to be seen. Id.
The launch of DeepSeek has unveiled China’s growing ambition in the AI realm and has introduced a serious competitor into the AI market. However, DeepSeek is not an immediate threat to U.S. tech companies. Like Temu’s entry into the market, DeepSeek represents foreign competition that U.S. companies should take seriously but not shy away from or fear. If DeepSeek is as cost-effective and efficient as its developers claim, then U.S. AI developers should view the model as inspiration that can be improved upon for their own models. Several U.S. CEOs have done just that and have pledged to continue spending to develop their AI infrastructures. (Ben Cohen, The Wall Street Journal).
Innovation is driven by competition, whether domestic or foreign. Should they be up for the challenge, U.S. companies should sleep soundly at night knowing DeepSeek is not the threat they fear it to be, but rather a motivator for their own growth.