China's DeepSeek AI Takes the Lead Over ChatGPT on App Store

On Monday, the Chinese artificial intelligence startup DeepSeek achieved a notable milestone by becoming the most-downloaded free app in the U.S. on Apple's App Store, surpassing OpenAI's ChatGPT. This shift in popularity has sent ripples through the tech stock market, leading to significant declines and potential losses in market capitalization totaling billions.

Industry experts, analysts, and tech investors suggest that the growing excitement—and underlying concern about keeping pace in the fast-evolving AI landscape—might indeed be justified. The competition among tech giants and emerging startups is intensifying, as they race to secure a foothold in a market projected to reach over $1 trillion in revenue within the next decade.

What is DeepSeek?

DeepSeek was established in 2023 by Liang Wenfeng, who co-founded High-Flyer, a hedge fund focused on AI technologies. The startup was developed from High-Flyer’s AI research unit with the aim of creating large language models and eventually achieving artificial general intelligence (AGI), a target that many in the industry, including OpenAI, are striving for. Analysts from Jefferies indicate that DeepSeek remains fully owned and funded by High-Flyer at this time.

The momentum for DeepSeek began earlier this month following the launch of R1, its reasoning model designed to compete with OpenAI's offerings. R1 is an open-source model, allowing any developer in the AI community to access and utilize it. It has quickly ascended to the top of various app stores and industry performance rankings, with users praising its reasoning and performance skills.

However, similar to other chatbots developed in China, DeepSeek does have its limitations. For example, when users inquire about policies related to Chinese leader Xi Jinping, the AI tends to redirect the conversation away from such topics.

Another point of discussion is that DeepSeek's R1 was developed despite the U.S. implementing three rounds of restrictions on chip exports to China over the past three years. Estimates regarding the investment in DeepSeek's R1 vary, though analysts from Jefferies quoted a possible development cost of approximately $5.6 million. This figure is reported to be less than 10% of the cost associated with Meta's Llama model. Regardless of exact costs, there seems to be consensus that R1 was created at a significantly lower expense than competing models from OpenAI, Anthropic, and Google.

This shift raises critical questions within the AI sector, such as the sustainability of escalating funding rounds and high valuations, as well as the possibility of an investment bubble.

Industry Reactions

The news has prompted reactions from major American technology companies, who are eager to respond before potential customers shift to more affordable alternatives offered by DeepSeek. Reports indicate that Meta has initiated four task forces within its generative AI department to address the newfound competition.

Microsoft CEO Satya Nadella commented on the developments on X, referring to DeepSeek’s rapid growth as an example of the Jevons Paradox, where increased efficiency and accessibility in AI lead to heightened demand. OpenAI CEO Sam Altman expressed views on the unfolding AI revolution, emphasizing its unstoppable nature.

Yann LeCun, Meta's chief AI scientist, noted on LinkedIn that DeepSeek's success highlights a shift in the industry towards open-source technology, asserting that the advancements drawn from open research speak to the power of collaboration in AI development.

In a recent statement, Alexandr Wang, CEO of Scale AI, pointed out that DeepSeek's latest model is groundbreaking and that its R1 release performs comparably with leading American models. Wang describes the ongoing competition between the U.S. and China in the AI field as akin to an