AI War: Why China's DeepSeek is a Threat to OpenAI, Google, Meta

China’s AI start-up DeepSeek is creating buzz in the tech world. It has overtaken ChatGPT as the most downloaded app on Apple’s US store.

DeepSeek recently gained prominence when it unveiled its large language models DeepSeek-V3 and DeepSeek-R1. This comes at a time when Donald Trump just took charge as the president. Trump’s love for protectionism is not unknown. For instance, recently, the US targeted China with new AI curbs. 

So, now, China's DeepSeek is posing a challenge to the prominence that the USA holds over AI. Many netizens have highlighted that the DeepSeek start-up could pop the AI stock frenzy in the USA. Currently, OpenAI, Google, and Meta are all originally from the USA. Before delving any further, let's first understand what DeepSeek is.

What is DeepSeek?

Chinese AI start-up DeepSeek was established in May 2023 by Chinese billionaire and hedge fund master Liang Wenfeng. The start-up operates independently under a Chinese quant fund, High-Flyer. Founded in 2015 in China, High-Flyer started as a hedge fund as well as an AI company that was focused on using machine learning for stock trading. However, in 2023, it spun off a research arm to develop efficient AI models using open-source technology.

Read: Artificial Intelligence Takes the Spotlight At TiEcon Delhi 2024

The start-up first started in November 2023 with the release of DeepSeek Coder, which is an open-source model that consists of a series of code language models. One of its most successful launches was DeepSeek V3 in December 2024. DeepSeek-V3 is an advanced and open-sourced AI language model. This model excels in generating text, answering questions, and creating content while working efficiently. It rivals top AI systems like OpenAI and Meta, earning global recognition for its advanced capabilities.

DeepSeek-V3 is built on the Mixture-of-Experts (MoE) system, which activates only the most relevant parts of its 671-billion-parameter model for each task, which in turn makes it efficient and powerful.

The biggest stir in the tech world occurred when DeepSeek R1 was launched on January 20, 2025. Now, considered a disrupter in the AI world, DeepSeek R1 is an advanced model focused on reasoning tasks. It uses large-scale reinforcement learning for training and directly competes with OpenAI's O1 model.

What is interesting about DeepSeek is the low cost associated with building the model as compared to the cost needed to build Meta and other models. The start-up claims that it created its low-cost model in just two months with less than $6 million. This is in huge contrast to the $100 million that OpenAI spent to train its GPT-4 model.

Read: AI Models Face-off: Chinese AI Startup DeepSeek Challenges OpenAI’s o1 & GPT 4o

To add to it, DeepSeek used older, cheaper Nvidia H800 GPUs. On the other hand, AI companies in the USA generally use powerful, expensive Nvidia H100 GPUs. One reason behind the same was the fact that due to U.S. export restrictions, Chinese companies like DeepSeek couldn't access the best AI chips, like Nvidia's. As a result, DeepSeek's engineers had to create smarter, more energy-efficient algorithms to make up for the lower computing power they had.

How did the stock market react to it?

The world has now taken notice of the new development in China. Speaking at the World Economic Forum, Microsoft CEO Satya Nadella said, “We should take the developments out of China very, very seriously.”

Meanwhile, the stock markets have witnessed an interesting rally with the Nasdaq 100 futures dropping by 1.9 percent during Asian trading hours. NVIDIA shares dropped more than 5 percent in 24-hour markets, according to Robinhood data on January 26. To add to it, Japanese chip stocks also saw a downfall from the surge of DeepSeek. For example, Advantest, which is a chip-making tool maker and has Nvidia as its client, saw a fall. Meanwhile, Softbank Group, which recently highlighted its plan to invest around $100 million in the USA, also saw around a 6 percent drop. 

Speaking about the stock market performance, Chamath Palihapitiya, venture capitalist, wrote on X, “There will be volatility in the stock market as capital markets absorb all of this information and re-price the values of the Mag7. Nvidia is the most at risk for obvious reasons. That said, markets will love it if Meta, Microsoft, Google, etc., can win WITHOUT having to spend $50-80B PER YEAR.”

Meanwhile, speaking about the recent breakthrough, Jim Fan, Senior Research Manager at Nvidia, wrote on X, “We are living in a timeline where a non-US company is keeping the original mission of OpenAI alive—truly open, frontier research that empowers all.”

The rush for AI has been increasing, and geopolitics is no exception to it. For instance, in October last year, the Biden administration finalized rules to limit the USA’s investments in China's tech sectors, targeting AI, semiconductors, and quantum technologies for national security reasons. To add to it, in September last year, China also urged local companies to stay away from Nvidia chips. The intent behind the same was to promote local Chinese chipmakers as per a Bloomberg report.

A blog post by Goldman Sachs says, “And while the United States, China, and Russia do not agree on many things, they all acknowledge that AI could reshape the balance of power.” It will be interesting to see how China increases its prominence in the AI world and the effect that it will cause.