The ranks of the heavyweights of the artificial intelligence world were shaken up seemingly overnight with news of an AI program made by a small Chinese start-up that can seemingly rival the best that OpenAI and other major AI leaders can provide.

Enter DeepSeek, the company that is set to change the game for AI and threaten the unrivaled dominance US-based tech companies have had thus far.

The fact that this is disruptive should come as no surprise. Just a week ago, most of the attention in the AI field was focused on the advancements of OpenAI, Meta, and more. Talk of AI super-agents, which are programs that can handle complex tasks and make decisions with a level of intelligence that could theoretically rival some of humanity's best and brightest, and the reveal of OpenAI's new Operator agent all seemed to point to continued American dominance in the sector.
AI super-agents were poised to disrupt the labor market, with Meta CEO Mark Zuckerberg having said that within 2025, it was possible that AI will become advanced enough that it could do the work of a mid-level software engineer.
But the revelation of DeepSeek has seemingly blindsided the entire industry and sent quakes shaking through Wall Street. In just a week after being launched, it became the most downloaded free app in the US, the BBC reported.
But what is DeepSeek? What makes it different from other AI companies? And more importantly, why is this making such major waves in the tech world?
Here is everything you need to know.
DeepSeek is a Chinese start-up established in 2023, owned by the hedge fund High-Flyer. The company released its first models for free in November of the same year, with continued and further refined models put out in 2024.On January 10, 2025, it released its first AI chatbot to the public.  
Its AI program is set up to work on the same level as OpenAI's ChatGPT, the AI platform said to be the gold standard for AI chatbots.
But what makes it so impressive is how it was able to be made at a fraction of the cost that larger US firms pay for the same results, as well as with a fraction of the high-end chips that its competitors use. According to The New York Times, DeepSeek only needed around 2,000 computer chips to train its AI model, specifically chips from Nvidia, whereas other companies use around 16,000 chips. This ended up costing around $6 million. For context, Meta spent around 10 times as much to develop its own AI.
Already back in May 2024, DeepSeek made waves in China for its success at making a successful AI model at such a low price, which forced the country's preeminent leaders in the industry such as ByteDance and Baidu to slash their own AI model prices just to stay competitive.
Now, it is making a similarly major impact on the US market, with Nvidia, one of the most valuable publicly traded companies in the world and a major leader in chip production, having lost nearly $600 billion in market value, the BBC reported.
According to Reuters, Nvidia's stock tumble is the biggest single-session loss for any company.
But what is especially significant is not only that a Chinese company made such major strides in AI development at a fraction of the cost compared to major US firms, but that it did so despite major US sanctions designed to curb China's AI advancements.

How did China's DeepSeek develop despite US sanctions?

The US had previously placed restrictions on the export of advanced AI chips to China in an effort to maintain American AI dominance compared to its chief economic rival.
As a result, it was far more difficult for Chinese firms to get access to the necessary chips.
However, as some experts noted, that seemed to only spur innovation as developers had to find ways of getting around these restrictions.
This was, in part, helped by DeepSeek founder and CEO Liang Wenfeng, who stockpiled a few thousand advanced Nvidia AI chips before US sanctions went into place.
However, just as important is the role of innovation in getting around the limited hardware Chinese companies have at their disposal, working with what they have and making breakthroughs as a result.
How is the US reacting to DeepSeek?
US President Donald Trump has long been hawkish about the US trade rivalry with China and has previously stated the importance of the United States being a dominant force in AI.
DeepSeek seems to upend these claims.

Buddy AI (credit: PR)
Buddy AI (credit: PR)

Speaking to reporters on Tuesday, Trump called DeepSeek a "wake-up call" directed at US tech firms. However, he also said it was a positive development, noting that it's good for US firms, too, if a company is able to do less and spend less on AI development and still get the same result.
OpenAI CEO Sam Altman also noted DeepSeek's "impressive model" and referred to them as a new competitor, though also said that his company would "deliver much better models" in the future.
Are the claims about DeepSeek true?
While DeepSeek has been making ambitious claims, others claim that the AI's successes need to be taken with a grain of salt.
One of the biggest arguments made by skeptics is that DeepSeek's AI breakthrough was only possible due to hoarding Nvidia chips or due to buying other advanced chips illegally,
Speaking to CNN, Yale School of Management senior associate dean for leadership studies Jeffrey Sonnenfeld said that DeepSeek's breakthrough "is either truly disruptive or deceptive, relying on hoarded Nvidia chips and others before the export sanctions."
In addition, others have voiced concern over data privacy and censorship due to policies by the Chinese government.According to multiple reports, the DeepSeek AI chatbot refused to discuss events such as the 1989 Tiananmen Square protests and the Uyghur genocide in Xinjiang and responded to questions about Taiwanese independence by saying there is no such thing as an independent Taiwan and it is a part of China.
This also comes as the US is currently grappling with concerns surrounding Chinese firms having access to US citizens' data. It remains to be seen how the tech sector will be impacted in the long term by these broader concerns.