The fluttery new AI technology called DeepSeek is dominating the market. The Chinese startup’s strong large-language model, which is based on open-source technology, has impressed the tech industry.

DeepSeek has also sent shockwaves through the Artificial industry, showing that it’s possible to build a strong AI for million in technology and education, when American firms like OpenAI, Google, and Microsoft have invested billion.

What is DeepSeek?

Liang Wenfeng, a Chinese federal who studied digital information and communication executive at Zhejiang University, is the founder and founder of DeepSeek. Liang began his career in AI by using it for statistical buying, co-founding the Hangzhou, China-based wall account High-Flyer Quantitative Investment Management in 2015. In 2023, Liang launched DeepSeek, focusing on advancing artificial general intelligence.

Image: DeepSeek

DeepSeek launched its first big speech design, DeepSeek-Coder, on November 29, 2023.

But it wasn’t until January 20, 2025, with the launch of DeepSeek-R1, that the business upended the Artificial business.

With a team of just 200 people and a budget of$ 6 million, DeepSeek released its free, open-source model, which was on par with OpenAI’s much-ballyhooed GPT 01 model—a project that cost as much as$ 600 million and took an an estimated 3, 500 people two years to build.

In contrast to large tech companies with large payrolls in the west, DeepSeek has optimized its choosing to focus on just graduated students:” Those with more than eight years of work experience are essentially rejected,” a recruiter told 36kr, a well-known Chinese technology blog.

And, whereas OpenAI and other powerful Artificial models were generally available as subscription products, DeepSeek’s code is available source, available for public scrutiny and can be downloaded to a local computer via AI playground Huggingface, or as a phone app, for free.

DeepSeek’s underlying technologies was viewed as a significant advance in AI, and its launch shocked the US tech industry, wiping out$ 1 trillion in value in a single day.

Image: DeepSeek

What’s so special about DeepSeek?

DeepSeek’s victory comes from its strategy to design style and education. Like a huge horizontal supercomputer that divides tasks among several processors to work on them together, DeepSeek’s Mixture-of-Experts system deliberately activates just about 37 billion of its 671 billion parameters for each task. This approach significantly improves efficiency, reducing computational costs while still delivering top-tier performance across applications.

DeepSeek improves its training process by utilizing Group Relative Policy Optimization, a reinforcement learning method that improves decision-making by contrasting a model’s choices with those of similar learning agents. This allows the AI to refine its reasoning more effectively, producing higher-quality training data.

By releasing its models under the MIT license, which makes it unique from competitors that maintain closed and proprietary systems, DeepSeek also demonstrated a commitment to open-source accessibility by demonstrating its commitment to open-source accessibility. Open-source also makes it possible for developers to enhance and share their work with other developers, who can then build on that work in a never-ending cycle of evolution and improvement.

DeepSeek’s development is helped by a stockpile of Nvidia A100 chips combined with less expensive hardware. Some estimates put the number of Nvidia chips DeepSeek has access to at around 50, 000 GPUs, compared to the 500, 000 OpenAI used to train ChatGPT.

Reactions to DeepSeek

Many AI technologists have lauded DeepSeek’s powerful, efficient, and low-cost model, while critics have raised concerns about data privacy security.

We are in a time where a non-US company is preserving the original goal of OpenAI: to conduct truly open, frontier research that empowers all. It makes no sense”, Nvidia Senior Research Manager Dr. Jim Fan wrote on X ( formerly Twitter ). The most likely outcome is” the most entertaining outcome.”

Even Sam Altman, the CEO of OpenAI, acknowledged that DeepSeek is impressive.

” We will undoubtedly deliver much better models, and having a new competitor is legit invigorating.” On X, Altman said.

Days later, though, the firm claimed to have found evidence that DeepSeek used OpenAI’s proprietary models to train its own rival model.

Critics have also raised questions about DeepSeek’s terms of service, cybersecurity practices, and potential ties to the Chinese government. Others have highlighted the extensive amount of user data collected by DeepSeek, including device models, operating systems, keystroke patterns, and IP addresses—data that’s stored on DeepSeek’s China-based servers, according to the firm’s privacy policy.

” Privacy is an issue because it’s China. It’s always about obtaining user data. So user beware”, Kevin Surace, CEO at AI software developer Appvance, told . It will make “everyone reevaluate how models are trained and how much power is needed for inference.”

What does the future hold for DeepSeek?

DeepSeek’s explosive rise challenges Western tech giants ‘ dominance and raises important questions about the future of AI: who controls it, who controls it, and how accessible and affordable it all should be.

However, questions remain regarding the long-term effects of DeepSeek and whether U.S. President Trump will implement a ban in the manner of TikTok to counteract China’s ostensible overnight dominance in the AI sector. Did High-Flyer misrepresent how to use GPUs to make DeepSeek appear more effective than it actually is? Was DeepSeek’s sudden public launch timed to drive down Nvidia’s stock for the benefit of well-positioned investors?

As competitors, including Meta and Perplexity AI, scramble to adapt to DeepSeek’s methodology, the full impact of this AI breakthrough remains uncertain. However, one thing is certain: DeepSeek resurrected the tech sector by demonstrating once more that sometimes, resource constraints force novel ideas to be created and that potent technology can be created without multi-billion dollar price tags.

Generally Intelligent Newsletter

A generative AI model’s voiceover for a weekly AI journey.

Share This Story, Choose Your Platform!