DeepSeek Sends Shock Waves to Global Tech Sector

Sitting under a palm tree in Mexico reflecting on the major AI developments this past week, DeepSeek is the top story to share a perspective on.

DeepSeek woke up the tech industry with the launch of its AI LLM Model, open source model - DeepSeek-R1. Developed by Liang Wenfeng, based in Hangzhou. Founded in late 2023, DeepSeek's capabilities raised eyebrows as it rivaled the creations of Google and OpenAI.

DeepSeek touted that its new AI LLM model was developed using less advanced and fewer computer chips than employed by tech giants in the United States. In their research paper, DeepSeek stated they had spent less than $6 million to train their model – a fraction of the multibillion-dollar AI budgets funded by OpenAI and Google, creators of ChatGPT and Gemini.

Marc Andreessen, a well-known VC, declared the DeepSeek as “AI’s Sputnik moment”.

What transpired was legitimate fears that the high market valuations of Nvidia, Meta, and Open AI are inflated too high. Last week, Nvidia lost nearly $600bn in market capitalization after its shares plummeted 17 percent. Apart from Nvidia’s dramatic slide, Google parent Alphabet and Microsoft also saw their stock prices fall 4.03 percent and 2.14 percent.

US President Donald Trump said DeepSeek should serve as a?“wake-up call”?on the need for US industry to be “laser-focused on competing to win”.?

It is important to appreciate that DeepSeek’s arrival on the scene challenges the assumption that it takes billions of dollars to be at the forefront of AI.?

For example, OpenAI was founded 10 years ago, has 4,500 employees, and has raised $6.6 billion in capital. DeepSeek was founded less than 2 years ago, has 200 employees, and was developed for less than $10 million.”?

Although the markets reacted strongly to this announcement, many others were not surprised by China’s progress. Former research director at Stability AI, Tanishq Abraham, said China has always been forward-thinking in AI, given the rollout of various models by Chinese firms such as Alibaba and Baichuan.

“DeepSeek’s apparent progress is almost an example of this: by not having enough computational power to build models as large as ChatGPT, they had to be smart. Necessity is the mother of invention.”

DeepSeek's latest model V3 has a 671B parameter model and boasts solid performance while requiring significantly fewer resources than its peers. DeepSeek-R1 focuses on reasoning tasks and challenges OpenAI's model with its advanced capabilities.

DeepSeek’s API pricing is significantly lower than that of its competitors. DeepSeek-R1’s API costs just $0.55 per million input tokens and $2.19 per million output tokens, compared to OpenAI’s API, which costs $15 and $60, respectively.

DeepSeek’s open-source approach further enhances cost efficiency by eliminating licensing fees and fostering community-driven development. This allows developers to freely access, modify, and deploy DeepSeek’s models, reducing the financial barriers to entry and promoting wider adoption of advanced AI technologies.

These points, however, should be put in perspective, as the security risks with DeepSeek are very high. For example, cybersecurity firm?KELA reported?that they were able to jailbreak DeepSeek's model, enabling it to produce malicious outputs, including ransomware development, instructions for creating toxins, and the fabrication of sensitive content.

Recognize that all your data, including your personal information and, even worse, your employers, is going and staying in China.

It is important to appreciate that the war for AI LLM innovation has heated up even more with DeepSeek’s arrival. This also opens up the research consideration that large companies can create their own LLM models, without having to invest billions of dollars.

References:

Janakiram. January 27, 2025. “All about DeepSeek”? Source: Forbes

Power, J. Jan 28, 2025. What’s DeepSeek?? Source: Aljazeera.