What is DeepSeek and why is it revolutionary?
DeepSeek is turning the world of AI upside down. With its powerful V3 and R1 models, the Chinese company combines pioneering efficiency with impressive language processing, offering a surprising alternative to better-known players such as OpenAI (ChatGPT) and Google (Gemini). But where did it come from and what makes DeepSeek different from all those other models?
The rise of DeepSeek
On 27 January, the tech world was rocked by Nvidia’s stock drop, the chip giant behind cutting-edge AI processors. In just one day, the company lost 17% of its value (nearly $600 billion), an amount comparable to Belgium’s entire GDP. The cause? The sudden emergence of a relatively unknown Chinese AI lab: DeepSeek.
The market panic was fuelled by a fundamental question: what if Chinese AI undermines the dominance of Western players like Nvidia? Despite Nvidia regaining some of its losses (the stock has since partially recovered), DeepSeek’s advance continues to cause worry. A breakthrough whereby high-performance AI is developed without reliance on US chips could dramatically shift the balance of power in the world. In addition, there are concerns about privacy protection of the online app and DeepSeek’s political censorship is very obvious (see Is DeepSeek safe?).
But what makes DeepSeek’s models different? And why is it causing such a stir worldwide?
“ The question for us is not why, but how. ”
Who is behind DeepSeek?
Outside China, Liang Wenfeng’s name is hardly known, but in the financial world he has long been known as the founder of High-Flyer, a hedge fund company. Liang’s fascination with mathematics and algorithms led to the establishment of DeepSeek, an AI lab owned by High-Flyer, in April 2023.
Initially, DeepSeek did not seem like a direct competitor to OpenAI (ChatGPT), Anthropic (Claude) or Google (Gemini). It started out as a research project, funded by a hedge fund company that focused on fundamental AI research rather than the race for the most powerful models. However, it soon became clear that DeepSeek was becoming a serious player. Barely a year after its founding, DeepSeek’s V2 model managed to impress. Wenfeng’s strategy –innovation over imitation – and massive investment in chip capacity gave the lab an edge over other Chinese players, while also challenging US companies.
This rapid rise did not come out of nowhere. High-Flyer already had years of experience with AI and founded another AI company back in 2019, investing $28 million in a deep-learning platform with an infrastructure that was impressive even for big tech companies. Wenfeng himself succinctly sums up his approach: ‘The question for us is not why, but how.’ (Interview with Liang Wenfeng, November 2024)
DeepSeek changes the rules of the game
DeepSeek’s can-do attitude is there for a reason. Its rise takes place within a broader geopolitical AI battle, in which China could potentially gain dominance. According to the Australian Strategic Policy Institute (ASPI), there is a high risk of Chinese monopoly in three out of the six major AI domains:
- machine learning,
- advanced data analytics,
- AI algorithms.
In the remaining three areas (chips, adversarial AI and NLP), China shares a leading position with the US, putting the global power struggle in artificial intelligence on edge.
But there is an Achilles’ heel: hardware. While China leads the way in AI algorithms, the chip industry is still dominated by Western (or allied) companies. Nvidia (US) develops the most powerful AI chips, TSMC (Taiwan) is the world leader in semiconductor manufacturing and ASML (the Netherlands) has a monopoly in essential semiconductor manufacturing machines. Until 2024, having a hardware advantage was assumed to be crucial for AI dominance, and America stopped selling the highest-performance chips to China in 2022. And therein lies DeepSeek’s real revolution: despite the limited number of chips in China, they were still able to emerge as a key player by being much more efficient with limited computing power.
Have you heard of these chatbots?
DeepSeek Chat
tool - online - DeepSeek
Le Chat
online chatbot - Mistral AI
ChatGPT
Online chatbot - OpenAI
Claude
chatbot - online - Anthropic
Gemini
tool - online - Google
Microsoft Copilot
chat tool - online - Microsoft
Jasper AI
copywriting tool - online - Jasper.AI
Vinnie De Craim
I believe technology only truly matters when it creates societal impact. With a background in sociology and technology, I work on digital innovation in the public sector—where thinking and doing go hand in hand.
My approach is hands-on and impact-oriented. I bring critical analysis, but I’m equally driven by building concrete, meaningful results. Whether it's generative AI, policy development, or science communication, I’m always looking to bridge knowledge and real-world application—with pragmatism and a keen eye for opportunity.
I also believe in the power of language and storytelling to make complex ideas accessible. My background in the arts helps me not just understand technology and policy, but translate them in ways that resonate with different audiences.
Expertise: statistics, generative AI/AI, data science
Johan Desseyn
Johan Desseyn is founder and researcher at Mpiris. He started Mpiris 2017 to help governments and organisations make good decisions. He does so by providing policy-oriented social science research. With Mpiris, Johan also responds fully to the digitalisation of our society. Where possible, he applies technology-based methods to answer his clients' questions.
For Johan and his team, the emergence of Generative AI meant not only looking for ways to use these tools internally in an effective and responsible way, but also committing to sharing the accumulated expertise with his clients in government and education.
Expertise: policy-oriented research, education and training, governments, labour market policy, social policy
Related courses
Machine learning and deep learning fundamentals
Workshop - Leuven - VIB, Training geselecteerd in de Belgische AI Factory Antenna (AIFA)
HPC Café
Seminarie - Online, Karlsruhe - KIT, HammerHAI, Training geselecteerd in de Belgische AI Factory Antenna (AIFA)