What is DeepSeek-R1, which spooked global AI market?

DeepSeek-R1's launch caused a significant decline in technology shares shaking investors' confidence in the AI sector's strong demand for high-tech chips

DeepSeek AI, a Chinese tech startup last week released its open-source AI model, DeepSeek-R1, which soon became the centre of attraction in the global market.

DeepSeek-R1, use multi-stage training and large-scale reinforcement learning (RL) techniques. ( Lam Yik/Bloomberg)

Its launch caused a significant decline in technology shares worldwide during the pre-market trading and reportedly shook investors' confidence in the AI sector's strong demand for high-tech chips.

While Nasdaq futures fell over 3%, S&P 500 futures tumbled nearly 2%.

Dominant AI chipmaker Nvidia's 8.4% slide led declines among heavyweight mega-cap stocks in premarket trading, with Microsoft off by 4%, Meta Platforms down 3.7% and Alphabet shedding 3.1%.

Such is believed to be the impact of DeepSeek AI, which has rolled out a free assistant it says uses lower-cost chips and less data, seemingly challenging a widespread bet in financial markets that AI will drive demand along a supply chain from chipmakers to data centres.

What is DeepSeek AI's DeepSeek-R1: All you need to know

• Fire-Flyer, a deep-learning division of High-Flyer, a Chinese quantitative hedge fund, gave rise to DeepSeek, an AI research lab.
• Last week, DeepSeek AI made headlines throughout the world when its open-source AI model, DeepSeek-R1, was released.
• An improved reasoning model called DeepSeek-R1 asserts that it outperforms current standards on several crucial tasks.
• To accomplish these capabilities, the model and its variations, like DeepSeek-R1, use multi-stage training and large-scale reinforcement learning (RL) techniques.
• High-Flyer, which was founded in 2015, rose to fame by using cutting-edge computing to analyze financial data. Liang Wenfeng, the company's founder, shifted efforts to DeepSeek by 2023 intending to produce ground-breaking AI models.
• Given the limitations imposed by the continuing tech battle between the US and China, DeepSeek's achievement is especially remarkable.
• Like all other Chinese-made AI models, DeepSeek self-censors on topics deemed politically sensitive in China.
• Unlike ChatGPT, DeepSeek deflects questions about Tiananmen Square, President Xi Jinping or the possibility of China invading Taiwan.
• The exact cost of development and energy consumption of DeepSeek are not fully documented, but the startup has presented figures that suggest its cost was only a fraction of OpenAI’s latest models.