Chinese AI startup DeepSeek, founded by hedge fund manager Liang Wenfeng in 2021, has made waves in the global tech landscape, prompting alarm in Silicon Valley and beyond.
The launch of DeepSeek’s new AI model has sent shockwaves through financial markets, leading to Nvidia’s largest single-day stock drop for a public company.
Nvidia’s shares plummeted to a four-month low of $118.42 on Monday (27), erasing nearly $600 billion from its market value. This decline, fueled by concerns over DeepSeek’s cost-efficient AI training methods, triggered a tech sell-off across global markets, with Nasdaq dropping over 3 per cent and ripple effects reaching Tokyo.
Interestingly, Chinese tech companies like Tencent, Alibaba, and Baidu saw gains as Hong Kong’s Hang Seng index opened higher on Tuesday (28), underscoring the shifting dynamics in the AI race.
DeepSeek’s innovation lies in its ability to train a foundational AI model to achieve results comparable to leading competitors like ChatGPT and Meta’s Llama while using significantly fewer resources.
Reports suggest DeepSeek’s AI was trained using only 2,000 second-rate Nvidia chips, compared to 15,000 high-end chips typically required by its Western counterparts.
This breakthrough challenges the conventional understanding of the investment-heavy nature of training large language models (LLMs).
The realization that AI models can be developed at a fraction of the previous costs has raised concerns about the sustainability of demand for specialized hardware like Nvidia’s GPUs. This development also highlights China’s accelerating progress in AI, narrowing the perceived technological gap between the US and China.
The success of other Chinese models, such as Alibaba’s QwQ, reinforces the notion that these advancements are not isolated but part of a broader trend. Both DeepSeek and Alibaba’s achievements demonstrate that the cost barriers for AI development can be significantly lowered, a notion that has disrupted market sentiment.
The advancements by DeepSeek prompted US President Donald Trump to describe the development as a “wake-up call” for American industries. Speaking in Miami on Monday (27), Trump stated, “The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries that we need to be laser-focused on competing to win.” While acknowledging the potential of DeepSeek’s model, Trump emphasized the need for heightened competition in the AI sector.
Nvidia, reacting to the market rout, noted that DeepSeek’s work complied with US chip export controls. The company stated, “DeepSeek’s work illustrates how new models can be created using that technique, leveraging widely available models and compute that is fully export control compliant.” Nvidia also reiterated its strong demand for AI inference, a crucial phase in AI model deployment.
DeepSeek’s advancements have significant implications for nations like India. If foundational AI models can be trained more cost-effectively, the barriers to entry for developing proprietary AI systems are lowered.
This development is particularly crucial for countries with limited access to high-end GPUs and financial resources. It opens up opportunities for nations like India to establish themselves in the global AI ecosystem.
India’s AI strategy is currently under debate. While Infosys co-founder Nandan Nilekani argues against building foundational AI models from scratch, others, like Perplexity AI founder Aravind Srinivas, strongly disagree.
Srinivas stated, “He’s (Nilekani) wrong on pushing Indians to ignore model training skills and just focus on building on top of existing models. Essential to do both.”
Srinivas also highlighted DeepSeek’s achievements, urging India to reconsider its stance. “I hope India changes its stance from wanting to reuse models from open-source and instead trying to build muscle to train their models that are not just good for Indic languages but are globally competitive on all benchmarks,” he added.
India’s decision on whether to invest in foundational AI development or adapt existing open-source models will shape its role in the AI race.
DeepSeek’s cost-efficient methods could inspire Indian stakeholders to pursue both approaches simultaneously, ensuring competitiveness on a global scale.
DeepSeek began as an AI side project for Wenfeng’s trading fund, High-Flyer, using AI to identify patterns influencing stock prices.
Over time, the project evolved into a standalone AI venture, reflecting Wenfeng’s vision for leveraging AI beyond financial markets. The company’s breakthrough is a testament to the innovative potential of startups in redefining industry standards.
DeepSeek’s achievements have disrupted the global AI landscape, challenging the dominance of American tech giants and sparking debates on the economics of AI evolution.
For countries like India, the success of DeepSeek underscores the need to reassess AI strategies and capitalize on opportunities to build globally competitive models.
As AI continues to evolve, the ability to adapt and innovate will determine which nations and companies lead the next phase of technological advancement.
