DeepSeek’s Shock Debut Shook Silicon Valley – But Has It Changed AI?
When Chinese AI firm DeepSeek released its chatbot, DeepSeek-R1, in late January, few in the United States had heard of it. Within 48 hours, it was the most downloaded free app on Apple’s US store – and by Monday morning, it had rattled global markets.
The company claimed its system could match OpenAI’s ChatGPT while costing only a fraction to build. Investors took notice: shares in Nvidia, the chipmaker at the centre of the AI boom, dropped 17% in a single day, erasing $600 billion in market value – the largest one-day loss in US stock market history. Other AI-linked stocks fell in its wake.
For some, the launch was a “Sputnik moment” for AI, signalling that China might not be trailing the US in large language models after all. DeepSeek’s engineers said they had built a competitive chatbot for just $5.6 million, compared to the billions spent by American rivals in 2024. The claim challenged the prevailing belief that only vast infrastructure, enormous datasets, and massive computing power could produce cutting-edge AI.
Six months later, DeepSeek’s name rarely dominates headlines. Yet the app still has users – including some US startups keen to save on licensing fees. Many, wary of Beijing’s reach, run the model locally to avoid sending data to servers in China. The US State Department has warned that the company could provide support to Chinese military or intelligence services, an allegation DeepSeek has not addressed publicly.
The app’s sudden success forced American AI leaders to reconsider their approach. OpenAI’s recent release of smaller, open-source models was seen by some analysts as a response to DeepSeek’s efficiency-first strategy. Still, the industry’s biggest players are doubling down on scale: OpenAI has rolled out GPT-5, Meta is pouring billions into AI talent, and hyperscale data centre construction is accelerating.
Meanwhile, DeepSeek’s momentum has slowed. Reports suggest its follow-up model, DeepSeek-R2, has been delayed due to a shortage of high-end chips – the same bottleneck affecting AI firms worldwide.
For now, the AI race appears to have reverted to its pre-DeepSeek trajectory: ever-larger models, more data centres, and higher spending. Whether the Chinese upstart’s lean-engineering message has left a lasting mark remains an open question.
