Million? Billion? Trillion? China's DeepSeek Sparks AI Costs Debate
The world is currently in the midst of a global AI arms race, with both businesses and countries vying for dominance. The frequent use of the word \"safety\" in AI discussions highlights the significant implications for humanity, especially if things go awry. While the stakes are sky-high, there's an undeniable buzz—or perhaps hype—surrounding the colossal sums of money being mentioned in conversations about AI development. The debate over the monetary cost of AI spans a range from about \$6 million to \$500 billion, fueled by two main factors.
At the lower end of the spectrum is China's DeepSeek, an unheralded AI lab based in Hangzhou, Zhejiang Province. DeepSeek has gone viral in app stores and within AI business and social media circles following the release of its DeepThink R1 reasoning model in its AI chatbot application. In the competitive and ever-evolving landscape of AI chatbots, where version numbers and technical jargon can be overwhelming, DeepThink R1 stands out by demonstrably explaining how it thinks. To the average person, unfamiliar with the nuances between compute tokens and crypto tokens, DeepThink R1 appears impressively intelligent and organically reasoning. Remarkably, this achievement by DeepSeek is said to have cost around \$6 million—a figure that has shaken up the industry.
As more details about DeepSeek emerge, comparisons are already being drawn. This Chinese lab is reported to have trained its model on just a fraction of the resources used by leading U.S. companies in the field, such as OpenAI, which kick-started the generative AI revolution with the launch of ChatGPT about two years ago. Nvidia, once known primarily for gaming graphics cards (GPUs), has become one of the most valuable companies globally by supplying essential AI chips. However, DeepSeek's success utilized a comparatively limited number of Nvidia's H100 chips, challenging U.S. efforts to build AI supercomputer clusters comprising hundreds of thousands of more advanced Nvidia chips.
DeepSeek's achievement raises pertinent questions: Do monumental budgets genuinely equate to superior AI models? Is the industry's emphasis on massive spending perhaps overrated? As the AI arms race intensifies, DeepSeek's innovative approach could redefine how we perceive the cost and accessibility of advanced AI development.
Reference(s):
Million? Billion? Trillion? China's DeepSeek sparks AI costs debate
cgtn.com
