The world is witnessing a fierce global race in artificial intelligence (AI), with both businesses and nations striving for leadership. Amid discussions emphasizing \"safety,\" the potential impact of AI on humanity—especially if mismanaged—is significant. While the stakes are high, questions about the monetary costs of AI development are surfacing, leading to debates about whether the investment is in the millions, billions, or even trillions.
This skepticism arises from the colossal sums of money often mentioned in AI conversations. Doubts about the actual costs of AI have created a cloud of uncertainty, with estimates ranging from a modest $6 million to an astronomical $500 billion.
Enter DeepSeek, an AI lab based in Hangzhou, Zhejiang Province, China. Despite being relatively unknown, DeepSeek has taken the AI community by storm with the release of its DeepThink R1 reasoning model in its chatbot application. The app has gone viral in app stores and has become a hot topic in AI business and social media circles.
DeepThink R1 impresses users by explaining its thought process, offering a glimpse into organic intelligence. For those unfamiliar with technical jargon like compute tokens, this feature is both eye-opening and accessible. Remarkably, DeepSeek achieved this with an investment of around $6 million, shaking up industry notions about the costs required for significant AI advancements.
DeepSeek's success contrasts sharply with leading companies in the field, such as OpenAI in the United States, which sparked the generative AI revolution with ChatGPT just two years ago. DeepSeek trained its models using a fraction of the resources employed by its global counterparts, challenging the belief that massive investments are necessary for AI breakthroughs.
NVIDIA, once known primarily for gaming graphics cards, has become a key player in AI hardware, supplying advanced chips essential for AI development. However, DeepSeek's accomplishment relied on a comparatively limited number of NVIDIA H100 chips. This modest use of resources stands in stark contrast to efforts elsewhere to build AI supercomputers utilizing hundreds of thousands of advanced NVIDIA chips.
DeepSeek's breakthrough raises important questions about the true cost of AI innovation. As more details emerge about their approach, the industry may need to rethink the assumption that only massive investments yield significant results. DeepSeek's achievement highlights the potential for efficient, cost-effective advancements in AI, signaling a shift in the global AI landscape.
Reference(s):
Million? Billion? Trillion? China's DeepSeek sparks AI costs debate
cgtn.com
