DeepSeek, a Chinese AI company, has created advanced AI models at significantly lower costs than expected, raising concerns about the U.S. maintaining its AI leadership. The company claims it trained its models for $5.6 million using less powerful chips, yet experts suggest that the actual cost, including prior experiments, was likely much higher. Their engineering innovations, while notable, rely on established AI techniques. This situation could impact future investments and the perceived necessity for high-end AI hardware, indicating a potential shift in the competitive landscape of AI technology.
The $5 million refers to the final training run of the system, but it likely cost significantly more in earlier experimentation stages.
DeepSeek's methods are complex yet rooted in established techniques and may signal a shift in the AI development landscape.
Collection
[
|
...
]