DeepSeek, a Chinese startup, has created a powerful AI chatbot, R1, that rivals Google and OpenAI's creations, claiming to have developed it for a fraction of the cost. This has sparked debate about DeepSeek's claims and the effectiveness of US sanctions on China's AI sector.
DeepSeek claims to have trained R1 using 2,000 Nvidia H800 GPUs and spent $5.6 million, while OpenAI's GPT-4 cost over $100 million to train.
Some experts, including Palmer Luckey and Alexandr Wang, doubt DeepSeek's claims, suggesting they had access to more advanced chips and funding.
DeepSeek's founder, Liang Wenfeng, has defended the company's claims, stating that they have been "doing things according to our own pace" and prioritizing cost-effectiveness.
The debate highlights the ongoing competition between the US and China in AI development and the effectiveness of sanctions in slowing down China's progress.