DeepSeek has revolutionized AI training by dramatically cutting computing costs – reducing standard 32-bit model training to just 8 bits without sacrificing performance. This breakthrough, achieved mostly by a China-educated team, challenges the notion that Western-trained talent is essential for major tech advancements, says CCG senior researcher Andy Mok. More importantly, he believes that by making its model open-source, DeepSeek not only rivals OpenAI's ChatGPT but also empowers the Global South with accessible AI technology.