Nous Research has made significant advancements in decentralized artificial intelligence by pre-training a 15 billion parameter language model using its innovative DisTrO approach. This new methodology allows for distributed training over the internet, enabling users globally to connect their computers for collaborative training. The DisTrO technology reportedly enhances communication efficiency between GPUs by a factor of 10,000, marking a substantial leap in the field. The growing interest in decentralized training is seen as a pivotal development, potentially reducing the dominance of large tech companies in AI. Enthusiasts and experts within the AI community have expressed excitement about these advancements, highlighting the importance of decentralized training as a key innovation for the future.