AlphaGo Zero uses substantially less power to play, but it used astronomically more compute power to train & learn. According to OpenAI, AlphaGo Zero used more

AlphaGo Zero uses substantially less power to play, but it used astronomically more compute power to train & learn. According to OpenAI, AlphaGo Zero used more than 1,000 Petaflop/s-day, or about 4x higher than Alpha Zero and 100x more than Dota 1v1 or 10,000x more than VGG/ResNet. [1] The combination of better algorithms + more efficient hardware has significantly reduced the energy waste of that additional compute power. [1] https://openai.com/blog/ai-and-compute/ https://openai.com/blog/ai-and-compute/