Resolves YES if among frontier AI models released in 2027, a majority used FP8 as their primary pretraining precision. Frontier models = flagship from OpenAI, Anthropic, Google DeepMind, Meta AI, xAI, Mistral. FP8 means majority of compute in FP8 not BF16/FP16/FP32. Mixed precision where FP8 dominates counts. Post-training quantization does NOT count. Based on official reports or credible reporting. Undisclosed models excluded. Resolves NO if fewer than half qualify or insufficient info by March 2028.