“Arvind Krishna asserts that Large Language Models (LLMs) provide a 100x advantage in speed, tuning, and deployability compared to previous bespoke deep learning methods.”