Answering The "Trillion Dollar Question": How China's AI Models Compete On Cost Efficiency

In recent weeks, there have been two key tensions surrounding the ongoing historic AI rollout:

• the first one is the increasingly explosive debate over surging token costs (which we discussed most recently in "The Value Didn't Arrive": Bain Finds Cost-Savings From AI Falling Far Short Of Projections", "From Singularity To Tokenomics: The AI Narrative Hit A Snag" and "From Singularity To Tokenomics, Part II: The Subsidy Just Ran Out", and which the FT addressed today in "We created a monster’").
• The second one is the latest crackdown by the US government on Anthropic's models, banning foreign usage of Claude Fable 5 which forced Dario Amodei to disable all access to the new LLM (Anthropic is currently in talks with the Admin to address security flaws in the model).