[Coinbase CEO: Company Reduces AI Spending by Nearly Half Through Better Default Settings, Routing, and Caching Mechanisms] Coinbase CEO Brian Armstrong stated on the X platform that the company has reduced AI spending by nearly half through infrastructure optimization (default settings, routing, and caching mechanisms), even as token usage has grown exponentially. The measures include: - **Better Default Models**: Defaulting to open-source/cost-effective models (e.g., GLM 5.2, Kimi 2.7). 91% of employees previously never reached usage limits. - **Smart Routing**: Automatically routing tasks to the most suitable model (cutting-edge models for planning, cost-effective models for execution), allowing AI to replace manual selection. - **Efficient Caching**: Significantly improving cache hit rates (e.g., LibreChat increased from 5% to 60%).