Ep. 036 - $200 Buys $12,000 of Opus Tokens, We Bought Every Plan (AI Cloud TCO) | Max Kan, Jordan Nanos, Andrew Megalaa

AI subscription plans provide significant value not through direct subsidies, but because providers maintain high API margins while most users operate at roughly 10% utilization. Effective inference costs hinge on cache hit rates, where cache reads dominate token consumption and significantly lower overall expenses. Anthropic’s Opus models currently outperform competitors like OpenAI in value-per-dollar, particularly when accounting for token efficiency and cache management. While open-source models are increasingly capable of handling simpler tasks, frontier models remain essential for complex, agentic coding and cybersecurity workflows. The market for these subscriptions is evolving rapidly, with providers balancing competitive pricing against the need to maintain profitability as users shift between proprietary and open-source alternatives.