Estimated Maximum Cost of Using Gary Tan's AI Agents: $50,000–$100,000 Annually

Garry Tan, CEO of Y Combinator, estimated on Andreessen Horowitz's August 12 podcast that running AI agents at maximum performance could cost $50,000–$100,000 per year (approximately ₩66.65 million–₩133.3 million), assuming heavy token and compute resource usage. The figure is an estimate of peak-performance operating costs, not a disclosure of Tan's actual annual spending; claims that he personally spends $200,000 a year were not confirmed in the original podcast remarks. The estimate matters because it frames AI agent usage as a significant recurring expense comparable to a full-time salary, with implications for how startups and enterprises budget for AI tooling. Watch for follow-up data on real-world agent spending patterns and whether actual enterprise AI agent costs align with these projections.