US Increases Investment in AI, Capital Market Divided

However, as models are increasingly applied to traditional workflows, corporate attitudes are becoming more rational. A survey by Ernst & Young found that 82% of US executives are concerned about the use of tokens and related costs.

A survey by S&P Global found that the proportion of companies abandoning model projects has doubled compared to the previous round of statistics. As model efficiency improves and market competition intensifies, model calling prices continue to decline, and technological progress has narrowed the profit space to some extent.

Venture capital firm a16z estimates that due to algorithm optimization and the popularization of open-source models, the average annual decline in reasoning prices for large models with equivalent performance is around 90%.