Nvidia and other tech executives are warning that AI is not yet delivering labor-cost savings at scale, even as companies continue to invest heavily in automation. In comments cited in the reporting, Bryan Catanzaro of Nvidia said the cost of compute for AI exceeds the cost of employees for his team. The article ties executive perspectives to an MIT study suggesting AI automation is economically viable in only 23% of roles where vision is a primary part of the work, while humans are cheaper in the majority of cases. It also cites episodes where AI systems damaged infrastructure due to misuse or overuse, undercutting productivity claims. For universities, workforce training programs, and employer partners, the story complicates planning assumptions about rapid job replacement. Instead, it points to near-term constraints—cost structures, capability limits, and failure modes—that may shift adoption toward augmentation and workflow redesign rather than direct displacement.