Analysts at a16z found that in the top 1% of companies, spending on AI tokens per employee nearly equals the average annual salary of an engineer. This cost has grown exponentially recently, suggesting engineers could soon be 'cheaper' than AI usage by the end of the year.

AI tokens are the computational credits companies pay to run large language models and other AI services. Despite high token costs, companies heavily using AI increased their workforce by 10.2% within two years, while those spending less on AI saw almost no growth.