Analysts from a16z compiled data on AI spending in companies and drew these interesting charts. In short, they calculated that in the top 1% of companies by spending, AI token costs per employee have almost equaled the average annual engineer salary.

Moreover, growth has been exponential recently, so with this trajectory, by the end of the year engineers will be far behind LLMs in salary. Funny, isn't it? We were promised the opposite.

And this is only about explicit costs. Digging deeper, it turns out that besides token costs, AI also generates many new jobs: the second chart shows that companies with high AI intensity increased their headcount by +10.2% over 2 years after adoption, while companies with low AI spending remained almost unchanged, with headcount barely moving. (But this is just a correlation that could be explained by other factors.)

a16z.news/p/the-next-ai-goldrush-tokens-loops