← Back to context

Comment by bob1029

11 hours ago

I do think token demand might actually become satiated at some point. Machine to machine scenarios imply the potential for infinite sprawl, but machines don't run for free or forever. There is always a human with interest involved somewhere. You can try to hand wave the human away, but at some point you would be forced to reveal who is paying for the electricity, computers, real estate and internet connectivity.

Sure, but presuming next generation AI is nearly as high utility as a good as a junior engineer, a team of 1-3 seniors could be running projects with 10-30 junior engineers as is. Then consider that much of the throttle on adding more is cost per engineer matching cost of living oriented salary and demotivation effects of giving engineers projects that don't ship and don't further their career path.

I could easily see the equivalent of tens of thousands of engineers being used for companies that had dozens today, making prototypes of every imaginable solution to every RFE in case they are needed by management.

Cost will naturally serve as a limiter but I think most companies will find it hard to compete in their field with anyone who keeps costs similar to today by replacing salaries with energy and hardware capital and continues that cost level by turning all efficiency gains into new build out.