How do the frontier labs plan to make money?
retained text
To start with, AI is here to stay no matter what happens with the frontier labs, but I am still interested in knowing what people think of their fate. At least to me, their business model seems quite strange in a market where open-source models are competitive? Suppose the current models are good enough to automate most jobs to a good degree. That sounds like a good pitch, until you realize these models have an advantage only in the present. In a year or two (or maybe even sooner), we could have Astra-level lightweight open-source models. The frontier models of the time could very well be far ahead, but what practical difference does it make to 90% of the businesses out there? If a couple of humans + an Astra-level model can automate an accounting department very well, then why would the company pay extra for a frontier model? This is also beside the point that a lot of jobs don't even need Astra-level models to be made much more efficient... It seems like the top labs are banking everything on AGI/ASI. What if it's not achieved? Or at least, what if it's not achieved before the current rate of funding slows down? I don't see any way for them to survive in such a future. Maybe I am missing something? Happy to hear perspectives!
document fetched 2026-09-07 00:29 · canonical https://reddit.com/r/singularity/comments/1w90sll/how_do_the_frontier_labs_plan_to_make_money