“AI scaling laws have proven to be remarkably robust, and betting against them has historically been unsuccessful.”