In the sense that the incremental improvements in capabilities that we've been seeing in recent models seem to taking exponentially growing amounts of compute to achieve.