Provocation: 'Today's Frontier AI Companies Will Never Exceed the AI Capability Frontier Again'

Andrew Trask (Substack)·low signal

Andrew Trask argues that the current crop of frontier labs may have structurally peaked relative to the moving capability frontier, contending that future gains will come from elsewhere in the ecosystem. The thesis sparked debate on Hacker News about whether incumbents' lead is durable or eroding. It's a speculative but discussion-worthy counter to the assumption that today's leaders compound indefinitely.