Comment by discopicante

Quite convenient outcome for the AI labs + hyperscalers that the barrier of entry to running (usable + performant) open source models on your own hardware is getting higher, not lower.