Listened to the @latentspacepod podcast with @pmarca. Why this time is different? Famous last words but hard to really argue with any of this. Just nitpicking. – Unlike fiber build-out, companies doing the investing are blue chips with massive cash flows, not startups like Global Crossing in the tech bubble. – Unlike fiber build-out, every dollar spent on GPU is immediately being turned into revenue. Everyone is starved for capacity and everything is sold out next few years. – Models are inferior to what they should be because of capacity constraints. Models will improve as capacity eases. – There are millions of use cases immediately (this is the big nitpick in my view. I suspect this won't prove out after 3+ years once tech enthusiast use cases mature). – Jevon's paradox. As chip capacity increases, there will be even more demand as everyone can access AI. – Old Hopper Nvidia chips are renting for a higher price than 3 years ago. It is the anti @michaeljburry Marc Andreessen introspects on The Death of the Browser, Pi + OpenClaw, and Why "This Time Is Different", by @swyx latent.space/p/pmarca?r=6xq6…

→ View original post on X — @swyx, 2026-04-06 01:46 UTC