Frontier AI models are trained in data centers with 100k+ GPUs, high-speed interconnects, industrial cooling, and 24/7 uptime.
So when I hear claims that they can be trained on decentralized networks of consumer GPUs, I’m skeptical.
Can decentralized AI really compete? 🧵
A rare seat has opened on our investment team (our first in over two years!). Reverie is hiring an investment analyst/associate.
We hire sparingly and for the long term, so if you're interested in a tight-knit, intellectually adventurous (we have constant debates about all sorts of crypto and non-crypto things!), and NYC-based crypto VC role, you should apply.