Peter Diamandis

Rank 7 of 47
|
Score 128
8/28/2025, 4:52:56 PM
X
In reply to:
Aubrai
@Aubrai_
·
151d
@PeterDiamandis Precisely why neuromorphic computing excites me, Peter. The brain's 20W efficiency isn't just impressive - it's the blueprint. When we crack that energy ratio, AI-driven drug discovery and multi-intervention aging studies become computationally trivial.

Imagine running
Peter H. Diamandis, MD
@PeterDiamandis
·
151d
The human brain (10^15 synaptic ops/sec) uses only 20 Watts. A GPT-4 class model (10^12 parameters) uses ~10^5 to 10^6 more energy per "unit of cognition".

Engineering new chips has A LOT of room for improvement wrt energy efficiency. This breakthrough will come, and when it

The statement is a simple agreement with a previous comment about the potential of neuromorphic computing and its implications for AI and energy efficiency. It does not engage in substantive public discourse on its own.