Narayanan Questions Effect of Data Center Bans on AI
Princeton professor says efficiency gains outpace data center growth for AI advances.
Arvind Narayanan posted that efficiency improvements drive an order of magnitude more AI progress than data center growth, citing 5x per year conservatively versus 25 percent annual gains from new centers. He added that bans would leave existing capacity untouched. Alex Volkov replied that the measures would not pause AI development yet remain effective for scoring political points. The exchange centers on the limited reach of state level restrictions according to the posts shared.
What impact do data center bans or moratoria have on AI progress? Let's do some napkin math. A number such as "1 GW of capacity blocked" can be unintuitive. But we can take advantage of the fact that the efficiency of AI inference at a given capability level has been…
Reactions
3 postsAN 
Arvind Narayanan@random_walkerIf these numbers feel shockingly small, it may help to remember that 1. efficiency improvements drive an order-of-magnitude more AI progress than data center growth (5x/year *conservatively* vs 25% per year) 2. bans do nothing about all the capacity already built.
- likes: 23
- replies: 8
- bookmarks: 7
- reposts: 5
AV Alex Volkov@altryne@random_walker it's ineffective way to "pause AI" but it's a hella effective way to score political points.
- likes: 3
- replies: 0
- bookmarks: 0
- reposts: 1
NX nxthompson@nxthompsonThis suggests that datacenter bans are not as effective at slowing AI as advocates think. Which also means the pro-datacenter crowd doesn't need to flip out as much.
- likes: 14
- replies: 5
- bookmarks: 4
- reposts: 3
