Discussion about this post

User's avatar
The AI Architect's avatar

The VRT mention caught my attention bc most analysis overlook how data center plays scale differently than chip makers. Had a conversation with someone running infra for a mid-size AI firm—they said cooling and power distribution constraints are becomming the real bottleneck, not compute. The perps angle makes sense here since infrastructure capex cycles move slower but more predictably than semiconductor hype cycles.

1 more comment...

No posts

Ready for more?