Presentation: From Fab To Token - The State Of The Market
How semiconductor constraints are reshaping AI software architecture—from fab yields to token economics.
Why it matters
A deep look at the physical and economic constraints driving AI infrastructure decisions: where GPU scaling hits limits, how data-center expansion is bottlenecked, and what it means for model inference costs.
The key facts
10 to knowSpeaker: Jordan Nanos (SemiAnalysis research)
Topics: semiconductor constraints, data-center expansion, networking bottlenecks
Focus: GPU scaling performance, chip fab yields, tokenomics impact on inference
Source: InfoQ presentation (educational/technical deep-dive format)
Date: August 2026
Source: SemiAnalysis research on semiconductor constraints
Topics: GPU scaling, benchmark performance, networking bottlenecks
Scope: chip fab through model inference token economics
Speaker: Jordan Nanos
Format: presentation (likely recorded or slidedeck-based, lower depth than written analysis)
Go to the source
InfoQ AI/MLinfoq.com
Publisher excerpt: Jordan Nanos discusses how semiconductor constraints, data center expansion, and networking bottlenecks impact AI software architecture. Drawing from SemiAnalysis research, he shares insights on benchmark performance, GPU scaling, and tokenomics from chip fab to model inference. By Jordan Nanos