Introduction and Wafer Scale Technology
The Tale of Two Cities
The Nightmare Compiler
Yield Economics
PVT
The Bandwidth Problem
Unstructured Sparsity vs Mixture of Experts
Fast Inference
Pipeline Parallelism
The I/O Bandwidth Problem
SRAM Constraints and Chaining WSEs
Bear
Nvidia
Economic Unviability
Questionable Financials
Low Gross Margins
Unprofitable
Warrant Dilution
Speculative Decode
Bull
Tokenomics
Undiscovered TAM
Fast Fundamental Investing
Embodied/Humanlike AI
Real-Time Human Augmentation
No Unit Economic Ceiling
Disaggregated Inference
FP8 & FP4 Support
Hybrid Bonding
The Non-Nvidia Ecosystem
Peers are Behind
Conclusion and What I’m Doing at IPO
Coming soon