@vgr: AI stack vs crypto stack, 15 layers from power to regulation

Submit
vgr@vgr

AI stack: 1. Power sources 2. GPUs 3. Cuda/alt kernel levels 4. Data movement magic sauce <— largest models 5. PyTorch 6. Frameworks (transformers, CNNs, diffusion) 7. Models 8. MCP etc <—- still fragile 9. Apps, agents, and APIs 10. Paranoid and lonely people <— corner case forcing function 11. Deepfake and disinformation markets (politics, porn) 12. Information economy slop 13. Visionary FAFOers 14. Techlash pearlclutchers 15. Cynical regulation politicians Crypto stack: 1. Power sources [PoW] 2. GPUs, CPUs, ASICs 3. Cryptography 4. Zero-knowledge magic sauce 5. [ feels like there’s a layer here ] 6. Engineering crypto primitives (merkle trees etc) 7. Blockchains and blueskyish anti-chain things <—- comparable to models? 8. RPCs, sequencers, IPFS, bridges type things 9. Wallets, exchanges and other chain agnostic tech 10. Degens and grifters <— corner case forcing function 11. Money laundering, ransomware crime markets 12. Stablecoins <— this is crypto slop 13. Visionary FAFOers 14. Techlash pearlclutchers 15. Cynical regulation politicians

farcaster.xyz

@vgr: AI stack vs crypto stack, 15 layers from power to regulation

submitted bymisha13,1831y
Read original article
AI summary of the linked article

@vgr sets out the AI stack and the crypto stack as parallel 15-layer models, running from power sources to cynical regulation. Both begin with power sources and GPUs, and the crypto list asks whether blockchains are comparable to AI models at layer 7. Both stacks put a "corner case forcing function" at layer 10, paranoid and lonely people for AI and degens and grifters for crypto, and both end with visionary FAFOers, techlash critics and cynical regulation politicians. The crypto list also flags a missing layer and labels stablecoins as "crypto slop."

Recommended by 1 curator
@tum4y.eth
Comments
No comments yetWhat caught your eye? Start the discussion.
Top stories that week →