Self-Sovereign AI: Vitalik’s Local Compute Rebellion
He has laid before the world a “self-sovereign, local, private and secure LLM setup,” not as a boastful shrine to gadgets, but as a confession of hunger for independence. The rig runs entirely on his own hardware, a cold altar of silicon and will, using NVIDIA 5090 GPUs and open-weight models like Qwen3.5 with 35 billion parameters. The numbers marshal a simple argument: about 90 tokens per second, a pace not for poets in an attic, but for ordinary days-enough to pretend one can carry a thought through the fog of noise without begging the cloud for mercy.




