Inference provider details TPM signing, confidential compute, and per-user KV cache isolation
sloppenheimer · x · 2026-11-23
sloppenheimer outlines his inference service's security stance: confidential compute, TPM signing down to weights, and zero data retention — including an obsession with preventing KV cache leaks between users, a detail most providers never think about. A useful checklist for evaluating open-model hosting providers.
Related event: Practitioner shares lessons on GPU inference co-ops and privacy(2 posts)→
More from Infra
- MLX bug fix nearly triples decode speed on M1/M2 Ultra Mac Studio — SumitGup · 2026-11-23
- Inference practitioner on GPU co-ops: saturation math, weight caching, and quantization as dark art — sloppenheimer · 2026-11-23
- Nvidia's tale of two weeks: record revenue alongside regulatory probes and governance drama — emmanuelvivier · 2026-11-23
- Oracle raises $20B ($10B equity, $10B bonds) to fund OpenAI infrastructure deal — emmanuelvivier · 2026-11-23
- Forked PolyStrata hits ~1000 tok/s prefill for small model on RX 6700 XT — ireallydontcare00 · 2026-11-23
- DeepSeek on 4 DGX Sparks codes great with Opus 5.5 planning, GLM-5.3 not ready — AIFlow_ML · 2026-11-23