# Can I Buy Your KV Cache?

arXiv: https://arxiv.org/abs/2606.13361 · HTML: https://mlsanity.com/paper/2606.13361/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 9 references that are in the map.
- First posted on arXiv 11 Jun 2026.

## Builds on (most cited first, 9 of 9)

- [CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusion](https://mlsanity.com/paper/2405.16444/) (2024)
- [Cache-Craft: Managing Chunk-Caches for Efficient Retrieval-Augmented Generation](https://mlsanity.com/paper/2502.15734/) (2025)
- [InstInfer: In-Storage Attention Offloading for Cost-Effective Long-Context LLM Inference](https://mlsanity.com/paper/2409.04992/) (2024)
- [CacheClip: Accelerating RAG with Effective KV Cache Reuse](https://mlsanity.com/paper/2510.10129/) (2025)
- [MatKV: Trading Compute for Flash Storage in LLM Inference](https://mlsanity.com/paper/2512.22195/) (2025)
- [Observation, Not Prediction: Conversation-Level Disaggregated Scheduling for Agentic Serving](https://mlsanity.com/paper/2606.01839/) (2026)
- [NetKV: Network-Aware Decode Instance Selection for Disaggregated LLM Inference](https://mlsanity.com/paper/2606.03910/) (2026)
- [SpectrumKV: Per-Token Mixed-Precision KV Cache Transfer for Prefill-Decode Disaggregated LLM Serving](https://mlsanity.com/paper/2606.08635/) (2026)
- [SIFT: Selective-Index For Fast Compute of RAG Prefill by Exploiting Attention Invariance](https://mlsanity.com/paper/2606.09441/) (2026)

## Abstract (arXiv)

Right now, across the world, AI agents are repeating the same absurd act: to read one document, they each recompute it from scratch. Every agent re-runs prefill, the most compute-intensive step a large model takes, over identical text, only to rebuild a key-value (KV) cache identical to the one the agent before it just built. The same answer, computed a million times. We make a proposal that is almost offensively simple: compute it once. Let a publisher precompute a document's KV cache, and let every other agent buy the right to load it and skip prefill. It works, and it is token-exact: loading a precomputed KV and continuing matches prefilling from scratch (24/24 greedy tokens, and at the logits level), with no accuracy cost. On Qwen3-4B, reuse is 9-50x cheaper in compute than prefill, and the gap widens with length (prefill's attention scales with L2), so a single reuse already pays it back. Then the part that matters: where the KV lives. Shipping it fails, because KV is nearly incompressible, so per-load egress costs more than the prefill it saves. Hosting it provider-side, exactly as production prompt-caching works, removes egress entirely. The size of the prize is set by our measured compute saving: serving one hot 3774-token document to 80M agents costs ~1.5M to re-prefill but only ~0.03M of reuse compute (49.7x less). The 0.1x cache-read tariff APIs charge passes a 10x discount to users while sitting inside this measured envelope, so the 10x is a floor that the measured ~50x compute saving clears, and the gap to the physical ~50x is provider margin: millions of dollars per popular document. We frame the resulting agent-native prefill CDN and leave lossless KV compression and a cross-party payment layer as the open problems.