# Memory as a Wasting Asset: Pricing Flash Endurance for Embodied Agents, and the Limits of Doing So

arXiv: https://arxiv.org/abs/2606.18144 · HTML: https://mlsanity.com/paper/2606.18144/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 64 references that are in the map.
- First posted on arXiv 16 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023)
- [SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics](https://mlsanity.com/paper/2506.01844/) (2025)
- [Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws](https://mlsanity.com/paper/2401.00448/) (2023)
- [Split Computing and Early Exiting for Deep Learning Applications: Survey and Research Challenges](https://mlsanity.com/paper/2103.04505/) (2021)
- [Device Placement Optimization with Reinforcement Learning](https://mlsanity.com/paper/1706.04972/) (2017)
- [ReMEmbR: Building and Reasoning Over Long-Horizon Spatio-Temporal Memory for Robot Navigation](https://mlsanity.com/paper/2409.13682/) (2024)
- [MemER: Scaling Up Memory for Robot Control via Experience Retrieval](https://mlsanity.com/paper/2510.20328/) (2025)
- [KARMA: Augmenting Embodied AI Agents with Long-and-short Term Memory Systems](https://mlsanity.com/paper/2409.14908/) (2024)
- [An Imitation Learning Approach for Cache Replacement](https://mlsanity.com/paper/2006.16239/) (2020)
- [Life-Cycle Emissions of AI Hardware: A Cradle-To-Grave Approach and Generational Trends](https://mlsanity.com/paper/2502.01671/) (2025)
- [Agent Memory: Characterization and System Implications of Stateful Long-Horizon Workloads](https://mlsanity.com/paper/2606.06448/) (2026)
- [Junkyard Computing: Repurposing Discarded Smartphones to Minimize Carbon](https://mlsanity.com/paper/2110.06870/) (2021)
- [Assessing the embodied carbon footprint of IoT edge devices with a bottom-up life-cycle approach](https://mlsanity.com/paper/2105.02082/) (2021)
- [Managed-Retention Memory: A New Class of Memory for the AI Era](https://mlsanity.com/paper/2501.09605/) (2025)
- [Memory-Bound but Not Bandwidth-Limited: The Physical AI Inference Gap in Batch-1 LLM Decode](https://mlsanity.com/paper/2605.30571/) (2026)
- [Toward Robust and Efficient ML-Based GPU Caching for Modern Inference](https://mlsanity.com/paper/2509.20979/) (2025)
- [AEGIS: A Backup Reflex for Physical AI](https://mlsanity.com/paper/2606.06660/) (2026)
- [Agentic AI Systems Should Be Designed as Marginal Token Allocators](https://mlsanity.com/paper/2605.01214/) (2026)
- [AURA: Action-Gated Memory for Robot Policies at Constant VRAM](https://mlsanity.com/paper/2606.02775/) (2026)
- [Beyond Per-Token Pricing: A Concurrency-Aware Methodology for LLM Infrastructure Cost Estimation](https://mlsanity.com/paper/2606.11690/) (2026)
- [On the Promise and Pitfalls of Optimizing Embodied Carbon](https://mlsanity.com/paper/2306.15816/) (2023)
- [Nemo: A Low-Write-Amplification Cache for Tiny Objects on Log-Structured Flash Devices](https://mlsanity.com/paper/2603.09605/) (2026)

## Abstract (arXiv)

A robot's flash endurance is a non-renewable stock: every persisted write spends one of a few thousand program/erase cycles and never refills, yet no fielded robot memory system prices which memories are worth an erase cycle. We treat embodied memory as depreciating capital and price that stock with a single endurance shadow price η, which makes cost-minimizing placement across a RAM / on-board NVM / cloud hierarchy a threshold in a wear-augmented per-byte index. The index is cost-optimal whatever the sign of the value-write association chi; only when chi > 0 does the optimum turn non-monotone, sending a robot's most valuable memories off its flash. The pivot is thus empirical, and we measure chi on real robot logs at a pre-specified gate: its sign is a property of the deployment regime -- positive on recurrent long-horizon manipulation (hatchi ≈ +1.0 × 10-3, replicated at full power), null on a shorter-horizon suite, and negative on non-recurrent teleoperation. Two boundaries scope the result. The endurance budget is dormant on premium 3,000-P/E TLC at datasheet prices and binding on the commodity QLC/eMMC (~1,000 P/E) that cheaper edge robots run. And where it binds, a learned wear-aware controller only ties price-based routing on task value, because realized value is tier-invariant across RAM, NVM, and cloud: the rent governs device lifetime and cost, not task performance. Whether wear-aware placement improves task value remains open -- chi is measured against a value proxy, and the non-monotone optimum, while proven, is not yet observed in data.