# Thermodynamic Limits of Physical Intelligence

arXiv: https://arxiv.org/abs/2602.05463 · HTML: https://mlsanity.com/paper/2602.05463/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 6 of its 17 references that are in the map.
- First posted on arXiv 5 Feb 2026.

## Builds on (most cited first, 6 of 6)

- [Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling](https://mlsanity.com/paper/2304.01373/) (2023)
- [From Entropy to Epiplexity: Rethinking Information for Computationally Bounded Intelligence](https://mlsanity.com/paper/2601.03220/) (2026)
- [Transformers As Approximations of Solomonoff Induction](https://mlsanity.com/paper/2408.12065/) (2024)
- [Large Language Models as Computable Approximations to Solomonoff Induction](https://mlsanity.com/paper/2505.15784/) (2025)
- [Universal AI maximizes Variational Empowerment](https://mlsanity.com/paper/2502.15820/) (2025)
- [Scenarios and branch points to future machine intelligence](https://mlsanity.com/paper/2302.14478/) (2023)

## Abstract (arXiv)

Modern AI systems achieve remarkable capabilities at the cost of substantial energy consumption. To connect intelligence to physical efficiency, we propose two complementary bits-per-joule metrics under explicit accounting conventions: (1) Thermodynamic Epiplexity per Joule, new bits of structure about a specified environment-instance variable encoded in an agent's state per unit energy, and (2) Empowerment per Joule, sensorimotor channel capacity per expected energetic cost over a fixed horizon. These give two axes of physical intelligence, recognition versus control, but the resulting numbers are benchmark-relative rather than universal. Drawing on stochastic thermodynamics, we formulate a Landauer-scale closed-cycle benchmark for epiplexity acquisition by combining a thermodynamic-learning inequality with data processing, and clarify why boundary closure is required; conversely, a decoupling construction shows that without such assumptions information gain and in-boundary dissipation need not be tightly linked. For empirical settings where the latent structure variable is unavailable, we recommend compute-bounded MDL epiplexity / compression-gain surrogates. Finally, we propose a unified efficiency framework with a minimal checklist of conventions for relative bits-per-joule comparisons, and give a compact language-model reporting example.