CN101 - A Digital Thermodynamic Computer for Generative AI

2026/08/01 by Lars Holdijk, Denis Melanson, Zier Mensch +14
Computer Science · #cs.ET #cs.AI

paper · pdf

arxiv created 2026/08/01 · arxiv updated 2026/08/04

Abstract

Thermodynamic computing is an emerging hardware paradigm, in which stochastic physical dynamics serve as the direct computational primitive. The recent explosion of generative AI has only sharpened the search for alternative approaches to compute, and, as we show in this work, thermodynamic computing turns out to be well suited to this space. An important class of methods realises a function as the stationary expectation of an ergodic stochastic process: the answer is encoded in the time-averaged statistics of an equilibrating trajectory. To date, this equilibration-style class has been formulated exclusively through Langevin dynamics, restricting its implementations to analogue substrates and the engineering challenges those bring. In this work, we propose a substrate-independent formalisation of the equilibration-style formulation, in which the only object of design is the dynamical generator L* of an arbitrary ergodic process. The formalisation makes three hardware-level properties of the formulation explicit: the precision of a result is a knob set by how long the dynamics are run, sample averages decompose across independent trajectories, and dependent stages of a computation operate concurrently rather than serially, a property we call sequential parallelism. We instantiate the formalisation by fabricating a prototype digital thermodynamic computing chip, named CN101, that implements the formulation through discrete accumulator dynamics on standard CMOS using stochastic computing principles. We characterise CN101's success across conventional generative AI workloads in the form of VAEs and flow matching, applied to both image generation and scientific problems. Together, the formalisation and its digital instantiation show that the equilibration-style formulation is substrate-independent, and that its computational properties can be exploited on standard digital hardware.

Citations

• An efficient probabilistic hardware architecture for diffusion-like models
• Solving the compute crisis with physics-based ASICs
• Generative thermodynamic computing
• Nonlinear thermodynamic computing out of equilibrium
• Thermodynamic Natural Gradient Descent
• Thermodynamic Computing System for AI Applications
• Thermodynamic Linear Algebra
• Parallel Sampling of Diffusion Models
• Thermodynamic AI and the fluctuation frontier
• Flow Matching on General Geometries
• Flow Matching for Generative Modeling
• Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
• Probabilistic computing with p-bits
• Score-Based Generative Modeling through Stochastic Differential Equations
• Denoising Diffusion Probabilistic Models
• The Hardware Lottery
• FFJORD: Free-form Continuous Dynamics for Scalable Reversible Generative Models
• Neural Ordinary Differential Equations
• Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
• Hardware-Driven Nonlinear Activation for Stochastic Computing Based Deep Convolutional Neural Networks
• SC-DCNN: Highly-Scalable Deep Convolutional Neural Network using Stochastic Computing
• Deep Residual Learning for Image Recognition
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• Stochastic thermodynamics, fluctuation theorems and molecular machines
• Nonphysical sampling distributions in Monte Carlo free-energy estimation: Umbrella sampling
• Efficient estimation of free energy differences from Monte Carlo data
• High-Temperature Equation of State by a Perturbation Method. I. Nonpolar Gases
• Gradient-based learning applied to document recognition