Thermodynamic Theory of Algorithmic Catalysis Developed Within Watts-per-Intelligence Framework
Researchers have developed a thermodynamic framework called 'algorithmic catalysis' that identifies fundamental physical limits on how efficiently AI and computational systems can perform tasks. The work, submitted to the AGI-2026 conference, proves that any speed-up achievable by reusable computational structures is bounded by the algorithmic mutual information between the system and its task class, and that encoding this information carries an unavoidable energy cost via Landauer erasure. The findings matter because they place hard theoretical constraints on the energy efficiency of intelligent computation, with implications for the long-term sustainability and design of AI systems.
A new preprint on arXiv introduces a thermodynamic theory of 'algorithmic catalysis' as part of a broader 'Watts-per-Intelligence' research program aimed at understanding the energy costs of computation and intelligence. The authors prove that reusable computational structures — analogous to catalysts in chemistry — can reduce irreversible operations for a class of tasks, but only up to a ceiling set by the algorithmic mutual information between the computational substrate and the task class descriptor. Encoding that information into the system necessarily incurs a minimum thermodynamic cost through Landauer erasure, the well-established principle linking information deletion to heat dissipation. Combining these two results, the paper derives a 'coupling theorem' that establishes a lower bound on how long an algorithmic catalyst must be deployed before it becomes energetically worthwhile compared to non-catalytic approaches. The framework is demonstrated on an affine SAT problem class and is used to situate contemporary learned systems — such as large neural networks — within these information-thermodynamic constraints. The work is a camera-ready version accepted for AGI-2026 and represents a second installment in the Watts-per-Intelligence series.
What's missing
The framework is illustrated on a specific affine SAT class, and it is unclear how tightly the derived bounds apply to more general or real-world AI workloads. The relationship between the theoretical lower bounds and empirically measured energy costs in deployed systems is not established. Additionally, the paper does not address whether the Landauer-erasure cost dominates in practice relative to other sources of energy dissipation in modern hardware.
What different sources said
- arXiv cs.AICenter
Watts-per-Intelligence Part II: Algorithmic Catalysis
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.