Bit2Watt: A CyberPhysical Vulnerability Exploiting GPU Workloads Across Power and Computing Infrastructures

Jul 16, 2026

Upload by Oliver

Bit2Watt identifies a cyberphysical vulnerability in which an adversary leverages legitimate GPU workload execution to generate controlled, high-frequency power modulations. These modulations destabilize local power infrastructure and can propagate to disrupt computing services, increasing fluctuations, harmonic distortion, and damping degradation in scenarios with high DER penetration. The attack is difficult to detect because distinctive behavior concentrates in frequency components weakly captured by routine telemetry. The paper validates the risk via impedance-based analysis, power-system simulation, and real GPU plus grid-connected PV inverter experiments, including worst-case aggregation impacts and potential cascading failures.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages28

Reading Time1 h 11 min

AccessFree to download

Abstract

Introduction

Motivation and Threat Model

Challenges and Proposed Methods

Bit2Watt and Watt2Bit Risks

arXiv :2607 .05993v 1 [ cs .CR] 7 Jul 2026 IACR Transactions on Cryptographic Hardware and Embedded Systems ISSN XXXX-XXXX, Vol. 0, No. 0, pp. 1–28. DOI:XXXXXXXX Bit2Watt: A CyberPhysical Vulnerability Exploiting GPU Workloads Across Power and Computing Infrastructures Zhouhao Ji, Kaikai Pan∗ and Wenyuan Xu Zhejiang University, Hangzhou, China [zhouhao0779@zju.edu.cn](zhouhao0779@zju.edu.cn), [pankaikai@zju.edu.cn](pankaikai@zju.edu.cn),[wyxu@zju.edu.cn](wyxu@zju.edu.cn) Abstract. Modern data centers increasingly rely on large-scale GPU clusters and on-site renewable energy resources, resulting in a tightly coupled cyberphysical system between computing workloads and power-electronic-dominated grids. In this paper, we reveal Bit2Watt, a previously unexplored vulnerability in which an adversary manipulates GPU workloads to induce controlled, high-frequency power modulations that destabilize local power infrastructure and propagate back to disrupt computing services. Unlike traditional attacks that compromise grid-side devices or communication channels, Bit2Watt operates entirely within the cyber layer as a legal tenant, which could amplify ﬂuctuations, harmonic distortion, and damping degradation, particularly in high-DER-penetration scenarios. This risk is diﬃcult to detect under routine cloud- and facility-side monitoring because Bit2Watt exploits legitimate workload execution paths and concentrates much of its distinctive behavior in high-frequency components that are weakly captured by common telemetry. We validate Bit2Watt through impedance-based analysis, power system simulations, and real-world experiments on GPUs and grid-connected PV inverters. Under the synchronized worst-case aggregation model studied in the paper, manipulating 1,000 GPUs in a 1-MW local power system with 90% DERs raises current THD to 46.8% and results in a damping ratio of-0.27. We further show that the resulting power-quality degradation can stress data-center power-delivery equipment, trigger protection mechanisms, and, in extreme simulated cases, induce cascading failuresin transmission-scale systems. In addition, we analyze a plausible Watt2Bit feedback path, including denial-of-service risks and covert information exﬁltration via EMI side channels. This work exposes a vulnerability at the intersection of modern computing and renewable-integrated power systems, highlighting the urgent need for cross-layer defenses that jointly consider workload scheduling and power electronics. Keywords: Cyber-physical security · Systemic vulnerability · Electric power networks · Computing infrastructure · GPU power modulation 1 Introduction Modern digital infrastructure is increasingly built upon the deep integration of largescale computing systems and electrical power networks. The rapid proliferation of GPUaccelerated workloadsdriven by artiﬁcial intelligence and high-performance computing (HPC)has dramatically increased both the magnitude and volatility of the power demand in data centers, which varies at sub-millisecond timescales. In parallel, power networks worldwide are transforming toward high penetration of inverter-control-based distributed ∗ Corresponding author. Licensed under Creative Commons License CC-BY 4 .0. energy resources (DERs), such as photovoltaics (PVs) . As a result, modern data centers and renewable-integrated grids together constitute a networked cyberphysical system, in which cyber-level workload scheduling directly drives physical-layer actuation through power-electronic interfaces and fast control loops. While prior research [SCB22, SMP18 , DUW17 , NNQAG18 , JFK17 , ZL25 , SZ21] has independently highlighted the security issues in power systems and computing systems, existing cyberphysical threat models largely overlook adversaries who do not compromise sensing, control or communication infrastructure, but instead exploit legitimate cyberlevel control inputs to induce cyberphysical security issues. M

FAQs

What is Bit2Watt, and what makes it different from traditional grid attacks?

Bit2Watt is a vulnerability where adversaries manipulate GPU workloads to induce controlled high-frequency power modulations. Unlike attacks that compromise grid-side devices or communications, it operates within the cyber layer as a legal tenant.

Why is the Bit2Watt risk hard to detect using routine monitoring?

It exploits legitimate workload execution paths and concentrates distinctive behavior in high-frequency components that common telemetry captures poorly, allowing it to evade typical cloud- and facility-side monitoring.

How does the paper validate the impact of Bit2Watt?

Validation combines impedance-based analysis, power-system simulations, and real-world experiments using GPUs and grid-connected PV inverters. It also studies synchronized worst-case aggregation effects on power quality and damping.

Jul 16, 2026

Upload by Oliver