New SPLICE publication on Confidential Virtual Machines

CAEC enables protected, plaintext memory sharing between Confidential Virtual Machines (CVMs), addressing a fundamental issue in how CVMs communicate today. CVMs are increasingly used to veil sensitive information from “snooping” hypervisors, but their disjoint memory model creates a performance bottleneck. A CVM’s memory is either shared with the hypervisor or isolated from everyone, including other CVMs. As a result, any data exchanged between CVMs must pass through hypervisor-accessible memory, forcing expensive encryption and decryption on every inter-CVM transfer just to preserve confidentiality and integrity.

Comparison of hypervisor-mediated communication, hypervisor-accessible shared memory, and CAEC confidential shared memory between CVMs
Fig 1. Communication modes between two CVMs.

CAEC removes this bottleneck. Building on Arm’s Confidential Compute Architecture (CCA), CAEC extends the firmware to introduce Confidential Shared Memory (CSM), a memory region that multiple CVMs can access directly while remaining inaccessible to the hypervisor (and to any CVM not part of the sharing group). This gives collaborating CVMs a fast, protected channel for exchanging data without ever exposing plaintext to untrusted parties.

To learn more, read the paper. To receive notification of SPLICE news, subscribe to this blog!

Sina Abdollahi, Amir Al Sadi, David Kotz, Marios Kogias, and Hamed Haddadi. CAEC: Confidential, Attestable, and Efficient Inter-CVM Communication with Arm CCA. Proceedings of the IEEE European Symposium on Security and Privacy (Euro S&P). July 2026. DOI 10.1109/EuroSP68448.2026.00056

Leave a comment

We’re SPLICE

SPLICE logo

The SPLICE research team consists of faculty, postdocs, graduate students, and undergraduate students from 8 different institutions across the United States. We look at smart-home security and privacy from a multi-disciplinary perspective, across the lifecycle of smart devices, with varied residential situations in mind.

Contact Us

← Back

Thank you for your response. ✨