AI and the Hunt for Ballistic Missile Submarines

The document provides a detailed scoring framework and analysis of how AI enhances the ability to search, localize, classify, and track ballistic missile submarines. It evaluates platforms ranging from fixed acoustic arrays (IUSS) and nuclear attack submarines (Virginia SSN) to unmanned systems like the Orca XLUUV and Slocum Gliders, as well as satellite-based SAR constellations. The analysis highlights that AI's impact is most significant in shallow waters for multi-modal platforms and in deep water for acoustic-heavy systems, while physical constraints of the ocean environment continue to limit AI's effectiveness at extreme depths.

• The Virginia-class nuclear attack submarine (SSN) is the only platform capable of effective operation across all three depth regimes and all four phases of the ASW detection chain. AI enhances its capabilities through automated target motion analysis (TMA) and the use of convolutional neural network (CNN)-based algorithms to distinguish target tonals from background shipping noise in shallow water.
• The Sea Hawk unmanned surface vessel (USV) achieves the highest AI impact score in shallow water (28), driven by its ability to use AI for autonomous maneuvering, contact re-acquisition, and the use of machine learning to improve active sonar classification via acoustic imaging.
• For fixed and deployable acoustic arrays (IUSS), AI impact is highest in the 'deep' regime (200–1,000 ft) because convergence zone and SOFAR channel propagation allow machine learning and distributed sensor fusion to exploit high signal-to-noise (S/N) signals over great distances.
• Unmanned underwater vehicles (UUVs) like the Slocum Glider and Orca XLUUV leverage AI primarily for persistence and swarm coordination. Slocum Gliders demonstrated the ability to autonomously concentrate around high-probability contacts and adapt course based on peer detection reports during the REPMUS exercise.
• Satellite-based Synthetic Aperture Radar (SAR) is limited to detecting surface-level signatures such as wakes, Bernoulli humps, and internal-wave wakes. Deep learning is applied to automate the detection and classification of these wakes, but the technology is ineffective for submerged contacts in deep or extreme depth regimes.
• The P-8A Poseidon aircraft's AI effectiveness drops significantly in deep water compared to shallow water because it loses its multi-sensor advantage; radar and electronic support measures (ESM) become ineffective below the thermocline, leaving only sonobuoys.

APA

Schlegelmilch, B., & Johns, E. (2026). AI and the Hunt for Ballistic Missile Submarines. The Council on Strategic Risks. https://councilonstrategicrisks.org/wp-content/uploads/2026/08/Schlegelmilch_Johns_ASW_Appendix.pdf

Chicago

Schlegelmilch, Barret, and Eliana Johns. AI and the Hunt for Ballistic Missile Submarines. The Council on Strategic Risks, 2026. https://councilonstrategicrisks.org/wp-content/uploads/2026/08/Schlegelmilch_Johns_ASW_Appendix.pdf.

Wikipedia

{{cite report |last1=Schlegelmilch |first1=Barret |last2=Johns |first2=Eliana |title=AI and the Hunt for Ballistic Missile Submarines |publisher=The Council on Strategic Risks |date=August 2026 |url=https://councilonstrategicrisks.org/wp-content/uploads/2026/08/Schlegelmilch_Johns_ASW_Appendix.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{schlegelmilch2026hunt, author = {Schlegelmilch, Barret and Johns, Eliana}, title = {{AI and the Hunt for Ballistic Missile Submarines}}, institution = {The Council on Strategic Risks}, year = {2026}, month = aug, url = {https://councilonstrategicrisks.org/wp-content/uploads/2026/08/Schlegelmilch_Johns_ASW_Appendix.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }