Education

Sailor’s First – Aligning the Leadership Continuum

Industrial Base

The United States Can’t Deter China Without Allied Shipyards

Force Structure

Mass Drones to Save Missiles: A High–Low Mix for the Pacific

Arctic

Can an Interagency Task Force Work in the Arctic?

Commentary

Enduring the Storm: Reflections on the U.S. Navy’s “Fat Leonard” Scandal

Industrial Base

Useful Lemons

Readiness

Archers Need Arrows: Deficiencies in U.S. Submarine Munitions

February 26, 2026 Guest Author 3 Comments

By Alana Davis

In 2023, the Center for Strategic and International Studies (CSIS) wargamed a conflict between the United States of America and the People’s Republic of China (PRC). Reflecting 24 iterations of the wargame, the study weighed if China could succeed in invading Taiwan in 2026 and examined the variables affecting the outcome. Although CSIS concluded that China is unlikely to succeed, it found the result to be highly contingent on posture, weapons, and platforms. Crucially, one of the most determinant factors is U.S. submarine dominance in the undersea domain.

The report recommended prioritizing full-spectrum undersea warfare in planning for a potential large-scale, cross-ocean military conflict. This prioritization reflects the potency of the submarine force: Fast Attack Submarines (SSNs) torpedoing adversary commercial shipping and naval forces as Guided Missile Submarines (SSGNs) strike key adversary infrastructure with long-range cruise missiles.

But what happens when the archers run out of arrows – when submarines expend their weapons in the first battle of the next war? Does the U.S. have the inventory to support necessary reloads? Are the ports, vessels, and personnel ready to conduct the rapid reloads required to maintain pressure through a protracted war? If the current munitions stagnation continues, the answer is no. The Navy should work with the Department of War (DoW) and Congress to increase weapons supply and reinforce the means to conduct expeditionary submarine weapons transfers.

Recent Weapons Production and Expenditure

The Fiscal Year (FY) 26 Defense Budget prioritizes revitalizing the defense-industrial base with a notable increase in ship and weapons production. The National Defense budget request rose 13% from last year, topping $1 trillion, while the President has called for a $1.5B topline. In December 2025 the DoW announced an expansion of an existing RTX contract to order 219 Block V Tomahawk Land Attack Missiles (TLAMs) – the largest order in years, and a nearly 10-fold increase from the 22 planned for purchase in FY25.

Unfortunately, this sharp increase barely covers recent expenditures. The Eisenhower Carrier Strike Group alone expended 125 TLAMs against Houthi targets in Yemen. SSGNs conducted multiple strikes against Houthi Targets and enabled the B-2 Bomber strikes on Isfahan, Iran. Assuming a TLAM stockpile of roughly 4,000, U.S. naval forces in the Middle East depleted this missile’s inventory by 3% in relatively limited strikes against Iran and its proxies. This is a frightening statistic when contemplating the expenditures from all-out war with a near-peer adversary like the PRC. This troubling consideration is not limited to land attack missiles: A House-commissioned CSIS simulation estimates that in a Chinese invasion of Taiwan, the Navy could run out of long-range anti-ship missiles in less than a week of fighting.

Weapons production and delivery holdups reflect 1990s production halts after the end of the Cold War, unstable procurement continuing into the 2000s, and an increasing scarcity of U.S.-based manufacturing of certain critical parts like rocket motors and processors due to obsolescence challenges. The limited missile inventory is not the only problem. Diminishing submarine weapons on-load readiness stems from aging submarine tenders (ASes), which were commissioned in the 1970s, and the logistical complexities of loading weapons in foreign submarine ports.

What should the DoW and the Navy prioritize to ensure continued lethal armament of the submarine force? Action should include a two-pronged focus: one, creatively and efficiently increasing TLAM and torpedo supply, and two, investigating and investing in the ports and support vessel ability to conduct submarine weapons transfers.

Action 1 – Advance Submarine Munitions Supply

Military leadership and civilian defense experts agree that submarines are a key asset enabling U.S. victory in future naval conflicts. Instead of throwing money broadly towards munitions production, the DoW should prioritize making weapons that the bulk of both U.S. naval forces and U.S. allies can deploy.

The U.S. should focus on TLAMs because they are versatile – launched from SSNs, SSGNs, Ticonderoga-Class cruisers (CGs), and Arleigh Burke-class destroyers (DDGs) – totaling approximately 55 submarines and 83 surface ships. The United Kingdom, Japan, Australia, and the Netherlands all use TLAM – greatly increasing weapon production efficacy through scale. For similar reasons, the U.S. should also focus on Mk-48 ADCAP production, utilized by all 69 submarines in the U.S. fleet plus many Australian, Canadian, and Dutch vessels.

Additionally, efforts must be made to expedite weapons stockpile growth through manufacturing contracts and partnerships that encourage “close enough” component solutions rather than perfection. The Navy should be allowed to make minor compromises on weapon specs without compromising safety or viability. In November 2025, the DoW’s Strategic Capabilities Office announced open solicitations for a new, affordable SSN heavyweight torpedo called the Rapid Acquisition Procurable Torpedo (RAPTOR) to augment the Mk-48 ADCAP. Producing a torpedo at $500,000 per weapon vice the current $4 million per weapon is certainly enticing, given the many potential targets, but it does not mean production efforts and methods should slow on parts for the Mk-48 ADCAP. Promoting newer, cheaper technology is key, but continuing production of the tried-and-true ADCAP is also essential.

Furthermore, if compromises must be made between TLAM and ADCAP production investment, the Navy should prioritize the Mk-48 ADCAP because of its greater efficiency in sinking enemy ships and reinforcing a strategy of deterrence by (sea) denial.

Another production avenue worth investigating is shared weapons production with allies. The U.S. continues to lean on co-manufacturing partnerships with Australia and South Korea to re-supply depleted 155-millimeter artillery shells from the Russo-Ukrainian war. Similar co-production agreements should be signed with Australia and the UK as part of the AUKUS submarine partnership, as well as with Japan for manufacturing of parts for the TLAM and/or the ADCAP. Production of critical weapons and weapons components in strategic foreign locations strengthens U.S. logistics networks and shortens operational timelines. Weapons stockpiling in strategic locations improves deterrence, as allied power projection becomes more credible with the proximity of weapons – though this forward staging must incorporate defense, dispersal, and deception to mitigate against enemy strikes.

Action 2 – Strengthen Submarine Munitions Re-Supply Capability

In the Western Pacific, the U.S. maintains three bases capable of submarine weapons handling of TLAMs and ADCAPs: Yokosuka and Sasebo, Japan and Apra Harbor, Guam. Additional foreign port reload sites may include Subic Bay, Philippines; Souda Bay, Greece; Sterling, Australia; and Diego Garcia. These reloads are aided by the two remaining Guam-based submarine tenders, the USS Frank Cable and the USS Emory S. Land, which were specially designed to travel to submarines and assist in conducting weapons transfers, repairs, and nuclear-level maintenance. This small but mighty AS fleet continues to demonstrate its utility, such as in 2022 when the Frank Cable supported the first TLAM reload conducted by a U.S. submarine in Australia on the USS Springfield (SSN-761).

But these tenders are over 45 years old. They have outlived their intended lifespan and their ability to deploy safely comes into greater question with each passing year. As of July 2025, the Pentagon awarded $72.6 million to General Dynamics-NASSCO to continue developing up to three “AS(X)” class submarine tenders. With both existing tenders slated to decommission by 2030, time is quickly running out to replace these unique and valuable assets. Still, a net of only one additional tender by 2030, assuming production deadlines are met, is not enough given that by 2028 the Navy aims to boost submarine production to three SSNs a year (one Columbia Class and two Virginia Class). Further, one must carefully consider where to homeport these assets, focusing on Japan and/or Australia for maximum operational flexibility.

Besides investing in the rapid production of the new AS(X) class, the Navy should invest more in the infrastructure of the submarine bases themselves – namely Apra Harbor, Guam. Apra Harbor relies on the island’s public power authority which supplies energy via import-reliant petroleum plants with 50-year old generators susceptible to natural disaster, not to mention deliberate attacks. The unreliable power supply alone threatens the likelihood of efficient weapons transfer and maintenance stops for submarines on their way to a fight in the Pacific. Additionally, concerns over adequate equipment like heavy-lift cranes and trained personnel to conduct efficient submarine weapons reloads also remain.

The Navy should thoroughly investigate the real capacity of its overseas submarine ports to conduct efficient and safe submarine weapons transfers in a simulated wartime scenario. This analysis should answer the questions: How long does it take to move weapons inventory, re-load equipment and crews, and a submarine tender as applicable to various ports? Which ports lack critical equipment or trained personnel to conduct short-notice reloads? What is each port’s and each tender’s maximum reload ability and fastest reload pace? The last publicly documented transfer of a Mk-48 training shape to a submarine was in 2021 between the Frank Cable and the USS Hampton (SSN 767). Five years may as well be ancient history when facing today’s emerging adversary threats. There must be steady effort to test these vessels and ports in wartime conditions and pace, but compromises can also be made. For example – the Navy may be able to withstand AS(X) delays by ensuring all foreign submarine port call locations have heavy-lift cranes.

Conclusion: Make More of What Works and Make What Works Better

U.S. submarines remain a dominant and lethal force, but in the 21^st century, their lethality is jeopardized by two weapons concerns: rapidly depleting TLAM and Mk-48 ADCAP inventories, and inadequate weapons reloading facilities. The solution is not just to throw more money toward the problem. Since FY24 the DoW has invested hundreds of millions into weapons development and submarine tender design. The DoW and U.S. Navy must make more of what works by continuing production of versatile and battle-proven weapons. The United States should make what works better by improving how allied foreign ports and strategic assets can perform in wartime.

For the U.S. submarine fleet to dominate in naval conflict, it must have ample weapons stockpiled in strategic locations with all enabling infrastructure ready to support time-sensitive reloads. The first steps in ensuring continued dominance include: acknowledging the submarine force has critical weapons-related shortfalls, and studying which inventories, which bases, and which production lines are most vulnerable.

Submarines can operate within Surface Weapons Engagement Zones and conduct long-range tactical fires. In a target-dense environment, submarine munitions will deplete rapidly. In a conflict with the PRC, some estimates suggest an SSN will expend its inventory of 20 to 50 torpedoes within two weeks on station, and an SSN or SSGN will launch all their 12 or 154 TLAMs, respectively, within three weeks. At such rates of fire, it is easy to see how weapons inventory and reload pace become critical to continuing, and winning, the future fight.

Archers need arrows. If Congress and the U.S. Navy do not act now to ensure submarines stay armed and ready for battle, munitions problems will only worsen – leaving the force, the fleet, and country more vulnerable.

Lieutenant Alana Davis, U.S. Navy, is a submarine officer serving as a Force Manpower Planner under OPNAV N1 in Arlington, VA. She is a graduate of Harvard University (BA ‘19) and The University of Florida (MBA ‘26). The views presented are hers alone and do not necessarily represent the views of Department of War or the Department of the Navy.

Featured Image: Conceptual drawing of the Virginia-class attack submarine from 2004. [credit: wikimedia]

Indo-Asia-Pacific

Taiwan’s Layered Air Defence and the Calculus of Deterrence

February 24, 2026 Guest Author 1 Comment

By Guarav Sen

In any future Taiwan Strait conflict, the opening phase would be decisive – not because it guarantees victory, but because it shapes escalation, operational momentum, and political decision-making. The identification of a centre of gravity in Taiwan’s defence is therefore contingent on the People’s Liberation Army’s (PLA) campaign objectives, which vary across firepower-strike, invasion, and blockade scenarios.

Taiwan’s integrated air defence system functions as a scenario-dependent operational centre of gravity, most clearly in a PLA firepower-strike or decapitation campaign. While unlikely on its own to determine the outcome in all contingencies, integrated air defence plays a central role in shaping the battlespace. An analysis of Taiwan’s air defence is particularly salient as nations assess lessons from the recent U.S. strike and leadership-targeting operations, and recognize that neutralizing defence air systems is a critical enabling capability in invasion and blockade scenarios.

By denying rapid air superiority and preserving Taiwan’s combat power, the integrated air defence complicates efforts to achieve a swift fait accompli and raises the costs and risks of PLA operations.¹This article examines the interplay between integrated air defence and the broader PLA campaign options, assessing how its survivability influences the feasibility of coercion, blockade, and amphibious invasion.

PLA Campaign Logic Across Scenarios

PLA operational planning emphasizes methodical sequencing of action rather than a single decisive engagement. A campaign could shift from initial firepower-strike and paralysis efforts toward a coercive blockade or, if conditions permit, an amphibious invasion. Each option places different demands on air superiority, command and control, and escalation management, making Taiwan’s air defence posture central to shaping the viability of PLA courses of action.

The Opening Salvo: Fire and Electrons

A PLA campaign could begin with a multi-domain strike designed to induce strategic paralysis, rather than a fleet posture offshore that is immediately detectable, attributable, and escalatory.² Roughly 900 short-range ballistic missiles fielded by the PLA Rocket Force are aimed at Taiwan, alongside hundreds of land-attack cruise missiles and long-range guided rockets. Such weapons place island targets at risk from mainland firing points.

However, as Russia’s campaign against Ukraine demonstrates, even sustained missile and drone saturation struggles to produce strategic paralysis against a defended state and instead yields diminishing returns as air and missile defences adapt.³ The experience of Ukraine offers a useful comparison. Large-scale saturation attacks using missiles and one-way attack drones have imposed high costs and strained air defences, but have failed to produce strategic paralysis, instead pushing the conflict toward prolonged attrition as a functional, integrated air defence remains in place.

Concurrent with missile barrages, the PLA Navy Air Force could unleash thousands of precision-strike sorties in the initial days.⁴ Its Eastern and Southern Theatre Commands already field a modern fleet of over 950 fighters and 300 bombers or attack aircraft.⁵ This includes J-16 multirole fighters and low-observable J-20s armed with long-range PL-15 air-to-air missiles.⁶ Air bases, ports, radar sites, command-and-control nodes, and surface-to-air missile batteries would be the primary targets in Taiwan to blind, break, and constrain Taiwan at sea and in the air.⁷

Kinetic barrages will be combined with non-kinetic operations. The People’s Liberation Army (PLA) would use a combination of electronic warfare and cyber tools to interfere with early warning radars, jam satellite communications, penetrate networks, and use decoys to drain finite interceptor stocks.⁸ The doctrine of “systems destruction warfare” seeks to collapse an adversary’s operational architecture rather than engage in platform-versus-platform attrition, although they possess the numbers to do so.⁹

Special Operations, Air Assault, and Shaping Actions

PLA special operations forces and pre-positioned networks would likely focus on targeting critical nodes—air defence command elements, sensors, communications infrastructure, and key political or military leadership—rather than holding terrain. Air assault forces would aim to seize or disrupt airfields, ports, and chokepoints to enable follow-on operations. Low-signature platforms such as helicopters or gyrocopters pose a detection challenge, but remain vulnerable to short-range air defences, visual acquisition, and networked cueing from integrated sensor systems.

Years of “grey-zone” activity or military actions below the threshold of open conflict set the table. Frequent PLA Air Force air defence identification zone incursions and surrounding naval drills signal, degrade Taiwan’s readiness, and enable ongoing reconnaissance. Every Taiwan radar activation reveals location, frequency, and modes, refining future targeting and prioritization.¹⁰ Grey-zone efforts normalize tension, compress warning time, and blur the distinction between exercise and attack, complicating mobilization and defence.¹¹ These preparatory activities negatively shape the environment in which Taiwan’s air defence system must function from the very first hours of conflict.

The Layered Shield: Architecture, Integration, and Vulnerabilities

If Taiwan’s air defence system survives—even in degraded form—it becomes a key enabler of Taiwan’s Overall Defence Concept, a planning framework that emphasizes force preservation, littoral denial, and the destruction of invading forces at the beachhead.¹²Amphibious success requires local air superiority.¹³ A functioning air defence system complicates PLA Air Force air dominance, forcing higher-altitude standoff operations that dilute close air support and defensive fires for landing forces.¹⁴

Under this air denial umbrella, Taiwan’s mobile Hsiung Feng anti-ship missile batteries and fast attack craft can turn the Strait into a kill zone.¹⁵ Without functioning air defence, China’s air forces could hunt down these dispersed assets; however, with air defence in place, lightly defended support ships are exposed, creating a sustainment dilemma for Beijing. Even in a blockade scenario, the PLA Navy must still sustain persistent surveillance, air–maritime coordination, and enforcement against blockade-running; functions that become more costly and escalation-prone when operating under a surviving, even degraded, Taiwanese air defence. Air defence undermines the notion that a blockade is a low-risk coercive option.

Taiwan has spent decades building one of the world’s most integrated air defence systems, which is designed to detect, track, and engage everything from ballistic missiles to low-flying drones.¹⁶ Its philosophy is defence-in-depth with multiple supporting layers. Each layer can be ablative, reducing incoming attacks and protecting key assets, allowing forces to continue fighting.¹⁷

Included in these layers, the AN/FPS-115 Pave Paws radar at Leshan provides early warning of incoming ballistic threats from deep within the mainland.¹⁸ Pave Paws also feeds into a resilient network of fixed and mobile long-range air-search radars.¹⁹

Six E-2K Hawkeye airborne early warning and control aircraft add a flexible, high-altitude “look-down” capability against low-flying cruise missiles and stealthier threats, pushing the detection envelope far into the Taiwan Strait. When integrated with passive sensors and ground-based networks to mitigate the vulnerabilities of emitting systems, together the system forces China to allocate scarce long-range interceptors and strike assets to hunt these platforms rather than employ them elsewhere.²⁰

Taiwan fields multiple layers of medium- and high-altitude air and missile defences designed to counter aircraft, cruise missiles, and short-range ballistic missiles. While some interceptors are optimized for engaging aircraft, the backbone of Taiwan’s ballistic missile defence consists of hit-to-kill interceptors. These interceptors are intended to destroy incoming missiles by kinetic impact rather than proximity detonation. The operationalization of an enhanced upper-tier interceptor, expected in the mid-2020s, is intended to expand the defended battlespace in both range and altitude, strengthening Taiwan’s ability to absorb and attrit large missile salvos in the opening phase of a conflict.

Taiwan’s air defence posture continues to rely on a layered architecture in which a mid-tier capability focuses on engaging aircraft and cruise-missiles, while a dispersed short-range air defence network provides terminal and point defence against low-altitude threats.

Observations from recent high-intensity conflicts, particularly Ukraine’s experience with contested airspace, have reinforced Taiwan’s emphasis on prioritising the defence of critical assets under conditions of constrained air-surveillance coverage. This has encouraged a defensive posture that relies heavily on high-end interceptors to manage high-value threats. While such a layered approach improves localised situational awareness and asset protection at the tactical level, it also deepens dependence on scarce and costly upper-tier capabilities.

The combined sensors and shooters are operational and integrated within the Republic of China (Taiwan) Air Force Air Defence and Missile Command. Looking ahead, the planned T-Dome project aims to further decentralize command and control by improving sensor–shooter integration and shortening decision-making timelines.²¹Rather than relying on a single command node, the system is intended to allow multiple sensors to cue interceptors more flexibly, improving resilience against a decapitation strike.

The systems of the combined forces are integrated so that each sensor can be activated and used to direct fire and engage enemy formations, creating a highly efficient, low-latency tactical engagement core. This is expected to create a far more active tactical posture in anticipation of a decapitation strike or a tactical battle.²²

Strategic self-reliance underwrites Taiwan’s defence concept. To offset delays in U.S. arms deliveries and the risk of wartime isolation, National Chung-Shan Institute of Science and Technology (NCSIST) has prioritized domestic production. NCSIST’s new goal of producing over 1,000 missiles yearly will be very useful. Programs that are hitting their goals early, such as the TK-3, signal credible sustainment in the wartime scenario and the capacity to sustain wartime operations without immediate U.S. resupply, which enhances deterrence through depth and resilience.²³

The Asymmetric Duel: Attrition and Adaptation

Taiwan’s integrated air defence system faces two principal challenges. First is the risk of saturation, as PLA doctrine emphasizes overwhelming defences through large volumes of drones and missile salvos intended to exhaust interceptors.²⁴ The second are non-kinetic pressures. PLA investments in cyber, electronic warfare, and data manipulation are aimed at degrading, rather than disabling, command-and-control and sensor–shooter links. Experience from Ukraine suggests that such non-kinetic operations can disrupt the effectiveness of air defence and impose friction, but have generally fallen short of paralyzing integrated systems, particularly when defenders employ redundancy, mobility, and rapid adaptation.

The Ukraine war offers lessons for the PLA as it refines drone-swarm tactics to saturate Taiwan’s defences.²⁵ As in Ukraine, tactics can be used to overcome a numerical disadvantage, including mobility, dispersal, and tactical innovation. Taiwan’s adaptation includes surface-to-air missile mobility, strategic hardening, command-and-control redundancy, and distributed teams equipped with man-portable air defence systems and other weapons to efficiently counter unsophisticated threats. During a blockade or attrition contest, the ability to sustain will be decisive, elevating NCSIST’s role in the mass production of missiles, drones, and spares from a matter of industrial necessity to one necessary for survival.²⁶

Should Taiwan survive the first strike and compel the PLA to transition to an adaptive phase of the conflict marked by a slower tempo, operational improvisation, and iterative adjustment rather than pre-planned shock operations, the possibility of a swift PLA victory is eliminated.

Conclusion: The Shield of Uncertainty

An integrated air defence system is a central enabling pillar that shapes campaigns, denies quick victory, and raises costs. It integrates domestic and foreign systems under a doctrine developed from the lessons of in-depth analysis of contemporary warfare. It is more than the capability to down missiles and aircraft. It aims to withstand the initial strike, disrupt China’s rapid decision-making in a conflict, and force any hostilities into a protracted, expensive war of attrition for Beijing.

This system serves as a multilayered complication for every stage of a potential cross-strait invasion. It denies an adversarial force the critical air superiority necessary to acquire an amphibious assault, and increases the risk of a military blockade. It also supports Taiwan’s more extensive asymmetric defence posture, which relies on dispersed, mobile defence systems. Success, in this context, is measured not by the system’s absolute performance but by its robust, sustained performance under stress and by the cognitive impacts of its existence on Chinese war planners. As such, the integrated air defence has a unique impact, increasing China’s calculative risk and introducing deterrence through the potential of a protracted, destructive war that Beijing is highly unlikely to win.

Gaurav Sen is a Senior Research Fellow at the School of International Studies, Jawaharlal Nehru University, New Delhi. He is the author of The Peril of the Pacific: Military Balance and the Battle for Taiwan. His research interests include Indo-Pacific security, great-power competition, strategic autonomy, and maritime geopolitics.

References

1. Lantes, Korey F. 2024. “’Strategic Disruption’ Can Thwart an Invasion of Taiwan.” Proceedings 150, no. 12 (December 2024). U.S. Naval Institute. https://www.usni.org/magazines/proceedings/2024/december/strategic-disruption-can-thwart-invasion-taiwan

2. Cancian, Mark F., Matthew Cancian, and Eric Heginbotham. 2023. The First Battle of the Next War: Wargaming a Chinese Invasion of Taiwan. January 9. Washington, DC: Center for Strategic and International Studies (CSIS). https://www.csis.org/analysis/first-battle-next-war-wargaming-chinese-invasion-taiwan

3. Goldstein, Lyle. 2025. “Target Taiwan: Prospects for a Chinese Invasion.” Defense Priorities, September 2025. https://www.defensepriorities.org/explainers/target-taiwan-prospects-for-a-chinese-invasion/

4. Ibid.

5. Xu, Tianran. 2025. “Taiwan’s Air and Missile Defence. Part 4: Long-range SAMs versus PLA Offensive Capabilities.” ThoughtRoom – Open Nuclear Network, April 29, 2025. https://platform.opennuclear.org/thoughtroom/quick-takes/taiwans-air-and-missile-defence-part-4-long-range-sams-versus-pla-offensive-capabilities

6. The International Institute for Strategic Studies (IISS). 2024. Asia-Pacific Regional Security Assessment 2024: Key Developments and Trends. London: IISS. May 2024. https://www.iiss.org/globalassets/media-library—content–migration/files/publications—free-files/aprsa-2024/asia-pacific-regional-security-assessment-2024.pdf

7. Lin, Sean and Wu Su-wei. 2025. “Taiwan Should Seek to Leverage PLA Satnav System to Counter Drone Threat: Experts.” Focus Taiwan (Central News Agency), September 2, 2025. https://www.focustaiwan.tw/cross-strait/202509020028

8. Lin, Sean and Wu Su-wei. 2025. “Taiwan Should Seek to Leverage PLA Satnav System to Counter Drone Threat: Experts.” Focus Taiwan (Central News Agency), September 2, 2025. https://www.focustaiwan.tw/cross-strait/202509020028

9. Wuthnow, Joel. 2025. PLA Systems Attack. Keystone 25-1, January 2025. Available at https://keystone.ndu.edu/Portals/86/PLA%20Systems%20Attack-%20Keystone%2025-1%20Jan%2025.pdf

10. The International Institute for Strategic Studies (IISS). 2018. China, Global Security & Taiwan. Research Paper. London: IISS. https://www.iiss.org/research-paper/2018/09/china-global-security/

11. Goldstein, Lyle. 2025. “Target Taiwan: Prospects for a Chinese Invasion.” Defense Priorities, August 25, 2025. https://www.defensepriorities.org/explainers/target-taiwan-prospects-for-a-chinese-invasion/

12. Hsi-min, Lee (Adm., Ret.). 2021. Taiwan’s Overall Defense Concept: Theory and Practice. Hoover Institution. September 27, 2021. Available at https://www.hoover.org/sites/default/files/210927_adm_lee_hoover_remarks_draft4.pdf

13. Revels, Matthew. 2023. “Denying Command of the Air: The Future of Taiwan’s Air Defense Strategy.” Journal of Indo-Pacific Affairs 6, no. 3 (March–April): 135–44 https://www.airuniversity.af.edu/JIPA/Display/Article/3371516/denying-command-of-the-air-the-future-of-taiwans-air-defense-strategy/

14. “Taiwan’s Air and Missile Defence. Part 4: Long-range SAMs versus PLA offensive capabilities,” Open Nuclear Network, accessed Nov 3, 2025

15. Dotson, John. 2025. “Taiwan’s Defense Policies in Evolution.” Journal of Indo-Pacific Affairs 8, no. 1 (Spring 2025). April 21, 2025. https://www.airuniversity.af.edu/JIPA/Display/Article/4164821/taiwans-defense-policies-in-evolution/

16. Xu, Tianran. 2024. “Taiwan’s Air and Missile Defence. Part 1: Tien Kung-1 and Tien Kung-2.” Open Nuclear Network(Thoughtroom). 18 September 2024. https://platform.opennuclear.org/thoughtroom/quick-takes/taiwans-air-and-missile-defence-part-1-tien-kung-1-and-tien-kung-2

17. U.S. Department of Homeland Security. 2016. Recommended Practice: Improving Industrial Control System Cybersecurity with Defense-in-Depth Strategies. Washington, DC: ICS-CERT / NCCIC.

18. Wolff, Christian. 2025. “Strategic Radar Systems — AN/FPS-115 ‘PAVE PAWS’.” RadarTutorial. https://www.radartutorial.eu/19.kartei/01.oth/karte004.en.html

19. Missile Defense Advocacy Alliance. 2018. “AN/FPS-117.” May 1, 2018. https://www.missiledefenseadvocacy.org/defense-systems/an-fps-117/

20. Missile Defense Advocacy Alliance. 2018. “AN/FPS-117.” May 1, 2018. https://www.missiledefenseadvocacy.org/defense-systems/an-fps-117/

21. Author unknown. 2025. “What is Taiwan’s multi-layered T-Dome air defense system?” The Japan Times, November 30, 2025. https://www.japantimes.co.jp/news/2025/11/30/asia-pacific/taiwan-air-defense-focus/

22. Author unknown. 2025. “Taiwan President Unveils ‘T-Dome’ Air Defence System to Counter China Threat.” The Hindu, October 10, 2025. https://www.thehindu.com/news/international/taiwan-president-unveils-t-dome-air-defence-system-to-counter-china-threat/article70146730.ece

23. “Taiwan’s Missile Production Program … Two Years Ahead of Schedule,” Global Taiwan Institute, 2024, accessed Nov 3, 2025.

24. Sen, Gaurav. 2025. “How Taiwan Must Prepare to Face Chinese Drone Saturation.” The Strategist (Australian Strategic Policy Institute), July 4, 2025. https://www.aspistrategist.org.au/how-taiwan-must-prepare-to-face-chinese-drone-saturation/

25. Ditter, Timothy. 2025. PRC Concepts for UAV Swarms in Future Warfare. Arlington, VA: CNA Corporation. July 2025. https://www.cna.org/reports/2025/07/PRC-Concepts-for-UAV-Swarms-in-Future-Warfare.pdf

26. Grieco, Kelly A., and Hunter Slingbaum. 2025. “Taiwan’s Squandered Defensive Potential.” The Henry L. Stimson Center, September 11, 2025. https://www.stimson.org/2025/taiwans-squandered-defensive-potential/

Featured Image: A People’s Liberation Army Air Force J-16 escorts a H-6 bomber during a routine deterrence patrol. [Credit: Japan Air Self-Defence Force]

Force Structure

A Concept of Operations for Achieving a Navy Fleet of 500 Ships

February 20, 2026 Guest Author 4 Comments

By Captain George Galdorisi

The U. S. Navy stands at the precipice of a new era of technology advancement. In an address at a military-industry conference, the then-U.S. Chief of Naval Operations, Admiral Michael Gilday, revealed the Navy’s goal to grow to 500 ships, to include 350 crewed ships and 150 uncrewed maritime vessels. This plan has been dubbed the “hybrid fleet.” In an address at the Reagan National Defense Forum, his successor, Admiral Lisa Franchetti, cited the work of the Navy’s Unmanned Task Force, as well numerous exercises, experiments and demonstrations where uncrewed surface vessels were put in the hands of Sailors and Marines, all designed to advance the journey to achieve the Navy’s hybrid fleet.

More recently, other speeches and interviews addressing the number of uncrewed surface vessels the Navy intends to field culminated in the issuance of the Chief of Naval Operations Force Design 2045, and subsequently the Chief of Naval Operations Navigation Plan for America’s Warfighting Navy, both of which call for 350 crewed ships and 150 large uncrewed maritime vessels. These documents provide the clearest indication yet of the Navy’s plans for a future fleet populated by large numbers of uncrewed surface vessels (USVs).

The reason for this commitment to uncrewed maritime vessels is clear. During the height of the Reagan Defense Buildup in the mid-1980s, the U.S. Navy evolved a strategy to build a “600-ship Navy.” That effort resulted in a total number of Navy ships that reached 594 in 1987. That number has declined steadily during the past three-and-one-half decades, and today the Navy has less than half the number of ships than it had then. However, the rapid growth of the technologies that make uncrewed surface vessels increasingly capable and affordable has provided the Navy with a potential way to put more hulls in the water.

However, the U.S. Congress has been reluctant to authorize the Navy’s planned investment of billions of dollars in USVs until the Service can come up with a concept-of-operations (CONOPS) for using them. Congress has a point. The Navy has announced plans to procure large numbers of uncrewed systems—especially large and medium uncrewed surface vessels—but a CONOPS, in even the most basic form, has not yet emerged. Additionally, while the composition of the future Navy’s crewed vessels is relatively well understood—based on ships being built and being planned—what those uncrewed maritime vessels will look like, let alone what they will do, has yet to be fully determined.

That said, the Navy has taken several actions to define what uncrewed maritime vessels will do and thus accelerate the journey to have uncrewed platforms populate the fleet. These include publishing an Unmanned Campaign Framework, standing up an Unmanned Task Force, establishing Surface Development Squadron One in San Diego and Surface Vessel Division One in Port Hueneme, CA, and conducting a wide range of exercises, experiments and demonstrations where operators have had the opportunity to evaluate uncrewed maritime vessels.

All these initiatives will serve the Navy well in evolving a convincing CONOPS to describe how these innovative platforms can be leveraged to achieve a hybrid fleet and gain a warfighting advantage over high-end adversaries. Fleshing out how this is to be done will require that the Navy describe how these platforms will get to the operating area where they are needed, as well as what missions they will perform once they arrive there.

A key part of this evolving CONOPS will involve integrating crewed ships and uncrewed maritime vessels. This means that both will need to operate as a synergistic fighting force, not all merely steaming together to perform a mission. This will require leveraging emerging technologies that can connect these platforms in a fashion now called man-machine teaming.

U.S. Navy’s Commitment to Uncrewed Maritime Vessels

It is beyond the scope of this article to attempt to detail the reasons for the precipitous decline in the number of crewed ships. Indeed, the most recent Navy Long-Range Shipbuilding Plan details 19 ship decommissionings during this fiscal year, more than the number of ships being commissioned. Many—especially the U.S. Congress—have encouraged the Navy to increase the number of ships it fields. Add to this such factors as the increasing cost to build ships, and especially the cost to man these vessels (Seventy percent of the total ownership costs of surface ships is the cost of personnel to operate these vessels over their lifecycle), and the fact that the Navy is literally wearing these ships out more rapidly than anticipated in order to meet the increasing demands of U.S. Combatant Commanders, and it is easy to see why the Navy has difficulty growing the number of crewed surface vessels.

The rapid growth of the technologies that make uncrewed surface vessels increasingly capable and affordable has provided the Navy with a potential way to put more hulls in the water. To support these goals regarding large numbers of uncrewed maritime platforms populating the Fleet, the Navy established an Unmanned Task Force to provide stewardship for Navy-wide efforts to accelerate efforts regarding uncrewed systems. From all indications, it seems that for the U.S. Navy, the intent is to go all-in on uncrewed maritime vessels and field a hybrid force of crewed ships and uncrewed maritime systems. Importantly, the intent is to have these uncrewed systems work in conjunction with manned platforms and achieve the goal of manned-unmanned teaming.

In a presentation at a Center for Strategic and International Studies/U.S. Naval Institute forum, Vice Admiral Jimmy Pitts, deputy chief of naval operations for warfighting requirements and capabilities (N9), put the focus on uncrewed maritime systems in these terms: “We are leading the way with unmanned systems. We are leveraging the success of the Navy’s unmanned task force as well as the disruptive capabilities office. Our goal is to get unmanned surface system solutions to the Fleet within the next two years.” Admiral Pitts went on to ask the questions: “What will unmanned systems do operationally? How will they get to the war at sea and littoral operating areas? How will they stay in those areas and remain ready for conflict?”

In an article in U.S. Naval Institute Proceedings, the U.S. Indo-Pacific Commander, Admiral Samuel Paparo, put the emphasis on scaling robotic and autonomous systems in an operational context, noting:

The CNO is focusing on rapidly developing, fielding, and integrating UxSs. These systems will augment the multi-mission conventional force to increase lethality, sensing, and survivability. Project 33 [part of the Navigation Plan] will allow the Navy to operate in more areas with greater capability. Unmanned systems provide the ability to project fires and effects dynamically, at any time, from multiple axes, and with mass.

Recognizing that the United States is in an “AI arms-race” with our peer adversaries, a report by the Navy’s Science and Technology Board: The Path Forward on Unmanned Systems, advises the Navy to fully leverage AI-technologies, noting: “As they design, develop and acquire new systems, DON will want to take advantage of rapidly changing technology such as AI and autonomy.” This builds on the Navy’s desire to lower total operating costs by moving beyond the current “one UxS, multiple joysticks, multiple operators” paradigm module that exists today.

A Concept of Operations for Getting Uncrewed Surface Vessels to the Fight

The concept of operations proposed is to marry various size surface, subsurface and aerial uncrewed vehicles to perform missions that the U.S. Navy has—and will continue to have—as the Hybrid Fleet evolves. The Navy can use evolving large uncrewed surface vessels as a “truck” to move smaller USVs, UUVs and UAVs into the battle space in the increasingly contested littoral environment. The Navy has several alternatives for this platform:

The Navy’s program of record LUSV. The Navy envisions these LUSVs as being 200 feet to 300 feet in length and having full load displacements of 1,000 tons to 2,000 tons, which would make them the size of a corvette.
Unmanned Surface Vessel Division One (USVDIV-1) has stewardship for two surrogates for LUSVs, the Ranger and Nomad, as well as two MUSV prototypes, Sea Hunter and Seahawk. The Navy was sufficiently confident in the operation of its LUSV and MUSV prototypes to deploy them to a recent international Rim of the Pacific (RIMPAC) exercise.
The MARTAC T82 Leviathan, a scaled-up version of the T38 Devil Ray, is an MUSV capable of either carrying an approximately 35,000-pound payload or, alternatively, carrying smaller craft and launching them toward the objective area.

While there are a plethora of important Navy missions this integrated combination of uncrewed platforms can accomplish, this article will focus on two: intelligence surveillance and reconnaissance (ISR) and mine countermeasures (MCM). There are many large, medium, small and ultra-small uncrewed systems that can be adopted for these missions. The technical challenge remains that they must be designed to ensure that the multiple sized UxSs associated with these missions can be adapted to work together in a common mission goal.

Rather than speaking in hypotheticals as to how uncrewed vessels might be employed for these two missions, this article will offer concrete examples, using COTS uncrewed systems that have been employed in recent Navy and Marine Corps events. In each case, these systems not only demonstrated mission accomplishment, but also the hull, mechanical and electrical (HME) attributes and maturity that Congress is demanding.

While there are a wide range of medium uncrewed surface vessels (MUSVs) that can potentially meet the U.S. Navy’s needs, there are three that are furthest along in the development cycle. These MUSVs cover a range of sizes, hull types and capabilities. They are:

The Leidos Sea Hunter is the largest of the three. The Sea Hunter is a 132-foot-long trimaran (a central hull with two outriggers).
The Textron monohull Common Unmanned Surface Vessel (CUSV), now renamed MCM-USV, features a modular, open architecture design.
The Maritime Tactical Systems Inc. (MARTAC), catamaran hull uncrewed surface vessels (USV) include the Devil Ray T24 and T38 craft. The two Devil Ray USVs, along with their smaller MANTAS T12 USV, all feature a modular and open architecture design.

All of these MUSVs are viable candidates to be part of an integrated uncrewed solution CONOPS. I will use the MANTAS, Devil Ray and Leviathan craft for a number of reasons. First, they come in different sizes with the same HME attributes. Second, the Sea Hunter is simply too large to fit into the LUSVs the Navy is currently considering. Third, the MCM-USV is the MUSV of choice for the Littoral Combat Ship (LCS) Mine-Countermeasures Mission Package, and all MCM-USVs scheduled to be procured are committed to this program.

If the U.S. Navy wants to keep its multi-billion-dollar capital ships out of harm’s way, it will need to surge uncrewed maritime vessels into the contested battlespace while its crewed ships stay out of range of adversary anti-access (A2/AD) systems. This will require robust command and control systems,

Depending on the size that is ultimately procured, the LUSV can carry several T38 Devil Ray uncrewed surface vessels and deliver them, largely covertly, to a point near the intended area of operations. The T38 can then be sent independently to perform the ISR mission, or alternatively, can launch one or more T12 MANTAS USVs to perform that mission. Building on work conducted by the Navy laboratory community and sponsored by the Office of Naval Research, the T38 or T12 will have the ability to launch unmanned aerial vessels to conduct overhead ISR.

For the MCM mission, the LUSV can deliver several T38s equipped with mine-hunting and mine-clearing systems (all of which are COTS platforms tested extensively in Navy exercises). These vessels can then undertake the “dull, dirty and dangerous” work previously conducted by Sailors who had to operate in the minefield. Given the large mine inventory of peer and near-peer adversaries, this methodology may well be the only way to clear mines safely.

Operational Scenario for an Integrated Crewed-Uncrewed Mission

This scenario and CONOPS are built around an Expeditionary Strike Group (ESG) that is underway in the Western Pacific. The ESG is on routine patrol five hundred nautical miles from the nearest landfall. An incident occurs in their operating area and the strike group is requested to (1) obtain reconnaissance of a near-shore littoral area, and (2) determine if the entrance to a specific bay has been mined to prevent ingress. The littoral coastline covers two hundred nautical miles. This area must be reconnoitered within twenty-four hours without the use of air assets.

Command staff decides to dispatch the three LUSVs for the mission. Two LUSVs are each configured with four T38-ISR craft and the third LUSV is configured with four T38-MCM vessels. The single supervisory control station for the three LUSVs is manned in the mothership.

The three LUSV depart the strike group steaming together in a preset autonomous pattern for two hundred and fifty nautical miles to a waypoint that is central to the two hundred nautical mile ISR scan area, two hundred and fifty nautical miles from the shore. At this waypoint, the LUSV will stop and dispatch the smaller T38 craft and then wait at this location for their return. Steaming at a cruise speed of twenty-five knots, the waypoint is reached in about ten hours.

Two T38-ISR craft are launched from each of the two LUSVs. The autonomous mission previously downloaded specifies a waypoint location along the coast for each of the four craft. These waypoints are fifty nautical miles apart from each other, indicating that each of the four T38 craft will have an ISR mission of fifty nautical miles to cover.
Two T38-MCM craft are launched from the third LUSV. The autonomous mission previously downloaded has them transit independently along different routes to two independent waypoints just offshore of the suspected mine presence area where they will commence mine-like object detection operations.
In this manner, each of the six craft will transit independently and autonomously to their next waypoint which will be their mission execution starting point.
Transit from the LUSV launch point, depending on route, will be about two hundred and fifty to three hundred nautical miles to their near-shore waypoints. Transit will be at seventy to eighty knots to their mission start waypoint near the coast. Transit time is between four and five hours.
The plan is for each of the T38-ISR craft to complete their ISR scan in four to five hours each and for the two T38-MCM craft to jointly scan the bottom and the water column for the presence of mine-like objects in four to five hours at a scan speed of six to eight knots.

The MANTAS and Devil Ray craft transit to the objective area and conduct their ISR and MCM missions. The timeline for the entire mission is as follows:

LUSV detach strike group to T38 launch point and launch six T38: – 10-12 hours.
T38 transit from launch point to mission ISR/MCM start waypoints: – 4-5 hours.
ISR Mission and MCM mission time from start to complete: – 4-5 hours.
T38 transit from mission completion point back to LUSV for recovery: – 4-5 hours.
LUSV recover T38s and return to strike group formation – 10-12 hours.

Even with the ESG five hundred nautical miles from shore, the strike group commander has the results of the ISR and MCM scan of the shoreline littoral area within approximately twenty-four hours after the departure of the LUSVs from the strike group.

A Bright Future for Uncrewed Surface Vessels

This is not a platform-specific solution, but rather a concept. When Navy operators see a capability with different size uncrewed COTS platforms in the water successfully performing the missions presented in this article, they will likely press industry to produce even more-capable platforms to perform these tasks. This, in turn, will enable the Navy to field a capable Hybrid Fleet that will be the Navy’s Future Force.

While evolutionary in nature, this disruptive capability delivered using emerging technologies can provide the U.S. Navy with near-term solutions to vexing operational challenges, while demonstrating to a skeptical Congress that the Navy does have a concept-of-operations for the uncrewed systems it wants to procure.

Captain George Galdorisi (U.S. Navy – retired) is a career naval aviator and national security professional. During his 30-year career he had four tours in command and served as a carrier strike group chief of staff for five years. Additionally, he led the U.S. delegation for military-to-military talks with the Chinese Navy. He is the Emeritus Director of Strategic Assessments and Technical Futures at the Naval Information Warfare Center Pacific. He is the author of seventeen books, including four New York Times bestsellers. His most recent novel, Fire and Ice, was eerily prescient as it foresaw Russia’s invasion of Ukraine.

Featured Image: T38 Devil Ray USV

Feature Image Credit: Martak

Readiness

Optimizing Reactor Plant Maintenance: The Case for Shipboard SLMs

February 18, 2026 Guest Author Leave a comment

By LT P.J. Greenbaum and LT Vince Freschi

Introduction

Operational availability is the Nuclear Navy’s bread and butter, yet shipboard technicians are currently prevented from improving maintenance outcomes by an archaic data bottleneck. Scheduled maintenance to prevent system degradation or failure is planned years in advance and kept up-to-date utilizing detailed records boards which meticulously track maintenance completion. When a system fails or is degraded, corrective maintenance often requires hours of reading, symptom elaboration, and troubleshooting, before a component can be repaired. Fatigued watchstanders burn critical man-hours sifting through static technical libraries and disjointed databases to isolate casualty root causes – a complex, time consuming process akin to utilizing a card catalog to search tens of thousands of pages for a single sentence.

Consider an MMN2 (Machinist’s Mate, Nuclear) conducting a troubleshoot and repair of a noisy coolant pump. Currently, this Sailor must spend hours manually cross-referencing decades worth of material history logs with historical maintenance records, send manually-collected vibration analysis data off-ship for analysis by pricey contractors, and flip through multiple volumes of tech manuals, all to diagnose a possible problem that may or may not lie at the root cause of the issue. Generating a quality control package to repair the pump takes several hours, and generating the work authorizations and safety protocols (i.e. “tag-outs”) takes several more – all of which delays the time to repair for the pump, leaving critical propulsion plant components offline longer than needed. From the authors’ experience leading Sailors in the maintenance of nuclear systems, finding the right part alone can take hours, with many times a successful search becoming futile due to the part number turning obsolete. This results in an asset unavailable for tasking, reducing the operational force posture.

The solution to this issue is to incorporate existing technology to equip Sailors with the tools necessary to keep ships in the fight. Fortunately, the technology exists to provide locally hosted, air-gapped Small Language Models (SLMs) to enhance the nuclear propulsion plant troubleshooting process. By implementing a Retrieval-Augmented Generation (RAG) framework, Naval Reactors can transform its massive repositories of procedures, technical manuals, and material history into a live, up-to-date database of Reliable External Knowledge (REK). This approach will reduce manhours required for fault investigation, improve the accuracy of system troubleshooting, and result in reduced downtime for propulsion plant components.

The Structure: Coupling Small-Language Models with Retrieval Augmented Generation

Much attention has been garnered since the release of ChatGPT in late 2022 regarding the use cases for large language models (LLMs) – models that possess between 100 billion and 2 trillion parameters – which are able to digest large volumes of disaggregated data and generate human-like text in response.¹ While extremely powerful in their ability to synthesize and process loosely structured datasets, these machines require a high bandwidth connection to an off-ship data center – a non-starter for warships that frequently operate at EMCON in signal-denied environments. Even the Department of War’s release of GenAI.mil – a well-intentioned attempt at bringing generative AI capabilities directly to the warfighter – has limited applicability on a warship, where bandwidth and EMCON restrictions prevent its widespread use. Furthermore, LLMs can also be prone to “hallucinating” – irrelevant, incorrect, or misleading responses to queries – with potentially catastrophic consequences in a nuclear propulsion plant.²

The development and refinement of small language models (SLMs) possessing orders of magnitude fewer parameters (1 billion to 10 billion), however, allows for highly specialized, localized air-gapped models, capable of running on a single, high-functioning workstation – like a high-performance gaming laptop – which would be perfect for a rugged shipboard environment.³ These small-parameter SLMs would be most effective when coupled with a Retrieval-Augmented Generation (RAG) framework.⁴ This architecture replaces the model’s reliance on static, pre-trained memory with a dynamic system.⁵ By utilizing a local vector database to index every page of the ship’s electronic technical manuals – Reactor Plant Manuals (RPMs), Steam Plant Manuals (SPMs), machine-specific technical manuals, maintenance logs, and material history entries – the SLM no longer has to recall every answer from its training. All it has to do is find the answer within the verified, pre-uploaded documentation. RAG will enable it to return responses on the semantic meaning of queries and not the word-by-word match currently available with a “CTRL+F” search. This shift ensures that even a “small” 10 billion parameter model could deliver hull-specific, nuclear-grade technical guidance with a level of accuracy that matches or even exceeds its larger, data center-bound counterparts, all without the danger of hallucinations.⁶

The Nuclear Reliable External Knowledge Database

Key to the development of the vector database will be compiling and maintaining – with proper version control – a database of reliable external knowledge (REK) that the SLM can draw from when responding to Sailor queries. The widespread use of interactive electronic technical manuals (IETMs) like the Reactor Plant Manual and Steam Plant Manual onboard nuclear powered naval vessels simplifies the process of indexing the Nuclear Navy’s relevant data into a machine-readable format, but a machine which sees and understands procedural context – via the IETMs – without a thorough understanding of the system design bases – present in component technical manuals and procedural front-matter – runs the risk of providing inaccurate and incomplete recommendations. Therefore, in addition to the IETMs, legacy manuals for all propulsion plant components – like pumps and valves – as well as procedural guidelines – like the Joint Force Maintenance Manual (JFMM), the Radiological Controls for Ships, and the propulsion plant preventative maintenance system (PMS) – must be included in the REK database as well, allowing SLM recommendations to be pre-filtered through Nuclear Navy maintenance rules and regulations prior to arrival at the technician.

But the REK database cannot be static; it must be periodically updated to incorporate the latest feedback reports, Naval Reactors technical bulletins, and CASREPs , ensuring that the SLM maintains current hull and platform specific information. A monthly maintenance check, accomplished by uploading a “refresh” library – centrally created and made available via the Naval Reactors local area network – would allow for periodic updates in all but the strictest EMCON conditions. A critical advantage of incorporating fleet‑wide data is that equipment casualties encountered on one hull are often repeat iterations of known failure modes across other ships‑in‑class. By enabling Sailors to leverage prior diagnoses and corrective actions on sister platforms, the system reduces redundant troubleshooting, supports failure‑rate trending, and ultimately shortens equipment downtime.

Onboard the modern Ford-class engineering plant, the pre-existing automation and smart sensors open the doors to even more AI applications. Integrating the SLM with data log sets, vibrational analysis data, and material history, would allow the plant to be not only monitored through the eyes of its highly trained nuclear operators, but also through AI powered by the latest Silicon Valley advances. Coupling the SLM with Eulerian Video Magnification and installing fixed cameras on vital pumps and turbines would allow for early detection of failure on a minute-by-minute basis. The efficacy of preventative maintenance can be audited based on failure frequency and down time, empowering Sailors to continuously preserve the plant in the most efficient manner possible.

The Use Case

Consider once again the MMN2 from the introduction and the noisy reactor coolant pump. With a shipboard SLM co-pilot, the MMN2 could input the symptoms seen in a natural language prompt, accompanied by the equipment logs that preceded the failure along with the vibration data analysis for the pump. After a few seconds of “thinking,” the SLM would utilize its RAG framework to instantly pull the relevant drawings, scrape the log data for trends, and analyze the vibration data for potential failure mechanisms. But utilizing its live-updated REK library, the SLM could also flag CASREPs from other ships-in-class which noted similar failure modes. After a few more seconds, a quality assurance package could be generated in the ship-specific format, complete with all tools, parts, and materials required, required isolations for the work to be conducted, a step-by-step procedure for repair drawn from the relevant tech manual. But perhaps most importantly, each reference the SLM pulls its material from can be displayed on an adjacent window, with the MMN2 checking the SLMs recommendations each cited tech manual.

The Way Forward

The first step in deploying a shipboard RAG-equipped SLM at-scale must be accelerating the conversion of legacy technical manuals and drawings into machine-readable formats suitable for the creation of a vector database. Once all relevant technical publications, procedures, and best-practices have been compiled into a singular database (no small feat considering the Navy’s notorious compartmentalization), the data must be “chunked” into smaller sub-sections and embedded into a high-dimensional (due to computing restrictions, likely in the low hundreds of dimensions) numerical vector for retrieval by the model. Ensuring the model remains free of “data poisoning” – the injection of corrupted or incorrect data into a model’s training – will require a centralized organization, like Naval Reactors, to manage the training and development of the vector database. For optimal utility, shipboard technicians must be able to add data – like material history entries – into the vector database, although care must be taken to ensure that the model does not incorporate this potentially flawed shipboard data into its training phase. Fortunately, the recent establishment of the Propulsion Plant Local Area Network (PPLAN) technician school provides an avenue to train designated technicians on the procedures required to maintain the REK database. Designating these Sailors with a Naval Enlisted Classification (NEC) and designating those NECs as “critical” for billet-based distribution would ensure that ships have the requisite knowledge to always support their shipboard AI nodes onboard.

Next, the Navy must define the technical hardware specifications for a “Shipboard AI Node” that meets MIL-SPEC requirements for shock, vibration, and EMCON security. The Navy must prioritize maximizing video random access memory (VRAM) to maximize the processing power (the “intelligence”) of the SLM. Utilizing a laptop client already approved to handle classified information would save time and speed delivery of the system to the fleet.

Once the software and client have been paired together, the Nuclear Navy can utilize its already-existing training pipeline to begin integrating the system into maintenance and troubleshooting workflows. The Nuclear Power Training Units in Charleston, South Carolina and Ballston Spa, New York function as the perfect “proving ground” for the SLM in a fleet-like scenario. Treating the prototype rollout as a dynamic test – experimenting with model size, tokenization, chunking, and retrieval methods – would improve model performance and validate system functionality prior to fleet-wide deployment.

Lastly, and perhaps most importantly, Naval Reactors must establish a “Human-in-the-Loop” certification framework, ensuring that AI-assisted troubleshooting remains an advisory tool that works within the rigorous standards of the Nuclear Navy. Just as the SWO community still teaches its navigators paper charting at SURFNAV, despite the existence of well-proven Voyage Management Systems (VMS), the Nuclear Navy must work to ensure that any AI troubleshooting co-pilot utilized by its Sailors enhances their understanding of integrated plant operations, not replaces it. While some may argue that using such a co-pilot bypasses the deep-dive traditionally required to build system-wide expertise, a shipboard SLM acts as a learning accelerant, not as a crutch. By removing manual document searches, the tool allows technicians to spend more time synthesizing plant information and conducting high-level analysis – tasks that truly build systemic understanding – as opposed to spending hours searching through manuals for the “NIIN in a haystack.”. When integrated into a “Human-in-the-Loop” framework, this technology ensures that a Sailor’s learning is grounded in the most accurate, cited technical data available. The stakes in nuclear reactor plant operations are simply too high to outsource critical thinking. This technology must be viewed as a force multiplier that sharpens a nuclear technician’s judgment and reinforces the culture of procedural compliance that defines the program.

Conclusion

The integration of SLMs equipped with RAG architecture into the nuclear propulsion environment represents a significant upgrade to the maintenance capabilities of the individual nuclear operators. Incorporating AI into the shipboard troubleshooting and maintenance workflow is a fundamental shift designed to reduce equipment downtimes. In a future conflict characterized by denied communications and long stints at sea, a ship’s ability to remain self-sufficient – by diagnosing and repairing its primary propulsion and electrical generation capacity without racing back to stateside contractors – could mean the difference between sustained operability and taking a capital ship out of the fight. Ultimately, the goal is clear: a propulsion plant where easily accessible, machine-readable data works as hard as the Sailor. The technology exists already; the Nuclear Navy’s leadership must just deploy it. By embracing SLMs and RAG architecture, the Navy’s most valuable and complex nuclear assets will remain mission-ready, even in the most contested environments.

LT Vincenzo Freschi was born in Olbia, Italy, to an American Navy Officer and an Italian chef, where he enjoyed life in the countryside with his Border Collies until he attended Penn State University. There he earned a degree in Nuclear Engineering with a minor in Military Studies, and commissioned as a Naval Officer in 2020. Following commissioning, he served aboard the USS Stockdale (DDG-106). In 2023 he completed the Navy Nuclear Power School and Prototype training pipelines before reporting to the USS Gerald R. Ford (CVN-78), the world’s largest aircraft carrier, where he worked as the RE DIVO. Now he serves as an NROTC instructor at Carnegie Mellon University while pursuing a master’s degree in Artificial Intelligence.

LT Paul (P.J.) Greenbaum grew up in Boiling Springs Pennsylvania and attended Princeton University on an NROTC scholarship, where he studied international relations, public policy, and African studies. He commissioned and pursued a follow-on Master’s degree at Tsinghua University in Beijing, China as a Schwarzman Scholar, where he studied international relations and Chinese language. He served onboard the USS Benfold (DDG-65) homeported in Yokosuka, Japan as the Electronic Warfare Officer, and he completed his follow-on nuclear propulsion training in Charleston, South Carolina. He reported to the USS Abraham Lincoln (CVN-72) in October 2023 and served for two years as the Reactor Mechanical division officer. He currently serves as an NROTC instructor at the University of California, Berkeley while pursuing a degree in nuclear engineering.