AUKUS and Allied AI: Building Trilateral Defense Capabilities Across the AI Stack
Colonel Nicholas DiFiore, U.S. Army - Retired
Artificial intelligence (AI) is rapidly transforming military power by accelerating decision-making and enabling autonomy. The AUKUS partners – the United States, United Kingdom, and Australia – have each identified AI as critical to future military capability. In this context, the AUKUS trilateral security pact, announced in 2021, made emerging technologies a core pillar of cooperation. While Pillar I of AUKUS centers on helping Australia acquire nuclear-powered submarines, Pillar II focuses on co-developing advanced military capabilities: cyber, quantum, hypersonics, electronic warfare, and AI. Recognizing the critical role AI will have in future warfare, the AUKUS nations are committed to developing interoperable, secure, and mission-effective AI systems.
This paper analyzes AI collaboration across AUKUS using a layered “AI stack” framework– from foundational infrastructure and compute through data and modeling to decision-support applications and autonomous systems.[i] It argues that AUKUS is leveraging this full stack to enhance trilateral military power. After outlining national approaches, the paper examines each AI stack layer, highlights challenges, explores strategic implications, and concludes with policy recommendations.
Comparative National Approaches to Defense AI
The AUKUS nations bring diverse strengths and challenges to the co-development of defense AI. The U.S. leads in investment and operational experience. It established the Joint Artificial Intelligence Center in 2018 to accelerate AI adoption. In 2022, it elevated AI efforts under a Chief Digital and AI Office – signaling AI’s priority status for the Pentagon. Increased budget allocations for defense AI, reaching $1.8 billion in the FY2024 budget, reinforce this priority.[ii] Through programs like the Joint Warfighting Cloud Capability (JWCC), the U.S. leverages partnerships with major tech firms like Amazon, Google, Microsoft, and NVIDIA to bolster access to cutting-edge compute and cloud infrastructure. Yet integration with allies and partners remains a shortfall – recent Defense Innovation Board findings called for deeper partner integration to support networked defense capabilities.[iii]
The UK has strong academic and commercial foundations in AI, including firms like DeepMind. Still, its defense AI efforts are on a smaller scale. Recognizing AI’s importance, its 2022 Defence Artificial Intelligence Strategy set a goal to be the “world’s most effective, efficient, trusted and influential” defense organization in AI for its size. [iv] To implement this vision, it established the Defence AI Centre as a hub to champion and enable AI across the armed forces[v] and invested nearly £1 billion in AI research.[vi] However, the UK defense AI sector remains underdeveloped, facing infrastructure and workforce constraints.[vii] The UK views AUKUS as a force multiplier, offering access to larger data sets and joint experimentation opportunities.[viii]
Australia is in the early stages of defense AI adoption, guided by national data and autonomy strategies.[ix]Recent investments through the Defense Innovation Hub and Next Generation Technologies Fund have supported AI-enabled decision support and automated imagery analysis pilot projects. The Defence Science and Technology Group’s AI for Decision Making Initiative sponsors research targeting operational needs.[x] Despite this momentum, Australia’s small defense-industrial base, limited workforce, and reliance on foreign technology present challenges. Canberra views AUKUS as a critical mechanism for accessing advanced capabilities, mitigating its small industrial base, and contributing meaningfully to Indo-Pacific deterrence.[xi]
Together, the U.S. brings scale, the UK offers innovation, and Australia provides a regional perspective. Their combined efforts under AUKUS allow resource pooling, accelerate capability development, and reinforce collective deterrence.
AUKUS AI Cooperation Across the AI Stack
The AUKUS partners are aligning capabilities across the entire AI technology stack to achieve a coalition advantage in defense AI. This section examines each layer of the AI stack, highlighting national initiatives and trilateral collaboration.
Infrastructure Layer: The Digital Backbone for Coalition AI
AUKUS defense AI cooperation requires a resilient, secure, and interoperable digital backbone. Each nation invests in foundational infrastructure, such as cloud computing and networks, to ensure AI systems can be deployed and sustained at all echelons. The U.S leads with JWCC, while the UK’s MODCloud and Australia’s classified cloud investments mirror this effort. These infrastructures support global, distributed AI and are essential for real-time data sharing, joint mission planning, and the seamless deployment of AI models across allied platforms.
Infrastructure interoperability is being tested and refined through combined exercises. In 2023, the UK hosted the first AUKUS AI trial, where a secure coalition network connected U.S., UK, and Australian drones in real-time, sharing sensor data and retraining AI models in flight.[xii] In 2024, Project Convergence replicated this in a multi-domain U.S. trial.[xiii] These demonstrations showed that collective success requires shared network protocols, compatible security, and the ability to federate AI workloads across cloud environments. Ultimately, AUKUS Pillar II aims to create a shared digital backbone allowing any unit to access coalition data and AI models from any secure node.
Accelerated Computing & Device Layers: Enabling Edge AI
While infrastructure provides the foundation, it is accelerated computing and tactical devices running algorithms that bring AI to life in operational settings. To deploy AI in contested, communication-denied environments, AUKUS nations are shifting focus from centralized training to distributed inference, equipping platforms with enough computing power to run AI models at the edge.
The United States is testing and fielding AI-capable processors, like IBMs NorthPole neural inference accelerator chip, for uncrewed systems and dismounted units. The Tactical Assault Kit, for example, allows forward-deployed operators to receive AI-enhanced situational awareness on mobile devices. Other initiatives equip drones and sensors with on-device processing to support target recognition and navigation without connectivity to centralized networks.[xiv]
The UK is testing edge AI through its SAPIENT architecture, which allows sensors to detect, classify, and prioritize battlefield targets using embedded processors. This system supports human-machine teaming and is designed for interoperability with allied platforms.[xv] Australia is integrating AI mission systems into platforms such as the Ghost Bat (Loyal Wingman) and XT-8 Sky Shark UAVs. These aircraft are designed to process sensor data and execute mission logic autonomously. In AUKUS trials, Australian UAVs successfully ran allied AI models and executed ISR and strike coordination with minimal human oversight.[xvi]
A key outcome of these efforts is the emergence of modular, mission-adaptable AI kits. During trilateral exercises, allied drones demonstrated the ability to swap and run AI models across national platforms, proving both technical compatibility and operational trust.[xvii] Together, these initiatives reflect a shared commitment to field-deployable AI that is fast, secure, and coalition-ready. By embedding compute power at the edge, AUKUS partners are building a distributed architecture that allows each platform to sense, decide, and act independently while operating within a shared alliance framework.
Data Management Layer: Creating a Shared Foundation for AI
Data is the critical enabler that fuels AI performance. Each country has adopted defense data strategies emphasizing visibility, accessibility, and interoperability.[xviii] In joint trials, ISR data from UK and Australian drones was fused in real-time and processed using U.S.-trained models. Achieving this required shared ontologies, compatible metadata standards, and trust in the integrity of exchanged data. The result was faster, more accurate target identification and coordinated strike execution.[xix] To scale this capability, AUKUS is exploring federated data lakes and synthetic training data to mitigate classification barriers. [xx] This approach enables cooperative model development without compromising national control over classified datasets.
Machine Learning & Modeling/Simulation: Co-Developing the Brain of AUKUS AI
Machine learning and simulation lie at the heart of AI capability development. AUKUS partners are moving from isolated algorithm development to shared model training and experimentation. In 2023, the first co-developed AI ISR system was deployed, trained on shared data and validated in coalition trials.[xxi] This underscores a key advantage of the AUKUS framework – that shared access to real-world data and operational environments enables faster and better AI.
Combined simulation environments like the U.S. Army’s Project Convergence allow coalition partners to stress-test AI systems in realistic, multi-domain scenarios.[xxii] Such environments produce synthetic data and refine human-machine teaming concepts and joint tactics. Future efforts aim to create persistent virtual testbeds and digital twins, connecting labs in all three countries for continuous experimentation.
Decision Support & Autonomy: Toward Interoperable AI-Enabled Operations
At the top of the AI stack are applications that inform decisions and enable autonomous action. AUKUS is building the capacity for coalition forces to share AI insights and act on them as a unified team. For example, in Project Convergence 2024, a UK drone identified a target using an AI model trained on UK data, which was verified by a human controller and passed to an Australian UAV to strike.[xxiii] This scenario demonstrated the operational promise of human-AI teaming and coalition interoperability.
AUKUS also integrates AI-enabled applications into command-and-control systems and mission planning tools, such as algorithms that suggest optimal asset allocations or forecast adversary behavior.[xxiv] Partners are also fielding autonomous systems, from drone swarms to unmanned underwater vehicles. Trials like the AUKUS Undersea Robotics Autonomous Systems (AURAS) project aim to develop large, autonomous submarines for persistent surveillance and strike.[xxv]
Individual and collectively, the AUKUS nations are operationalizing AI cooperation across every layer of the technology stack. These efforts are introducing AI-enabled capabilities that are interoperable, edge-capable, and mission-effective into the AUKUS nations’ armed forces, enhancing overall coalition warfighting.
Challenges to Integration Across the Stack
While the AUKUS AI collaboration holds great promise, significant challenges and barriers exist to integrate fully the three nations’ efforts across all layers of the AI stack. These challenges span policy, technology, and institutional culture. Overcoming them will be crucial to realizing the vision of a seamlessly interoperable AUKUS AI ecosystem.
Security Classification and Information Sharing. Despite high levels of trust and robust intelligence sharing from the Five Eyes partnership, differences in classification systems, NOFORN restrictions, and export control laws hinder AI model and data sharing. Recent efforts, like U.S. International Traffic in Arms Regulations (ITAR) exemptions for AUKUS Pillar II technologies, are positive steps, but legal and bureaucratic constraints persist.[xxvi] Without harmonized security protocols or federated clearance processes, collaborative AI development will be confined to unclassified or synthetic domains without harmonized security protocols or federated clearance processes.
Export Controls and Intellectual Property. AI systems often blend commercial and military components, subjecting them to complex export regimes. The lack of fast-track licensing between AUKUS nations slows collaboration. Intellectual property (IP) concerns further complicate matters—defense contractors and startups need clarity on how jointly developed AI tools will be governed, licensed, or commercialized. Without clear IP governance frameworks for jointly developed AI, firms may be reluctant to engage in cross-border R&D.
Interoperability and Technical Standards. Divergent data formats, APIs, and computing platforms can prevent seamless integration. For example, an AI model trained in one country may not function with another’s hardware or data. Shared technical standards for model formats, data labeling, and testing are essential for plug-and-play interoperability.
Cultural and Organizational Differences. Each nation’s defense bureaucracy has different risk tolerance levels, procurement flexibility, and tech adoption timelines. Coordinating programs means aligning budgeting cycles and contract vehicles. Organizational differences also extend to the industrial base, where corporate cultures of large U.S. primes and smaller UK and Australian firms may clash. These differences can slow joint development and deployment of AI-enabled capabilities. Effective collaboration will require incentives and trust-building across government and industry.
Resource and Workforce Imbalances. Disparities exist in resources and workforce capacity across AUKUS. The U.S. leads in AI funding, research infrastructure, and talent availability, while the UK and Australia face more acute shortages of defense AI specialists. This imbalance risks placing a disproportionate innovation burden on the U.S. and limiting the ability of partners to engage in trilateral initiatives fully. Moreover, sustaining trilateral programs will require consistent resourcing—political shifts or budget constraints in any partner country could threaten long-term initiatives.
In light of these challenges, it is evident that success in AUKUS AI cooperation is not guaranteed or easy. However, recognizing and addressing these issues is part of the process. Steps like revising export controls, emphasizing interoperability in exercises, and planning joint education initiatives are already positive signs.
Strategic Implications for the AUKUS Alliance
Effective AI integration under AUKUS Pillar II could significantly reshape coalition capabilities and regional security in four key ways:
Strengthening Deterrence and Combined Military Power. AI integration among the AUKUS partners will bolster their collective military capability and thus their deterrence posture in the Indo-Pacific. AI-enhanced capabilities across surveillance, targeting, and autonomy will raise the cost of aggression for an adversary by enabling faster and more accurate detection, coordination, and response. Shared AI-enabled systems, like autonomous surveillance drones or undersea sensors, will increase coalition presence in critical areas and improve situational awareness, reducing gaps that an adversary might exploit and enhancing conventional deterrence.
Operational Interoperability by Design. By building interoperable AI systems from inception, trained on common data, and running on shared infrastructure, AUKUS enables plug-and-play force integration. This reduces redundancy and ensures that coalition task forces can operate as a unified whole in conflict.
A Scalable Model for Innovation and Partnership. Pillar II offers a blueprint for alliance-based innovation—leveraging U.S. scale, UK creativity, and Australia’s Indo-Pacific focus to co-develop defense technologies. This architecture is also designed to expand: partners like Japan and Canada have expressed interest in collaborating on AI and emerging tech. AUKUS can serve as a modular platform where trusted nations integrate selectively, building a broader coalition of technologically aligned democracies.
Shaping Norms for Responsible Defense AI. AUKUS’s emphasis on ethical AI, human-in-the-loop design, and legal transparency positions it to lead on setting international norms. Trilateral standards for AI certification and operational use could influence broader coalitions like NATO or Five Eyes, countering authoritarian approaches to AI and autonomy.
In sum, AUKUS AI integration signals a dual shift in defense strategy: first, toward harnessing the full AI technology stack to augment traditional military platforms; and second, toward a collaborative model for developing advanced defense technologies. If fully realized, this approach could transform AUKUS into a digitally integrated coalition capable of countering China’s informationized warfare—and set the benchmark for collaborative development of next-generation military capabilities.
Policy Recommendations
To institutionalize progress under Pillar II and ensure sustainable trilateral AI integration, the following actions are recommended:
Expand Trilateral Testbeds and Simulation Environments. Transform joint trials like Project Convergence into permanent, networked test environments linking national labs. These should simulate contested conditions—cyber attacks, GPS denial, electronic warfare—to evaluate AI performance, refine interoperability, and inform acquisition decisions.
Harmonize Technical Standards and Procurement Requirements. Establish trilateral working groups to define shared standards for data formats, model interfaces, ethical review criteria, and edge hardware. Embed AUKUS interoperability requirements into each nation’s procurement processes, with mutual certification pathways and reciprocal market access for partner firms.
Streamline Information Sharing and Reform Export Controls. Accelerate reforms to ITAR and classification policies to enable license-free sharing of approved technologies within AUKUS. Build a secure enclave for AI collaboration based on Five Eyes precedents. Use tiered data access models and synthetic datasets to facilitate cooperation while preserving national control.
Conclusion
AUKUS Pillar II presents a rare opportunity to fuse advanced defense technologies with a deepening alliance structure. By integrating AI across the technology stack—from infrastructure to autonomy—the U.S., UK, and Australia are laying the foundation for a digitally enabled coalition capable of meeting 21st-century security challenges. This effort strengthens collective deterrence, enhances operational integration, and offers a scalable model for allied innovation. Yet success is not guaranteed. It will require harmonized policies, shared standards, and sustained investment in people, infrastructure, and trust. If these conditions are met, AUKUS could not only counter emerging threats in the Indo-Pacific but also shape global norms for the responsible development of military AI, setting a precedent for how democracies can lead in the next era of warfare.
About the Author
Nicholas DiFiore has served as a Department of Defense Civilian on the Joint Staff since 2019. He is currently a Strategic Planner in the J5 Directorate, where he integrates ally and partner nations into U.S. Joint Force strategy and operations. A U.S. Army veteran, he served 24 years as a military intelligence and foreign area officer, with assignments in Germany, Bosnia, Iraq, India, and Washington, DC. His career included roles as an intelligence officer, instructor, Army Attaché in New Delhi, and strategic planning on the Joint Staff. He is a 2025 graduate of the Eisenhower School for National Security and Resource Strategy.
[i] Shane Shaneman, “The AI Stack: A Blueprint for Developing and Deploying AI” (National Defense University, February 1, 2024).
[ii] Brandi Vincent, “Why the Pentagon Didn’t Request Higher Funding for AI in Fiscal 2025,” DefenseScoop (blog), March 11, 2024, https://defensescoop.com/2024/03/11/pentagon-ai-budget-request-2025/.
[iii] “Emerging Technology Horizons: Accelerate AUKUS Pillar 2 to Lead in Emerging Technologies,” National Defense Magazine, accessed April 16, 2025, https://www.nationaldefensemagazine.org/articles/2025/2/28/emerging-technology-horizons-accelerate-aukus-pillar-2-to-lead-in-emerging-technologies.
[iv] “Defence Artificial Intelligence Strategy” (UK Ministry of Defence, June 2022), https://assets.publishing.service.gov.uk/media/62a7543ee90e070396c9f7d2/Defence_Artificial_Intelligence_Strategy.pdf.
[v] “Defence Artificial Intelligence Centre,” GOV.UK, accessed April 16, 2025, https://www.gov.uk/government/groups/defence-artificial-intelligence-centre.
[vi] “Government Commits up to £3.5 Billion to Future of Tech and Science,” GOV.UK, accessed April 16, 2025, https://www.gov.uk/government/news/government-commits-up-to-35-billion-to-future-of-tech-and-science.
[vii] “Developing AI Capacity and Expertise in UK Defence,” UK Parliament, accessed April 16, 2025, https://publications.parliament.uk/pa/cm5901/cmselect/cmdfence/590/report.html.
[viii] “Developing AI Capacity and Expertise in UK Defence.”
[ix] Peter Layton, “Evolution Not Revolution: Defence AI in Australia,” in The Very Long Game: 25 Case Studies on the Global State of Defense AI, ed. Heiko Borchert, Torben Schütz, and Joseph Verbovszky (Cham: Springer Nature Switzerland, 2024), 581–603, https://doi.org/10.1007/978-3-031-58649-1_26.
[x] “Artificial Intelligence for Decision Making Initiative 2022,” Defence Science and Technology, March 15, 2022, https://www.dst.defence.gov.au/opportunity/artificial-intelligence-decision-making-initiative-2022.
[xi] Justin Hendry, “Why Australia Shouldn’t Pursue Sovereign AI,” InnovationAus.Com (blog), July 5, 2024, https://www.innovationaus.com/why-australia-shouldnt-pursue-sovereign-ai/.
[xii] “AUKUS Partners Demonstrate Advanced Capabilities AI Trial,” Department of Defence, May 26, 2023, https://www.defence.gov.au/media-room/media-releases/aukus-partners-demonstrate-advanced-capabilities-ai-trial.
[xiii] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations,” U.S. Department of Defense, August 9, 2024, https://www.defense.gov/News/News-Stories/Article/Article/3484698/aukus-pillar-ii-milestones/.
[xiv] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations.”
[xv] “Artificial Intelligence for Decision Making Initiative 2022.”
[xvi] “AUKUS Partners Demonstrate Advanced Capabilities AI Trial.”
[xvii] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations”; Tim Martin, “The AI Side of AUKUS: UK Reveals Ground-Breaking, Allied Tech Demo,” Breaking Defense, May 25, 2023, https://breakingdefense.com/2023/05/the-ai-side-of-aukus-uk-reveals-ground-breaking-allied-tech-demo/.
[xviii]“DOD Issues New Data Strategy,” U.S. Department of Defense, accessed April 17, 2025, https://www.defense.gov/News/Releases/Release/Article/2376629/dod-issues-new-data-strategy/https%3A%2F%2Fwww.defense.gov%2FNews%2FReleases%2FRelease%2FArticle%2F2376629%2Fdod-issues-new-data-strategy%2F; Layton, “Evolution Not Revolution.” “DOD Issues New Data Strategy,” U.S. Department of Defense, accessed April 17, 2025, https://www.defense.gov/News/Releases/Release/Article/2376629/dod-issues-new-data-strategy/https%3A%2F%2Fwww.defense.gov%2FNews%2FReleases%2FRelease%2FArticle%2F2376629%2Fdod-issues-new-data-strategy%2F; Layton, “Evolution Not Revolution.”
[xix] “AUKUS Partners Demonstrate Advanced Capabilities AI Trial”; “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations.”
[xx] Rojoef Manuel, “UK, Defense Partners Trial AI Recognition to Increase Target Information,” The Defense Post (blog), December 6, 2022, https://thedefensepost.com/2022/12/06/uk-defense-ai-recognition/.
[xxi] “AUKUS Partners Demonstrate Advanced Capabilities AI Trial.”
[xxii] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations.”
[xxiii] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations.”
[xxiv] “AUKUS Pillar II Milestones Hint at Future Integrated Autonomous, Artificial Intelligence Operations”; “AUKUS Pillar 2: Advanced Capabilities,” UK Parliament, September 2, 2024, https://commonslibrary.parliament.uk/research-briefings/cbp-9747/.
[xxv] “AUKUS Pillar Two: Advancing the Capabilities,” Center for Strategic and International Studies, July 24, 2023, https://www.csis.org/analysis/aukus-pillar-two-advancing-capabilities; “AUKUS Pillar 2: Advanced Capabilities.”
[xxvi] “Emerging Technology Horizons: Accelerate AUKUS Pillar 2 to Lead in Emerging Technologies.”