Alpha Vibes - AI’s defence dividend

Warfare’s new frontier

How AI is maximising capability 

For four years, the investment case for defence centred on greater volumes of artillery shells, interceptors and armoured vehicles, financed by budgets growing at their fastest pace since the Cold War. AI is changing how defence is being prioritised, from how much governments spend to what their spending connects. The opportunity for investors thus extends beyond autonomous weapons to infrastructure that translates information into an actionable military response.

Defence and technology equities are consequently trading on an increasingly shared set of drivers. In this edition of Alpha Vibes, we examine:

• Where AI is entering the defence chain
• How individual companies are re-engineering their products and partnerships around it
• Which business models can monetise the improvement
• How procurement and fiscal policy distribute the opportunity
• Which structural constraints could prevent operational demand from becoming shareholder returns

Key takeaways

AI is widening the defence opportunity beyond platforms and weapons to the sensing, software and communications tools that improve the effectiveness of the military kill chain. Defence exposure should be underwritten at programme and segment level, on funded demand, contractual protection, incremental margin and cash conversion, rather than on simply AI positioning.

• Content per system over platform count. AI raises the electronic and software content of installed platforms before it displaces them. That favours suppliers of sensing, communications and integration, and makes upgrade frequency a revenue driver.
• The incumbent-versus-challenger split is the wrong axis. Primes, the main defence contract holders responsible for the full delivery of weapons systems, monetise integration but absorb risk under a fixed-price default and buyback restrictions. Challengers retain more value through self-funded R&D and commercial terms, but have yet to prove repeat production and cash conversion. The useful distinction is between reusable capability and bespoke delivery.
• Digital content is not a margin guarantee. Conventional production can remain the stronger immediate earnings contributor. AI-related capabilities may improve strategic position without being its principal source of near-term margin expansion.
• Capital mobilisation is policy-contingent. A US continuing resolution, the midterm cycle, French debt near 119% of GDP and SAFE’s 35% cap on non-EU components decide when funding flows and who can capture it.

AI is changing the unit economics of military capability faster than the earnings power of the companies that supply it.
 

AI’s entrance into the defence chain

From intelligence to deployable capacity

The military kill chain follows a familiar sequence of detection, interpretation, decision, action and assessment. Mission completion has been traditionally done by kinetic weapons, those weapons which use direct physical force to maximise visible destruction. However, as the war in Ukraine has demonstrated, modern warfare is being rewritten. The methods are changing, with a rise in the use of non-kinetic applications, which use indirect, non-physical methods to effect, disrupt, or coerce an adversary. Therefore, when investors are considering the defence sector, they should be analysing these warfare methods together, without treating them as interchangeable.

AI’s role in this new warfare is that it can connect C4 (command, control, communications and computers) to improve reconnaissance and target decision-making across conventional and unmanned forces. For example, a drone can carry a reconnaissance payload, an electronic-warfare package or a weapon. Its role depends on the mission and supporting architecture, not simply its unmanned status. If an army launches a counter-drone engagement, that means that sensors identify an approaching threat; software combines observations and prioritises responses; electronic warfare attempts disruption; and an interceptor addresses threats that remain. The implication for investors is that the relevant market includes the sensing, communications and coordination around the effector, not merely the effector itself.

AI tools can increase the electronic and software content of conventional platforms before it replaces the platforms themselves. This favours an investment framework based on content per system, upgrade frequency and integration responsibility. For producers and investors it is no longer a binary choice between established contractors and technological challengers.

Four changes reshaping the defence production chain

AI’s relevance to defence becomes clearer when examined across the operating chain rather than through individual platforms. Company discussions at the 2025 and 2026 Association of the United States Army (AUSA) conferences highlighted four consistent complementary trends, comprising cheaper sensing, coordination driven by software, greater autonomy and more affordable effectors. Together, they connect information advantage with deployable capacity.

Sensing becomes cheaper and more effective. Faster processing delivers intelligence closer to the moment can shape an engagement.

• Leonardo DRS is developing uncooled infrared sensors that management targets to deliver 80% to 90% of cooled-sensor capability at approximately one-tenth of the cost, with annual volumes exceeding 100,000 units.
• HawkEye 360 aims to reduce satellite revisit times and data latency from 30 to 40 minutes to around 10 minutes through additional satellites, onboard processing and ground infrastructure.
   

Command, control and communication becomes more software driven. The value of these complementary architectural advances lies in making information usable across systems and reducing the time between threat detection and selection of an appropriate response.

• Anduril reported that integrating Lattice into an Army howitzer workflow reduced the firing process from approximately two hours to four minutes.
• Northrop Grumman’s Integrated Battle Command System connects previously separate sensors and defensive systems.
• L3Harris’ software-defined radios support resilient communications and subsequent upgrades.
   

Autonomy increases deployable capacity. The common aim is to extend operational reach and capacity without a proportional increase in personnel or reliance on newly designed platforms.

• AeroVironment’s Switchblade 600 Block 2 incorporates onboard computing for AI and machine-learning workloads; management envisaged capacity exceeding 1,000 units monthly across the Switchblade family.
• Lockheed Martin’s autonomous U-Hawk conversion provides 25% more cargo space than its crewed predecessor.
• Textron’s Ripsaw M1 is designed to complement crewed vehicles.
   

Effectors are redesigned around affordability. This combines non-kinetic disruption with physical interception, expanding response options. Cheaper systems complement high-end weapons, increasing the importance of each engagement’s cost and suitability.

• Leidos targets annual production of 3,000 low-cost containerised munitions by 2028–29 at below $500,000 per shot.
• Anduril aims to price Steelfin interceptors at roughly one-third of Patriot PAC-3 Missile Segment Enhancement’s (PAC-3) cost.
• AeroVironment’s layered counter-drone offering combines TITAN jamming, LOCUST directed energy and a low-cost interceptor still under development.
   

For investors, these changes establish the operational opportunity, not the financial outcome. The next distinction is between suppliers that enable greater capability and those that can deliver it repeatedly, profitably and without disproportionate capital requirements.

Who wins on the margins?

Prime contractors: integration has a price

The Pentagon’s revised terms of trade impose a structural ceiling on the margins of prime contractors, the companies that hold the main contract with a government and take responsibility for designing, integrating and delivering a complete weapon system, typically by coordinating a wide network of subcontractors. Fixed price contracts for Low Rate Initial Production, or LRIP, including those for the B-21 bomber, illustrate this constraint.¹ Two executive orders reinforce the pressure:

Executive Order (EO) 14402 of 30 April made fixed-price contracting the federal default. Department of War (DoW) contracts above $100 million that are not fixed-price now need written approval from the head of the agency, and each agency was told to restructure its ten largest contracts that are not fixed-price.¹ 

EO 14372 of 7 January bars ‘underperforming’ contractors from share buybacks and dividends.² A US Senate provision in the National Defense Authorization Act (NDAA) would put that restriction into law until 2031.³

The economic effect depends on technological maturity. A stable product, repeatable manufacturing process and well-defined requirement can allow a contractor to retain efficiency gains. An immature system with changing specifications can instead leave the supplier absorbing redesign, integration and production costs. 

Established contractors possess installed platforms and programme relationships through which new capabilities can be introduced. The primes are responding by becoming integrators of other companies’ innovation. The challenge is to capture AI-driven growth without giving up margin.

• Northrop Grumman is working with AeroVironment on Lumberjack, a one-way attack drone that carries ten smaller munitions made by Northrop. It is also working with Palantir on automatic targeting. Separately, it paid for the development of Raid Hunter, a gun-based air-defence system against drones and cruise missiles, itself.
• L3Harris is developing launched effects for kinetic and Electronic Warfare (EW) missions and fitting its radios into AeroVironment’s platforms. It is also working with Anduril on soldier-worn mixed-reality systems.
• General Dynamics sees Anduril and Palantir as complementary rather than competitive. It is positioning itself as the integrator that implements their systems, potentially at higher margins than in the past, because such programmes are run on more commercial terms.
• Textron supplies American Rheinmetall in the competition for the XM30 infantry fighting vehicle. It has prioritised a Modular Open Systems Architecture (MOSA) and command-and-control features. It is also bidding for Contractor-Owned, Contractor-Operated (COCO) drone services, a model the armed services increasingly favour.
   

Even shipbuilding is opening to new entrants. Anduril will put $3.7 billion of private capital into Arsenal-2, a Baltimore shipyard that will make components for Virginia-class submarines. Its Navy contract is worth up to $2.9 billion, with payments tied to production outcomes.

European contractors illustrate how exposure can differ within the same group. FactSet consensus estimates place Rheinmetall’s Digital revenue at €2.258 billion in 2026 and €3.298 billion in 2027, with the associated EBIT margin increasing from 12.3% to 12.9%.⁴ These are segment forecasts, not estimates of AI revenue. The investment relevance lies in the expansion of digital content alongside vehicles and air-defence systems; attributing that growth specifically to AI requires additional programme-level evidence.

The estimates also challenge the assumption that digital necessarily means more profitable. Consensus places Rheinmetall’s 2026 Weapons and Ammunition EBIT margin at 29.1%, compared with 12.3% for Digital.⁴ Conventional production can therefore remain the stronger immediate earnings contributor. AI-related capabilities may improve the group’s strategic position without being its principal source of near-term margin expansion.

Thales and Leonardo require the same segment-level discipline. FactSet consensus estimates place Thales’ 2026 Defence EBIT margin at 13.6% in 2026, while Leonardo’s Defence Electronics and Security business is forecast to generate revenue €10.150 billion of revenue at a 13.4% EBIT margin.⁴ Neither a strong defence franchise nor an AI strategy eliminates weakness elsewhere. Investors should connect programme exposure to the relevant segment before extending the argument to consolidated earnings.

In Europe, drones are a story of adaptation and integration rather than a replacement for conventional land systems. European discussions of battlefield adaptation emphasise the continued need for armour, artillery and infantry, supported by enhanced protection, surveillance, electronic warfare and unmanned systems.That supports primes as systems integrators, while widening the market for suppliers of sensors, effectors, secure communications and electronic warfare (EW). The European primes are also acquiring capability or partnering for it.

• Safran acquired the AI analytics firm Preligens.
• Hensoldt and Helsing are combining sensors and AI autonomy on the CA-1 Europa combat aircraft.
• Saab and Helsing are jointly developing AI-focused software for the Eurofighter.
• Rheinmetall has partnered with both Anduril and ICEYE.
• RENK is pairing its drivetrains with ARX Robotics’ autonomous ground vehicles.

Challengers: scalability must be demonstrated

The challenger model begins with a different proposition. Challengers fund development through Independent Research and Development (IRAD) and sell on commercial terms, retaining more of the value they create. The investment opportunity lies in spreading development costs across a larger installed base, while reducing the engineering needed to tailor each deployment to individual customers.

AI analytics, uncrewed systems, sensing, communications and electronic warfare are among the challengers’ ecosystem principal areas of activity. Private companies including Anduril, Helsing and Quantum Systems expand that ecosystem, but technological relevance and public-market investability remain separate questions.

These businesses can be competitors, customers or partners to listed suppliers, whose relevant exposure may lie in components, communications or integration instead of complete autonomous systems. Partnerships with private challengers should be assessed by the revenue opportunities they open, not treated as evidence that a commercially significant programme has already been secured. 

Listed challengers are turning product momentum into orders:

• Aevex booked about $175 million of orders in Q3, more than H1. It expects funding for uncrewed systems to survive the midterms, given support from both parties.
• Kratos sees its small-engine business growing several times over by the end of the decade, because engines have become a bottleneck across the missile and drone industrial base. Its hypersonic testing and target-drone businesses are also growing.
• AeroVironment’s counter-drone portfolio, including Titan, which it can deliver about a month after an order, positions it for the inner layer of Golden Dome 
   

Among the private challengers:

• Epirus’ high power microwave system, Leonidas, disables drone swarms, with electricity accounting for most of the cost per shot.
• Forterra’s revenue has more than doubled annually, and its autonomy kit is integrated into vehicles from BAE Systems, Oshkosh, Leonardo and RTX.
   

Challengers’ investment cases should be assessed as they move from development to repeat production, focusing on funded quantities, manufacturing throughput, delivery acceptance and the capital needed to expand capacity. A successful demonstration establishes technical potential, while a repeat order establishes demand. Sustained production and cash conversion establish the business model.

The corresponding test for Palantir and other software providers is whether deployments become broader, more repeatable and less costly to support. In a previous Alpha Vibes, Can the Defence sector be the new offensive line in your investment strategy? we discussed Palantir’s operational software layer and its ability to apply integrated data to customer problems.⁵ Now the appropriate financial bridge is customer expansion, implementation intensity, operating profitability and cash generation — not a retrospective share-price performance used as proof of future defence adoption.

Big Tech, but this time the compute layer is classified

Cloud and compute providers support model development, data processing, storage and secure deployment at the upstream layer. Our report, Beneath the Gold Rush places Microsoft, Amazon, Alphabet and Oracle within the broader capital-intensive AI infrastructure expansion.⁶ Defence offers a potential application for this investment, illustrated by the $9 billion Joint Warfighting Cloud Capability (JWCC) contract shared by AWS, Google, Microsoft and Oracle.⁷ 

Commercial AI scale alone, however, does not establish eligibility for a specific classified workload. AI models used for targeting and intelligence must run on networks accredited at Impact Level 6 (IL6), which covers classified information up to Secret.⁸ Above that are Top Secret environments, sometimes referred to as IL7. Each service should be assessed against its documented authorisation, security domain and deployment conditions.

AI adoption is not a consistently defined accounting category across these business models. Our proposed discipline is to move from capability to programme, programme to funded revenue and revenue to incremental cash generation. Where companies do not disclose an AI revenue measure, segment figures and programme milestones are more defensible evidence than an artificial estimate of thematic exposure.

Macro Realities

Burden sharing is not a funding guarantee

NATO’s spending commitment broadens the opportunity beyond conventional military equipment. The framework separates at least 3.5% of GDP for core defence requirements from up to 1.5% for broader security-related expenditure, including infrastructure protection, network defence, resilience, innovation and industrial capacity, by 2035.⁹ The distinction matters for investors because the second category broadens potential demand for cyber, communications and enabling infrastructure, but does not translate directly into weapons-contractor revenue.

The Hague Summit Declaration also requires annual spending plans and provides for a review of the trajectory and balance of expenditure in 2029.⁹ This introduces interim policy checkpoints, not an uninterrupted ten-year revenue commitment. The question remains if governments will translate capability targets into appropriations and procurement schedules predictable enough to support industrial investment.

Washington changed its terms of trade

US acquisition reform seeks to change how military capability is delivered, not just how much the government buys. Executive Order 14265, Modernizing Defense Acquisitions and Spurring Innovation in the Defense Industrial Base, signed on 9 April 2025, sets the broader acquisition modernisation agenda.¹⁰ Executive Order 14307, Unleashing American Drone Dominance, signed on 6 June 2025, adds an industrial policy dimension by directing agencies, where legally permitted, to prioritise US made unmanned aircraft and strengthen supply chain security.¹¹ This shifts attention to the full production chain, including airframes, sensors, communications, propulsion and control software, instead of only the final platform supplier. Domestic sourcing may improve competitive access, but can also raise qualification costs and the capital needed to replace established components.

Public Law 119-21, signed on 4 July 2025, allocates $1.4 billion to the small unmanned aircraft industrial base, $350 million to non kinetic counter unmanned aircraft programmes, $145 million to AI enabled one way attack aircraft and naval systems and $250 million to Cyber Command AI activities under Sections 20004 to 20005.¹² The allocations show a clear technological direction, with separate but complementary funding for autonomy, electronic disruption, cyber operations and physical engagement.

Playing by Europe’s rules to access its capital

The European Defence Agency (EDA) expects EU defence spending to rise from €418 billion in 2025 to €454 billion in 2026, equivalent to 2.4% of GDP, while R&D spending increases from €17 billion to €20 billion.¹³

Europe’s most direct procurement-financing instrument is Council Regulation (EU) 2025/1106 establishing Security Action For Europe (SAFE). This provides up to €150 billion in loans to member states and explicitly includes AI, electronic warfare, C4ISTAR, drones and air-defence capabilities within its scope.¹⁴

SAFE’s eligibility conditions also shape who can capture that demand. Article 16 generally requires contractors and relevant subcontractors to be established and managed within the EU, EEA EFTA states or Ukraine, subject to defined exceptions and safeguards. Components originating outside those jurisdictions cannot exceed 35% of the estimated component cost of the end product.¹⁴

For category-two products, comprising AI and EW, the regulation additionally requires contractors to control the definition, adaptation and evolution of the design without restrictions imposed by third countries, including authority to replace restricted components.¹⁴ Intellectual-property rights, component substitutability and design authority can therefore matter as much as manufacturing location. A European subsidiary or partnership does not necessarily satisfy every eligibility requirement.

Political continuity remains a timing risk

In the US, a continuing resolution (CR) has funded the government at FY 2026 levels since 1 October. It expires on 11 December. Under its terms, the DoW cannot start new programmes, raise production rates or begin certain multi-year procurements.¹⁵ As a result, more than 150 new programmes are on hold.¹⁶ Textron, for one, notes that the CR could affect the schedule of its MV-75 tiltrotor.

The FY 2027 request totals $1.5 trillion, a 44% increase. It is made up of $1.15 trillion of discretionary spending and $350 billion of mandatory funding through budget reconciliation.¹⁷ The current House reconciliation framework, however, includes only around $60 billion for defence, mostly to replenish stocks used up in the conflict with Iran.¹⁸ Industry participants at the 2026 AUSA expect only a small reconciliation or supplemental package on top of the $1.15 trillion base.

The 3 November midterms will decide who controls the FY 2027 outcome. Democratic control of either chamber would likely shrink the reconciliation component and put programmes such as Golden Dome at risk, although uncrewed systems continue to enjoy support from both parties.

Contracting is also shifting risk onto contractors. Beyond the fixed-price default, an August memo seeks more detailed cost and pricing data. Procurement is moving towards Other Transaction Authorities (OTAs) and Commercial Solutions Openings (CSOs) and away from traditional Federal Acquisition Regulation (FAR) routes.¹⁰ Together, these changes reward companies that fund their own development and penalise those that rely on cost-plus contracts.

European politics add further uncertainty. Defence spending by European NATO members rose at its fastest pace since 1953¹⁹, with the International Institute for Strategic Studies (IISS) estimating real growth of 12.6% in 2025, matching the 2024 rate.²⁰

Fiscal capacity varies more than political commitment, with France under the greatest pressure. France’s public debt reached about 119% of GDP in Q2, while this year’s deficit is projected at 5.4%.²¹ Prime Minister Sébastien Lecornu leads a minority government. His budget faces a contested passage through parliament ahead of the spring 2027 presidential election, in which the Rassemblement National (RN) is widely seen as a leading contender.²¹ Fundamentally, although initial commitments remain steadfast, the mobilisation of capital is increasingly contingent on fragile politics, especially in Paris and Washington.

Allied AI agreements provide another potential transmission channel, increasingly centred on data as well as hardware. On 24 August, UK Prime Minister Andy Burnham and President Volodymyr Zelenskyy signed an AI defence partnership. It makes the UK the first international partner to access Ukraine’s Avengers AI Labs²², a dataset of five million annotated battlefield images drawn largely from Ukraine’s DELTA battlefield management system. A first Ministry of Defence (MoD) competition, launched in September, targets drone-swarm AI that can keep working when communications and satellite navigation are degraded.²³

Transatlantic cooperation continues on two tracks.The US–UK Technology Prosperity Deal of September 2025 covers AI and quantum computing.²⁴ The advanced-capabilities pillar of AUKUS, the Australia–UK–US security partnership, covers AI and autonomy. However, investors should be mindful that cooperation becomes investable only when these arrangements generate funded work, and not when an agreement is announced on X.

Strategic Bottlenecks

Even where funding is secure, three constraints could cap sector returns

Interoperability and sovereignty. AI’s operational usefulness depends on whether information can move between systems and organisations. We have previously identified Europe’s fragmented procurement environment as a source of interoperability, training and maintenance difficulties.⁵ Adding AI to that environment does not automatically remove fragmentation. A capable model connected to incompatible data or restricted interfaces can remain commercially and operationally isolated.

This creates a tension between sovereignty and interoperability. Customers may want domestic control over sensitive data and technology while retaining access to allied capabilities. The European SAFE framework makes the industrial-policy dimension explicit through eligibility and sourcing conditions.¹⁴ Partnerships between prime contractors and start-ups are emerging as the practical compromise. For investors, sovereign positioning can protect access to some programmes while narrowing others. The opportunity should be assessed against actual procurement rules, and avoid assuming that either an exclusively domestic or universally open architecture will prevail.

Open interfaces also create a two-sided economic effect. Easier integration can expand the number of platforms a supplier serves, but easier substitution can reduce customer dependence on that supplier. The key considerations become 1. who actually controls a scarce capability 2. how easily it can be replaced without substantial operational disruption and 3. does interoperability expand its addressable market more than it weakens pricing power?

Governance is part of deployment. How autonomous targeting is governed is becoming a commercial issue. DoD Directive 3000.09 requires ‘appropriate levels of human judgment’ over the use of force.²⁵ The EU AI Act leaves military uses outside its scope.²⁶ United Nations talks on lethal autonomous weapons under the Convention on Certain Conventional Weapons (CCW) remain unresolved.²⁷

The DoW’s ‘any lawful use’ requirement has already produced a test case. In early March, the Pentagon formally designated Anthropic a supply-chain risk, after the company refused to allow its models to be used for mass domestic surveillance or fully autonomous weapons. A federal appeals court upheld the designation in September and litigation continues.²⁸ For vendors, terms of use have become a procurement risk.

For the investment case, validation and assurance are both potential costs and sources of differentiation. Our inference is that companies capable of satisfying those requirements may gain an advantage, but testing, documentation and secure maintenance can also make deployment slower and more expensive than commercial software comparisons imply.

Frontier models also raise biosecurity concerns. The US AI Action Plan of July 2025 responds with requirements for evaluation and screening.²⁹

Physical constraints remain decisive. The AI CapEx hyper-cycle remains dependent on physical infrastructure. Beneath the Gold Rush connects computing requirements with energy and mineral demand,⁶ while the AUSA 2026 discussions identify supply-chain diversification as a requirement for expanding production. Power availability constrains infrastructure deployment, while specialised components and materials constrain equipment manufacturing. Neither should be assumed to constrain every defence company equally.

On the minerals front, China’s export controls remain only partly suspended³⁰:

• The October 2025 rare-earth controls are suspended until 10 November
• The US-specific ban on gallium, germanium and antimony is suspended until 27 November
• The April 2025 licensing requirements on seven heavy rare earths are still in force
   

US Treasury Secretary Scott Bessent has said the truce runs to 10 January 2027, but China’s Ministry of Commerce (MOFCOM) has not confirmed an extension.³⁰

The exposure is direct. Teledyne cites rare earths as a bottleneck for drone gimbals, and Leonardo DRS is moving its germanium supply away from China. Suppliers such as Karman say restrictions on magnets and rare earths are adding to the strain on supply chains.

Washington’s response includes $7 billion of funding under the One Big Beautiful Bill Act (OBBBA) for critical-mineral stockpiles and supply chains.¹² It also includes the DoW’s July 2025 partnership with MP Materials, which combines³¹:

On the power front, the International Energy Agency (IEA) expects data-centre electricity demand to more than double to around 945 terawatt-hours (TWh) by 2030.³² The FY 2027 request pairs AI computing with $3.5 billion of Department of Energy (DOE) funding for firm baseload power.¹⁷

A growing order book can therefore precede a demanding investment phase. Yet, investors should examine cash conversion over several reporting periods, separating customer advances from sustainable operating improvement. A business can be strategically well positioned while temporarily earning poor returns on the capital required to meet that opportunity.

Final Considerations

The investment implication is a change in how defence exposure should be underwritten. Strategic relevance alone is insufficient. Investors must identify suppliers that control hard-to-replace capabilities, retain contractual value and scale without repeatedly rebuilding their cost base. Contrasting segment performance reinforces the point: greater digital content does not guarantee higher margins. Valuations should reflect proven earnings quality, not presumed technological leadership.

This favours a programme-level approach over a binary allocation between established contractors and challengers. An incumbent may capture attractive returns by integrating another company’s technology; a challenger may generate rapid growth while absorbing substantial development and manufacturing expenditure. The more useful distinction is between reusable capability and bespoke delivery.

The strongest counterargument is that AI improves military economics more reliably than supplier economics. Lower-cost equipment and more open architectures can broaden demand while increasing substitution and customer bargaining power. Fixed-price commitments can leave contractors carrying integration and execution risk. Adoption could consequently accelerate without supporting the margins embedded in valuations. The thesis would weaken materially if repeat orders failed to improve cash conversion or if maintaining technological relevance required development spending that consistently exceeded incremental profit.

Funding and industrial capacity introduce a second asymmetry. Expenditure can precede revenue by several years. Procurement uncertainty, sourcing restrictions and component shortages may require suppliers to commit capital before delivery schedules become dependable. A larger backlog should be evaluated alongside contractual protection, financing timelines and more importantly, the prospective return on the capacity being installed.

The next catalysts matter only if they narrow that asymmetry:

• The 11 December funding deadline and the lame-duck session after the midterms
• Q3 results, including Thales on 22 October and Rheinmetall, Hensoldt and Leonardo on 5 November (at time of writing)
• China’s mineral deadlines of 10 and 27 November
• The parliamentary approval process of France’s 2027 budget
   

To map the competitive landscape, we group the listed companies discussed in this edition into two baskets and divide them into US and European jurisdictions.

The classification is EXANTE’s own and is not a recommendation. It should not be interpreted as investment advice.

US incumbents: integrators adapting through open architectures and partnerships

Source: FactSet and EXANTE. Data as at 20 October 2026

European incumbents: platform owners partnering with start-ups for capability

Source: FactSet and EXANTE. Data as at 20 October 2026

US challengers: software, autonomy and compute on commercial terms

Source: FactSet, company data, and EXANTE. Data as at 20 October 2026

European challengers: a privately funded ecosystem scaling through prime partnerships

Source: Company Data and EXANTE. Data as at 20 October 2026

Private companies are shown for context and are not accessible through public markets.
 

References

Mayer Brown, “New Executive Order Makes Fixed-Price Contracts the Government’s Default”, May 2026.
Government Contracts Law Blog, “New Executive Order Bars Underperforming Defense Contractors from Stock Buybacks and Shareholder Dividends”, January 2026.
Holland & Knight, “Senate NDAA Provision Seeks to Restrict Defense Industry Stock Buybacks”, September 2026.
FactSet consensus estimates, segment revenue and EBIT margin for Rheinmetall, Thales and Leonardo. Data as at 20 October 2026.
EXANTE, Alpha Vibes, Can the Defence sector be the new offensive line of your investment strategy?, 21 February 2025.
Renée Friedman, EXANTE, Beneath the Gold Rush, 15 October 2026.
US Department of Defense, “Department of Defense Announces Joint Warfighting Cloud Capability Procurement”, 7 December 2022.
Defense Information Systems Agency (DISA), Cloud Computing Security Requirements Guide.
NATO, The Hague Summit Declaration, 25 June 2025.
The White House, Executive Order 14265, “Modernizing Defense Acquisitions and Spurring Innovation in the Defense Industrial Base”, 9 April 2025.
The White House, Executive Order 14307, “Unleashing American Drone Dominance”, 6 June 2025. US Congress, Public Law 119-21 (H.R.1, One Big Beautiful Bill Act), Title II, Sections 20004–20005, 4 July 2025.
European Defence Agency, “EU defence spending €418 billion in 2025, projected to €454 billion in 2026”, 16 July 2026.|
Council Regulation (EU) 2025/1106 establishing the Security Action for Europe (SAFE) instrument, Official Journal of the European Union, 2025.
Congressional Research Service, R49353, 2026.
Breaking Defense, “Can’t start, won’t start: over 150 new military programs face delay under a CR”, October 2026.
Office of Management and Budget, Budget of the US Government, Fiscal Year 2027, April 2026.
Air & Space Forces Magazine, on the CR, the December deadline and the $1.5tn request, October 2026.
SIPRI, “Global military spending rise continues”, 27 April 2026.
IISS, The Military Balance 2026, “Global defence spending”, February 2026.
Euronews, “France’s sovereign debt crisis explained: how dangerous could it be?”, 6 October 2026.
Reuters, “UK, Ukraine sign AI defence partnership linked to battlefield technology”, 24 August 2026.
Defense News, “Ukraine’s deadly AI training platform for drone tech opens to UK companies”, 25 September 2026.
UK Government, US–UK Technology Prosperity Deal, September 2025.
US Department of Defense, Directive 3000.09, Autonomy in Weapon Systems, 25 January 2023.
Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 2(3), Official Journal of the European Union, 12 July 2024.
United Nations Office for Disarmament Affairs, “Background on Lethal Autonomous Weapons Systems in the CCW”.
NPR, “Pentagon labels AI company Anthropic a supply-chain risk”, 6 March 2026; ABC News, appeals court ruling, September 2026.
The White House, America’s AI Action Plan, July 2025.
China Briefing, “China’s rare earth export controls: impacts on businesses”, 2026; Technology.org, 19 October 2026.
MP Materials, “Transformational public-private partnership with the Department of Defense”, 10 July 2025.
International Energy Agency, Energy and AI, April 2025. 

DISCLAIMER: This article is provided to you for informational purposes only and should not be regarded as an offer or solicitation of an offer to buy or sell any investments or related services that may be referenced here. Trading financial instruments involves significant risk of loss and may not be suitable for all investors. Past performance is not a reliable indicator of future performance. Forecasts are not a reliable indicator of future performance.

While every effort has been made to verify the accuracy of this information, EXT Ltd. (hereafter known as “EXANTE”) cannot accept any responsibility or liability for reliance by any person on this publication or any of the information, opinions, or conclusions contained in this publication. The findings and views expressed in this publication do not necessarily reflect the views of EXANTE. Any action taken upon the information contained in this publication is strictly at your own risk. EXANTE will not be liable for any loss or damage in connection with this publication.