Autonomous systems dominated the agenda at DSEI UK, but ethical issues concerns persist. A Cognizant-led roundtable with senior military and industry officials highlighted practical alternatives, including immediate gains from predictive maintenance.

 

Where should AI actually be deployed in defence? This is the question that opened the Cognizant roundtable with Belcan, the defence and aerospace engineering firm, at DSEI UK – its debut at the event.  

Titled “Can AI supercharge manufacturing and the development of kinetic platforms?”, the session brought together senior leaders from Rolls-Royce, Airbus, the Ministry of Defence's National Armaments Directorate, TechUK and other technology leaders and defence advisers.

The discussion began with a philosophical framing. The ethical questions around AI in defence have not been resolved; they remain an ongoing debate. This is challenged by autonomous weapons, which remain a vital area of exploration.

Yet, while the sector argues over AI’s role in combat, could the highest-value gains be hiding in plainer sight? In the operational, less glamorous domains that quietly define readiness and resilience, and keep critical missions moving?

Five key themes emerged from the roundtable:

 

1. Data foundations before algorithms

A technology leader made the point plainly. Data cannot be created or destroyed. AI simply transforms data to derive insight. This raises a question: does the needle you want to find actually exist in the haystack you have?

A Ministry of Defence (MoD) representative agreed. Too many focus on AI as the shiny feature on top, when in reality the logic runs backwards.

One defence supplier described welding machines that collect temperature data as they work. Years later, when a submarine weld fails, they trace it back to every other weld created under identical circumstances. That is practical AI applied to sustainment.

 

2. Knowing when automation beats AI

A different perspective came from an engineering services firm actively trying to convince clients to reframe the use of AI. Basic automation often delivers better return on investment.

Model-based systems engineering, now mandated across US Department of Defense programmes, can compress months of senior engineer effort into days of streamlined work by junior staff.

The Rolls-Royce executive described developing a small engine in 18 months instead of three to five years. The team follows two rules: be safe and don't damage the brand. Civil colleagues later adopted methods pioneered in defence.

When someone suggested defence teams could learn from commercial teams, in terms of pace of delivery, the Rolls-Royce response was swift: “actually, our defence teams showed the civil teams how to deliver faster.” Showing that some assumptions about the pace of defence development are not always well founded.

 

3. The translator gap

One startup leader predicted that software engineering as a distinct job may vanish within five years. Another participant pushed back. Organisations need people who translate between technical specialists and decision makers.

With 780 suppliers in the digital space and over 100 working on AI, the MoD faces a challenge in keeping smaller organisations competitive.

 

4. Procurement innovation

Defence procurement was not necessarily built for AI innovation cycles. One model often quoted is Unit X, the Pentagon's experimental unit that integrates Silicon Valley technologies into the military by engaging startups directly.

The UK is implementing something similar with Commercial X, with the aim of being more entrepreneurial. The MoD is establishing a digital commercial centre of excellence, with an annual GBP4.7 billion spend. Several frameworks designed to accelerate this process are being introduced with support from policy makers.

The structural challenge is real. The top third of the US defence system routinely rotates between government and industry. The UK civil service typically retains people for careers unless they become “revolvers”.

 

5. Enterprise thinking versus platform silos

A Submarine Delivery Agency delegate presented an interesting observation. Platforms are increasingly just accessories for the digital and technology inside them.

This reframes AI deployment. Instead of isolated initiatives for each platform, there is scope for cross-cutting capabilities. Supply chain visibility. Digital threads connecting design, suppliers, and sustainment. Condition-based maintenance across fleet types.

One participant was blunt. Defence needs to stop treating itself as special. Commercial manufacturing is moving quickly. Defence risks falling behind if it stays insular.

This thinking shaped what Cognizant and Belcan demonstrated at DSEI. They showed predictive maintenance for an Airbus A400 wing. The scenario: RAF Brize Norton, 06:00hrs. A humanitarian mission is pending, but inspection flags a potential fault. Can you guarantee mission readiness?

They combined Belcan's defence and aerospace engineering with Cognizant's agentic AI capabilities to answer that question. Using aircraft black box data – “the brick” that comes back without transparency – they showed how agentic AI could deliver predictive maintenance identifying faults before they become mission-critical. It demonstrated where AI delivers clear value in sustainment without ethical grey areas.

 

What needs to happen next

Defence has the capability to show leadership in AI, as it has historically. But that requires specific changes.

• Invest in data foundations before algorithms. Know when automation beats AI. Develop translation skills alongside technical expertise. Learn from procurement innovations like the US Unit X programme. Think in enterprise terms, not platform silos.
• Start where AI delivers measurable value. Predictive maintenance using black box data offers quantifiable risk reduction and mission readiness improvements without philosophical debates about machines making life-or-death decisions.
• The challenge is cultural, not technical. Whether defence organisations will choose applications that keep fleets operational, while understanding that aspects like lethality and the use of AI are still evolving.
• A proposal was suggested: a joint sandbox bringing together a prime contractor, the MoD, and AI partners to experiment with sustainment-focused applications. Not another pilot programme, but a focused effort to prove what works.
• Some primes will claim they are already doing this. But if 780 digital suppliers and over 100 AI companies are working with the MoD, and the sector is still debating when automation beats AI, something is still not connecting, perhaps.

 

 

---

Roundtable attendees:

•              Chris Arthurs, VP for Innovation, Hadean Supercomputing

•              Dave Gordon, Senior Vice President – UK, Europe and Africa, Rolls-Royce

•              Julian David, CEO, Tech UK

•              Lucy Lilley, Founder, Metis International and Strategic Advisor for Capital Park Partners

•              Paul Kahn, Executive in Residence, Renaissance Strategic Advisors, former CEO Airbus

•              Matthew Trimming, Senior Advisor to Babcock Marine

•              Marcus Lopez, President, Chief Growth Officer, Belcan, a Cognizant company

•              Victoria Cope, Commercial Director, National Armaments Director Group

•              Yatin Mahandru, VP, Head of Public Sector and Health, Cognizant UK (Moderator)