Emerging technology creates value only when an organisation can absorb it. As NATO compresses adoption timelines and ties innovation work to routes into procurement, the most useful line in a 2027 technology budget may pay for evidence, interfaces and people before it pays for deployment.
Andy Demir · 14 min read
Executive Summary
In April 2026, NATO's innovation accelerator DIANA placed its first research contract on behalf of an Ally, buying an engineering study and a simulation suite from a British undersea robotics company for Canada's defence research agency. Under the service that made the award, a successfully demonstrated prototype can move into production without a second competition. Read alongside NATO's 24-month adoption objective, the July 2026 Innovation Scale-Up Package and the September 2026 Quantum Technology Roadmap, the signal for aerospace and defence suppliers is that competition is moving earlier, to the stage where evidence, integration knowledge and test results are produced. The commercial opportunity is to fund that preparatory work as a distinct, bounded investment, separate from deployment, with an owner, an exit condition and a named route to a customer. My expectation, held with moderate confidence, is that by the end of the third quarter of 2027, more Allied innovation awards will be explicitly tied to such routes, with NATO's first quantum pilot report and assessment criteria in the first half of the year serving as checkpoints. The decision worth taking now is small: select two or three applications linked to real programme or customer priorities and commission a 90-day readiness review for each.
In This Essay
- A research contract with a production clause
- Three clocks behind one investment decision
- Readiness has an economic value
- Different applications deserve different commitments
- Where the friction sits inside the company
- The strongest case for waiting
- What the next twelve months will show
- A first decision for the 2027 plan
- Questions for the C-Suite and the Board
- References
A research contract with a production clause
In April 2026, NATO's Defence Innovation Accelerator for the North Atlantic, DIANA, awarded a research and development contract to HonuWorx, a British undersea robotics company, on behalf of Defence Research and Development Canada. The work is an engineering study to extend the operating depth of the company's undersea systems, together with a high-fidelity simulation suite intended to reduce technical risk on the way to a deployable system. It was the first contract DIANA had placed for an Ally under its Rapid Adoption Service, and the company reached it through DIANA's 2025 challenge programme, during which it worked with end users at the accelerator site in Nova Scotia (North Atlantic Treaty Organisation [NATO], 2026a).
The detail that matters commercially sits in the description of the service itself. DIANA can award research and prototype contracts on an Ally's behalf through an opt-in arrangement, and prototypes that are successfully demonstrated can then move into production without further competition (NATO, 2026a). The customer has paid to reduce specific uncertainties, the supplier has been paid to produce evidence, and the competitive moment has advanced to the point where that evidence is now available. A research contract of this kind is still a long way from a production order, and nothing in the announcement indicates that this system will reach that stage. What it shows is a pathway in which preparation is itself the contract.
For a chief technology officer building a 2027 agenda, I think this changes the question. The familiar exercise is to decide which technologies deserve attention. The more useful one is to decide where paying for preparation now creates a valuable option to deploy later, and where it would only absorb scarce engineering capacity. Readiness, meaning the evidence, interfaces, test data, and trained people that enable a capability to be accepted and used, is becoming something a company can budget for, test, and sell. Treated that way, an interesting technology list becomes an investment discussion that a CFO and a board can engage with.
Three clocks behind one investment decision
Every adoption decision runs on at least three clocks. The technical clock measures when a product can do the required task. The organisational clock measures when people, interfaces, data, security arrangements and operating procedures can support it. The commercial clock measures when a customer, a contract and a funding line allow it to earn money. The three interact, though they rarely move together, and many disappointments in technology adoption come from planning against one clock while the others run slower.
NATO's Rapid Adoption Action Plan, endorsed by Allied leaders in June 2025, makes the timing pressure explicit. It sets a general objective of moving from an identified need to the acquisition and integration of a new technological product within 24 months, with market research typically limited to 3 months and incremental testing, evaluation, verification, validation, and integration generally completed within 12 months of identifying potential solutions. The plan notes that these timelines may vary with technology readiness level, and implementation remains a national responsibility (NATO, 2025). They are objectives, and the useful way to read them is as the direction customers have committed to travel.
The implication for suppliers is conditional but material. If customers compress the purchasing schedule, the engineering work does not compress with it. A supplier that begins assembling qualification evidence, interface documentation or test data only after a requirement is published may find that a 12-month testing window leaves little room for work that has always taken longer. Some preparation, therefore, becomes more valuable before a tender, while other work still depends on requirements that only the customer can settle.
NATO's Quantum Technology Roadmap, published on 29 September 2026, shows how uneven the clocks can be within a single technology family. It treats quantum computing, communication and sensing separately because each has different readiness levels, standards, test centres and use cases. Its dated activities include an initial version of NATO's Quantum Assessment Criteria by the second quarter of 2027, a first pilot report on quantum technologies in NATO-relevant scenarios expected in the first quarter of 2027, and an industry advisory group study on readiness for quantum-resistant cryptography due in the second quarter (NATO, 2026c). These are planned milestones, and plans like these can slip.
That is a far better basis for planning than a single date for "quantum readiness". One application may justify participation in testing while another warrants monitoring, and a cryptographic migration can justify preparatory work long before any purchase of quantum computing makes sense. My reading is that the CTO's job increasingly involves translating those differences into distinct commitments, rather than a single judgement about an entire field.
Readiness has an economic value
Preparation is worth paying for when it changes the economics of a later decision. It can reduce engineering rework, expose an unsuitable supplier early, strengthen a bid or shorten the interval between customer acceptance and usable delivery. None of these benefits is automatic. Each depends on the preparation addressing a constraint that genuinely affects a programme or a customer, and on someone keeping the result in a form the business can use again.
This suggests two separate investment cases. The deployment case asks whether the operational benefit justifies the cost of acquisition, integration and support. The learning case asks whether a smaller commitment can resolve an uncertainty that materially affects the deployment decision. A successful learning investment can lead to procurement, modification, or a decision to stop, and its return is improved judgement, provided the evidence is retained.
NATO's Innovation Scale-Up Package, agreed in July 2026, gives the learning case a clearer destination. Allies state that their contracting for new technology gives non-traditional suppliers the confidence to scale and private capital the incentive to invest. They commit to linking innovation challenges and operational experimentation to pre-identified contracting pathways, including through systems integrators, so that tested solutions can be procured quickly where they meet essential requirements. At the Ankara summit, NATO also published its first public innovation demand signal, naming five areas: the combat effectiveness of large land formations, air and missile defence, deep strike, rapid medical treatment and evacuation, and adaptable logistics (NATO, 2026b). These are statements of intent whose effects will be reflected in contracts over the coming year. They nevertheless give a finance function something concrete to test a learning budget against: does this experiment sit inside a published demand area, and is there a named route from it to a buyer?
The integration problem beneath these intentions is large. McKinsey estimates that upgrading the more than 700,000 computing nodes in the US military's installed base to host the AI capabilities governments are already buying would cost between $160 billion and $230 billion, and argues that underinvestment in the layers connecting algorithms to platforms is constraining software at the tactical edge (Rodriguez et al., 2026). This is a consultancy estimate built on the firm's own analysis, and it covers the United States alone. Its relevance to a European supplier is directional: much of the value in the next few years may sit in making existing platforms able to accept new capability, which is readiness work by another name.
One question I would add to any investment paper is how much of the expenditure stays useful if the preferred technology disappoints. An evaluation dataset, an interface specification or a test environment can support an alternative supplier. A proprietary configuration that serves only one product keeps much less value. The comparison that matters is between the options preserved, the uncertainty resolved and the obligations created.
Different applications deserve different commitments
AI offers an immediate test of this reasoning. An assistant that retrieves engineering documents and a machine-learning function embedded in an aircraft can share underlying techniques while carrying very different consequences, permission boundaries and validation needs. Experience gained in a bounded internal workflow is valuable, and it is no evidence that an airborne function is ready for approval.
The regulatory work makes the distinction concrete. In June 2026, the European Union Aviation Safety Agency released Proposed Issue 03 of its AI Concept Paper for a ten-week consultation that closed on 12 August. The issue extends guidance to further techniques, including reinforcement learning and symbolic AI, and begins to address what the agency calls Level 3 AI, advanced automation in which the human end user may be remote or absent. EASA describes it as the final concept-paper deliverable under its AI Roadmap 2.0 and as part of a human-centric trustworthiness framework (European Union Aviation Safety Agency [EASA], 2026). It is guidance under consultation, which is why the useful unit of decision is the application in its operating environment. In AI-assisted engineering, a pilot might test whether retrieval still saves time once the cost of checking is factored in. For edge autonomy, preparation might concentrate on behaviour under degraded communications, uncertain inputs or constrained computing.
Post-quantum cryptography calls for another decision. It relies on algorithms designed to resist quantum attacks and requires no quantum computer to deploy. The US National Institute of Standards and Technology released its first three post-quantum standards in August 2024, and its migration project separates two pieces of work: discovering where and how cryptography is used across an organisation's systems, and testing the interoperability of the new algorithms in common protocols (National Cybersecurity Centre of Excellence [NCCoE], n.d.). For a long-lived aerospace or defence system, the first commitment is often an inventory of where cryptography sits, what suppliers plan for their own products and how easily algorithms can be changed. That work pays off well before anyone knows when a cryptographically relevant quantum computer will exist, and NATO's roadmap lists the adoption of post-quantum standards as ongoing work for the Alliance (NATO, 2026c).
Quantum sensing sits elsewhere on the clocks. In May 2024 the UK government reported flight trials at MoD Boscombe Down in which Infleqtion, working with BAE Systems and QinetiQ, flew an optical atomic clock and core elements of a cold-atom inertial sensor aboard QinetiQ's RJ100 demonstrator aircraft, with a national aim of deploying quantum navigation on aircraft by 2030 (Department for Science, Innovation and Technology & UK Research and Innovation, 2024). A demonstration more than two years old supports examining integration and operating conditions, and says little yet about the commercial suitability of particular systems. NATO's maritime quantum-sensing sea trials, planned for 2026 and 2027, will provide evidence of performance in environments where satellite navigation is denied (NATO, 2026c).
For portfolio discussion, I would use five commitments: scale an application that has proved itself, pilot a bounded uncertainty, build a prerequisite capability, watch a defined external milestone, or decline the proposition for now. The same technology can sit in several of these across one business, because requirements, existing capability and customer access differ by programme. The architecture beneath applications deserves equal attention, since an interface or test environment can serve several applications at once, and a single application can expose the need for a better interface. The aim is the minimum shared foundation that makes a credible use case possible, improved through evidence.
Where the friction sits inside the company
Readiness is also an organisational matter, and the friction is usually ordinary. Finance needs a return estimate, engineering needs a stable requirement, security needs a controlled interface and procurement needs a credible supplier. Each position protects something legitimate. A new capability that arrives without answers to all four tends to stall in the space between functions, where no one owns the next step.
A bounded experiment can help precisely there, if those functions shape it from the start. The measure worth agreeing is the complete accepted task, including review and correction. A tool that produces a faster first draft of an engineering analysis creates value only if the time saved exceeds the time it adds for checking, and only if it is clear who may approve the result. Training is then built around the real task. "AI skills" can describe anything from writing prompts to integrating machine-learning models into aircraft systems, and a training budget is more useful once it says which.
There is a quieter point about the middle of the organisation. Strategy set at the top and enthusiasm at the working level often meet a layer of programme and functional managers whose targets were set before the new capability existed and who carry the delivery risk if it disappoints. Their caution is rational. Giving them ownership of the evidence and a measure that rewards a well-reasoned decision to stop as much as a decision to proceed tends to drive adoption more than another demonstration.
The strongest case for waiting
The strongest objection to early preparation is scarcity. Engineering teams already carry delivery commitments, and an exploratory project incurs costs even when its equipment bill is small, because specialist attention cannot simply be scaled up. A company can also learn too early: standards change, suppliers disappear, and interfaces chosen for a prototype can create obligations that are expensive to unwind.
That objection argues for selectivity. Early learning is most attractive where an uncertainty matters economically, a prerequisite takes time to build, and a bounded activity can produce a useful answer. Waiting is reasonable where a proposition has no credible use case, an external milestone will settle the question more cheaply, or the activity would displace work of greater value. A decision to watch can be a disciplined allocation choice.
The distinction between reusable and perishable learning helps here. Understanding requirements, data quality, or a class of operational constraints tends to remain valuable across vendors, while detailed work tied to a single immature product can quickly become outdated. A review can weigh both, and ask whether working with a customer, a research institute or a test network offers a cheaper route to the same knowledge.
Every experiment therefore benefits from an agreed end point. It can produce enough evidence for deployment, reveal the need for a different approach or show that further work is unjustified. Continuing because a technology remains strategically interesting is a separate decision that needs its own reason. That discipline keeps room for exploration while giving every experiment a destination.
What the next twelve months will show
My reading is that by the end of September 2027, more Allied innovation awards will be tied explicitly to a named route into procurement, and that funding for the preparatory work around them, meaning test evidence, integration and interoperability, will be more visible in Allied programmes than it is today. I hold that view with moderate confidence: more likely than not, though well short of certain, because it rests on published commitments whose delivery depends on national budgets and procurement offices.
The signals I would watch are specific. Further Rapid Adoption Service contracts placed on behalf of Allies beyond the first Canadian award would support it. So would innovation challenges that name their contracting pathway at launch, as the Scale-Up Package proposes, and NATO's own quantum milestones arriving close to schedule: the first pilot report in the first quarter of 2027, then the assessment criteria and the industry readiness study on quantum-resistant cryptography in the second quarter (NATO, 2026b, 2026c). The view would be wrong if, by the third quarter of 2027, the first DIANA contract remains a showcase without successors, if challenges continue to end in demonstrations without follow-on commitments, or if deployments routinely succeed with much less preparation than this argument expects.
If the view holds, the suppliers that gain are those already holding evidence packs, test data and documented interfaces for applications inside the published demand areas, together with systems integrators able to carry smaller firms' products through to acceptance. Companies with promising demonstrators and no route to evidence are likely to find the window narrower than their technology deserves.
A first decision for the 2027 plan
The company I have in mind is a mid-sized European defence or dual-use supplier with an established product business and two or three emerging-technology lines that its management believes in. Were I part of its 2027 planning discussion, the option I would put forward is small enough to fit on one line in the minutes: select two or three applications tied to actual customer or programme priorities, and commission a 90-day readiness review for each, with a named owner and an agreed condition for stopping. It would replace a single emerging-technology budget line with separate learning and deployment cases, so that finance can judge each on its own terms.
The first month would establish a baseline with engineering, programmes, commercial, operations, security and finance: what prevents dependable use today, which uncertainty most affects the next commitment, and whether a published demand area and a named route to a buyer exist. A director or functional leader can run this within one programme, and the output is a short decision record with a proposed commitment, an owner and the evidence that would change it. The second month tests the prerequisite with the greatest consequence. If cost is the uncertainty, measure cost; if compatibility is the uncertainty, test the interface; if customer acceptance is the uncertainty, clarify requirements with the customer. By the third month, the evidence can be set against the original investment logic: what is now known, what spending remains, what can be reused and what obligation the next commitment would create. Within six months, the reviews that pass would move into bid preparation or customer-funded work, and within twelve the company would know which of its emerging lines has a credible path to revenue.
The economic improvement is real, but cannot be accurately assessed from outside the company. It would show in a small set of indicators: the time needed to produce evidence a customer accepts, review and rework effort, win rates on innovation challenges and rapid-adoption opportunities, and the conversion of demonstrators into funded follow-on work. A positive review earns further funding, and an adverse one releases engineers and money before a larger commitment. Both improve allocation.
The effect of delay is mostly about timing. Where the competitive moment moves to the prototype stage, and a demonstrated prototype can pass into production without another competition, a supplier that starts preparing when the tender appears may be starting after the decision that matters. There is also a reason to act this quarter: 2027 budgets are being set now, and NATO's first-half milestones will shape what Allied customers ask for during the year.
The trade-off is engineering attention. If the company's revenue rests mainly on long-cycle platform programmes whose national customers have not opened rapid-adoption routes, the same specialists may create more value protecting delivery, and I would then advise a lighter watching posture. Where those routes are open, the April contract for undersea robotics is a useful reminder: the work that came before deployment was what the customer paid for, and readiness was the product.
Questions for the C-Suite and the Board
Which of our emerging applications could become usable before the programme supporting them is ready? A CTO or programme director can answer by naming the longest prerequisite, whether evidence, an interface or trained staff, and the colleague whose involvement would shorten it.
Where does each emerging line sit against NATO's published demand areas and a named route to a buyer? The answer, which a chief commercial officer is well placed to give, shows which experiments have a commercial destination and which belong in research.
What single uncertainty, if resolved, would most change our next investment decision? Testing that uncertainty is likely to create more value than another demonstration of an already established capability.
How much of our preparation spending would stay useful if the preferred technology or supplier changed? For a CFO, the share held in reusable evidence, data and interfaces is a fair measure of the option value being bought.
Is the whole task becoming faster or cheaper once validation, correction and support are counted? A business-unit leader who measures the accepted task can connect a local improvement to programme value.
Who owns the result after the experiment ends? Clear ownership of the next technical and commercial steps ensures a good result reaches the customer.
What condition would lead us to release the funds for another use? A chair who agrees the exit condition in advance protects delivery capacity and makes a decision to stop a recognised outcome.
References
Department for Science, Innovation and Technology, & UK Research and Innovation. (2024, May 13). Un-jammable quantum tech takes flight to boost UK's resilience against hostile actors [Press release]. GOV.UK. https://www.gov.uk/government/news/un-jammable-quantum-tech-takes-flight-to-boost-uks-resilience-against-hostile-actors
European Union Aviation Safety Agency. (2026, June 3). EASA releases latest issue of its Concept Paper on Artificial Intelligence for comment. https://www.easa.europa.eu/en/newsroom-and-events/news/easa-releases-latest-issue-its-concept-paper-artificial-intelligence
National Cybersecurity Centre of Excellence. (n.d.). Migration to post-quantum cryptography. National Institute of Standards and Technology. Retrieved October 13, 2026, from https://www.nccoe.nist.gov/applied-cryptography/migration-to-pqc
North Atlantic Treaty Organisation. (2025, June 25). Summary of NATO's Rapid Adoption Action Plan. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2025/06/25/summary-of-natos-rapid-adoption-action-plan
North Atlantic Treaty Organisation. (2026a, April 21). NATO DIANA's Rapid Adoption Service enables first R&D contract between Allies and industry. https://www.nato.int/en/news-and-events/articles/news/2026/04/21/nato-dianas-rapid-adoption-service-enables-first-rd-contract-between-allies-and-industry
North Atlantic Treaty Organisation. (2026b, July 8). NATO Innovation Scale-Up Package. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2026/07/08/nato-innovation-scale-up-package
North Atlantic Treaty Organisation. (2026c, September 29). Quantum Technology Roadmap. https://www.nato.int/en/about-us/official-texts-and-resources/official-texts/2026/09/29/quantum-technology-roadmap
Rodriguez, C., Swartz, D., Chinn, D., Stöber, J., & Brukardt, R. (2026, February 12). Future defense tech: Multidomain stacks to build affordable mass. McKinsey & Company. https://www.mckinsey.com/industries/aerospace-and-defense/our-insights/future-defense-tech-multidomain-stacks-to-build-affordable-mass
About Andy Demir
Andy Demir is the editor of SwissTechnic and writes on strategy, international growth and capital in aerospace, defence, space and advanced technology. Drawing on a career in commercial leadership, the analysis examines how technology, sovereignty and industrial capability shape competitive power and long-term value.
