Government Agencies Lead 16.1% of 373 Defence Records at 4.1x the Corpus Rate as Military AI Carries a 1.24-Point Score Premium and Contract Awards Appear at 5.1x the Corpus Concentration
A corporate-led analysis of defence and dual-use innovation across 373 Corporate and 142 Research records finds that every one of ten segments sits below translation parity for a classified rather than commercial reason, that the industrial base has become the subject of one in six announcements, and that TRL 7 records at twice the corpus rate identify the cohort where procurement decisions are being made now.

InnoDexis has published its latest Corporate Intelligence Report covering defence and dual-use innovation, analyzing 373 validated Corporate-stream records and 142 validated Research-stream records on the common window from 1 July to 21 August 2026. The report reveals that defence carries a mean InnoDexis Corporate score of 5.60 against a corpus baseline of 5.26, with government agencies as lead entity on 16.1% of records against 4.0% corpus-wide — the highest government presence of any theme measured — and contract award announcements at 13.9% against 2.7%, a 5.1x concentration. All ten segments show translation ratios below parity, a uniform pattern the report attributes to classification rather than to an absence of science upstream.
Key Findings
The structural profile of defence is defined by three simultaneous measures that describe one procurement mechanism. Government agencies lead 16.1% of records — 4.1x the corpus rate. Contract awards account for 13.9% — 5.1x the corpus rate. Records sit at TRL 7 at 11.5% against 5.7% corpus-wide — twice the corpus concentration. Together these figures describe a market where revenue arrives as a government contract at low readiness rather than as a product at high readiness. Freeform Future's Department of War Innovation Unit award for adaptive space manufacturing at TRL 4, rated High Paradigm Shift, is the procurement signature in a single record: a government buyer funding a capability it does not yet possess, announced publicly by procurement rule rather than corporate choice.
Every one of ten defence segments sits below translation parity — and the report is explicit that this is a source property, not a market finding. In healthcare, sub-theme ratios spanned 0.25x to 3.03x. In space, 2.3x to 29.8x. Here every segment falls between 0.32x and 0.88x with no exception. Ten unrelated markets do not coincidentally share one characteristic unless something upstream of all of them is producing it. That something is classification — defence research is conducted under security restrictions that prevent publication on any timescale relevant to an innovation feed. Cross-stream analysis is therefore the wrong instrument for this theme, and the report states explicitly that sub-parity translation ratios should not be presented as market findings to any reader without that context.
AI and autonomy carry the largest premium measured in any theme across the July to August 2026 report series. Records with an AI or autonomy component score 6.44 against 5.20 without — a 1.24-point gap exceeding healthcare's 0.68-point AI premium. High investment attractiveness runs at 38% against 23%. The gap is wide because in defence, autonomy is not a feature layered onto a product but the product's reason for existing. The Department of War's USD 100 million award to NODA AI for scaling mission command across the joint force and the US Air Force's tenfold ceiling increase on the AcqBot contract to USD 99.9 million are software companies with government revenue at nine-figure scale — confirming that military AI has crossed from experimentation into programme-of-record funding.
The industrial base has become the subject of 64 records — 17% of the theme — describing production capacity, manufacturing, supply chain, and facility expansion rather than capability. The segment scores 6.02 with 38% High attractiveness and 77% at TRL 8–9. Northrop Grumman's USD 3 billion agreements to accelerate missile interceptor production and HII's submarine construction contracts for Block VI Virginia-class and Build II Columbia-class submarines are the large-scale expressions. The more structurally significant signals are geographic: 3D Systems' Saudi joint venture obtaining a military manufacturing licence for local production, and NordSpace investing in North Vector Dynamics for sovereign Canadian defence propulsion, both describe capability being relocated rather than expanded. Sovereign industrial capacity is being purchased alongside sovereign capability.
Space and missile defence is the smallest segment at 28 corporate records and simultaneously the highest quality, with the highest mean score in the theme at 6.54, the highest attractiveness rate at 56%, and the highest contract-award intensity at 39% — two records in five in this segment are a government award. Counter-autonomy became a productised category within the window: Lockheed Martin unveiled MORFIUS X-Rotor explicitly for swarm defeat, introduced PAC-3 ACE as a deliberately low-cost interceptor competing on cost-per-engagement, and Thales partnered with Destinus on integrated counter-UAS and long-range strike. A market where measure and countermeasure are being productised simultaneously tends to sustain procurement growth on both sides rather than resolving toward one.
Startups account for 10% of defence corporate records against 15.3% corpus-wide — the only theme where startup share falls materially below the corpus rate. Government agencies at 16% and large corporates at 16% each exceed it, with mid-size firms and SMEs at 49% between them. The barrier is not capital: Twenty, described as America's first venture-backed cyber warfare startup, raised a further USD 30 million from Khosla Ventures at a USD 1.2 billion valuation — the highest-scoring record in the theme at 9 — and Capitol Meridian Partners closed a USD 1.9 billion second fund at its hard cap in the same window. The barrier is clearance, accreditation, and programme-of-record access, which capital does not remove.
Strategic Insight and Trend Analysis
The most consequential structural finding of the Defence Cross-Stream Report is the identification of defence as the purest procurement market in the InnoDexis corpus — and the implication that the appropriate early-warning product for this theme is a procurement-tracking instrument rather than a technology-tracking one. In healthcare the gate is clinical; in AI infrastructure the gate is power availability. Here the gate is the contract award, and the award is published with a named recipient, an awarding body, and frequently a ceiling value. A tenfold ceiling increase on AcqBot to USD 99.9 million is a capability trajectory stated numerically in public — a form of forward intelligence that product launches cannot match.
The defence analysis also completes the three-report arc on the commercialisation-vehicle indicator. Space established that the indicator works where the route to market runs through company formation and fails where it runs through government contract. Defence confirms that reading from the most extreme example in the corpus: one patent application across 142 research records, government grants at 38%, corporate co-funding at 4%, and no segment above translation parity. The vehicle rate is not wrong; its domain of validity excludes any theme where classification removes the upstream before the data is collected.
The TRL 7 population at twice the corpus rate is the most actionable cohort in the theme for any organisation monitoring near-term procurement transitions. TRL 4 records show what has been funded; TRL 9 records show what is fielded; TRL 7 records — operational prototype demonstration — show what is about to transition. Stratolaunch's ten-plus reusable hypersonic flights, Lockheed's PAC-3 ACE, and Twenty's cyber warfare platform at USD 1.2 billion valuation are each at TRL 7 with committed funding or demonstrated operation behind them. Systems at operational demonstration typically reach fielding or cancellation within two to three years.
Global and Industry Implications
For prime contractors and defence suppliers, the mid-size corporate and SME band supplying 49% of announcement volume contains government-validated capability that would be slow to build internally. NODA AI's mission command award, Trenchant Analytics' AcqBot ceiling increase, MetroStar's Iris maintenance platform, Cohere Technologies' multi-waveform radio access network work, and X-Bow Systems' low-cost interceptor selection are each validated through government award — a form of due diligence that is public and unusually reliable. The competitive dynamic to watch is cost-per-engagement: Lockheed's simultaneous introduction of a counter-swarm effector and a deliberately low-cost interceptor signals that affordability has become a customer specification, and a supplier whose value proposition rests on performance at premium cost should expect that requirement to tighten.
For investors and capital allocators, quality in this theme is concentrated rather than distributed across segments. Space and missile defence at 6.54 mean and 56% High attractiveness across 28 records, munitions and effectors at 6.46 and 50% across 28 records, and air and aerospace platforms at 6.32 and 44% across 57 records are the three segments where quality clusters alongside high contract-award intensity — meaning revenue visibility is unusually good and the customer is a sovereign counterparty. The AI and autonomy premium of 1.24 points is the largest in the corpus and is being underwritten at nine-figure programme scale. Two explicit risks deserve weight: customer concentration is near-absolute, making growth a function of programme decisions rather than market adoption, and startup entry barriers are access-based rather than capital-based, which limits the addressable venture opportunity to a small number of positions of the kind Twenty represents.
For government and defence acquisition bodies, the sovereign-capacity pattern across Canada, Saudi Arabia, Japan, and Luxembourg confirms a competitive supplier environment for nations building domestic capability now — a window that may narrow as the industrial-base consolidation already visible in Northrop Grumman's production agreements and HII's submarine contracts proceeds. The 17% of records concerning production capacity rather than capability identifies magazine depth — how many effectors a force can actually field — as the constraint that has moved to the centre of procurement attention. For acquisition bodies seeking to broaden the supplier base, the evidence points to access barriers rather than capital or capability shortage as the primary constraint, with clearance and programme-of-record access the specific variables that startup share and Twenty's anomalous position together identify.
InnoDexis Statement
"Defence is the corpus's purest procurement market — where the contract arrives as public information years before the capability exists, where the industrial base has become the subject of one in six announcements, and where the appropriate early-warning instrument is a procurement tracker, not a technology one," noted InnoDexis in its latest intelligence report.
Conclusion
The Defence and Dual-Use Cross-Stream Report establishes that the sector runs on a procurement mechanism producing a 5.1x concentration of contract awards, that every segment sits below translation parity for a classified rather than commercial reason that must be stated explicitly in any customer-facing output, and that the AI and autonomy premium at 1.24 points — the largest measured in any theme — is now being underwritten at nine-figure programme scale. Across 373 Corporate and 142 Research records, the evidence confirms the TRL 7 cohort at twice the corpus rate as the most actionable near-term transition signal, sovereign industrial capacity as an active purchasing category across four countries in seven weeks, and structured contract-award extraction as the single highest-value schema addition available across all four cross-stream analyses produced this cycle. As awarding body, programme name, ceiling value, and period of performance are added to the contract-award schema, procurement sources expand into tender portals and award databases, and the classification gap is stated explicitly rather than presented as a market finding, the Defence and Dual-Use framework will provide the most structurally honest procurement-intelligence product the InnoDexis platform has yet produced. The complete Defence and Dual-Use Cross-Stream Intelligence Report July to August 2026 is available to InnoDexis subscribers and enterprise clients.
About InnoDexis
InnoDexis is a global Innovation Intelligence platform that tracks, analyzes, and interprets breakthrough innovations, prototypes, and emerging technologies across industries and countries. Its intelligence helps corporates, investors, and policymakers understand the true structure and direction of global innovation. Learn more at innodexis.ai.