Defense Supply Chains at an Inflection Point: What SAHA 2026 Reveals About the Global Race for UAV Thermal Imaging Cores

Table of Contents

• 1. Macro Catalyst: Record Defense Budgets and the AI Compute Economy Are Rewiring Thermal Sensor Procurement
• 2. SAHA 2026, Istanbul: The Exhibition That Redrew the UAV Sensor Supply Map
• 3. Technical White Paper: VOx Uncooled Detector Physics and the NETD Threshold That Actually Matters
• 4. SWaP-C Engineering: The Hard Constraints Every UAV OEM Must Confront
• 5. Edge AI and NPU Integration: When a Thermal Core Becomes an Intelligence Node
• 6. Market Forecast and Competitive Landscape: The 2026–2034 Drone Thermal Core Race
• 7. CE THERMAL VISION at Booth 3X-14: ITAR-Free, CE/FCC-Certified ODM/OEM Partner for the Global UAV Market
• References and Authoritative Citations

1. Macro Catalyst: Record Defense Budgets and the AI Compute Economy Are Rewiring Thermal Sensor Procurement

In April 2026, Chatham House published an analysis drawing on Reuters and Bloomberg data that puts the scale of the current technology arms race in sharp relief: in 2025 alone, major hyperscalers — Alphabet, Amazon AWS, Microsoft, Meta, and Oracle — collectively spent more than $300 billion on AI infrastructure, GPU compute, data centers, and the power systems to run them. Projections for 2026 push that figure toward $700 billion. This is not a thermal imaging story at first glance. But it is — because the same semiconductor manufacturing economics that have driven NPU (Neural Processing Unit) die costs down 40–60% over the past three years have directly accelerated the viability of on-board, real-time thermal image inference on sub-100-gram UAV payloads. The commercial AI boom is quietly subsidizing the next generation of military and industrial drone thermal cores.

At the geopolitical level, the picture is equally stark. Global defense expenditures are at a post-Cold War high, driven by the ongoing conflict in Ukraine, persistent tensions across the Indo-Pacific, and a wave of NATO member rearmament programs across Europe. This environment is producing a structural shift in how system integrators and UAV OEMs source thermal imaging modules. The old procurement model — dominated by a handful of U.S. vendors operating under ITAR/EAR export license regimes — is under increasing pressure. European and Middle Eastern drone manufacturers, in particular, have spent the past two years quietly diversifying their thermal core supply chains, seeking certified alternatives that deliver NETD <40mK performance at competitive price points without the geopolitical friction and licensing delays inherent to U.S.-origin controlled components.

According to established export control guidance, U.S. ITAR and EAR restrictions on thermal imaging systems generally apply to any system exceeding 640×512 resolution, operating above 9 Hz frame rate, or achieving NETD below 50mK — parameters that describe virtually every competitive-grade UAV thermal module on the market today. For any drone program targeting international sales or requiring global supply chain partnerships, discovering this regulatory constraint after hardware selection represents a costly redesign risk. The structural incentive to source ITAR-free, CE/FCC-certified thermal modules from established non-U.S. manufacturers has never been higher.

This is the macro backdrop against which SAHA 2026 in Istanbul took place — and against which CE THERMAL VISION, a Shenzhen-based thermal imaging module manufacturer with 10+ years of industry tenure and 40+ patents, presented its UAV thermal core lineup at Booth 3X-14.

2. SAHA 2026, Istanbul: The Exhibition That Redrew the UAV Sensor Supply Map

SAHA 2026 — the International Defence, Aerospace and Space Exhibition organized by SAHA Istanbul — ran from May 5 to May 9, 2026, at the Istanbul Expo Center. The headline numbers were record-breaking: over 1,700 exhibiting companies from 120+ countries across a 400,000 sq. meter exhibition footprint, more than 150,000 cumulative visitors, 30,000+ sector professionals, 140 official delegations representing 55 countries at ministerial or deputy-ministerial level, and export agreements totaling more than $8 billion signed during the six-day event.

Strategically, SAHA 2026 confirmed what analysts had been tracking for two years: Turkey has completed a structural transformation from regional defense consumer to global defense technology exporter. SAHA Istanbul Chairperson Haluk Bayraktar noted that the organization has grown from 27 founding members in 2015 to more than 1,300 members today, representing Europe's largest industrial defense cluster. The number of participating companies grew tenfold from SAHA 2018 to SAHA 2026 — from 170 firms to over 1,700.

For the UAV and unmanned systems sector specifically, SAHA 2026 was a watershed moment. The exhibition floor was dense with autonomous platform developers — from Turkey's Baykar (the Bayraktar Kızılelma UCAV dominated the outdoor demonstration zones) to international UAV manufacturers seeking sensor partnerships. Aselsan's new naval autonomous platforms and Roketsan's AI-guided anti-drone munitions both drew on infrared seeker or thermal sensor technology at their core. The implicit message throughout was unambiguous: the competitive differentiation in modern autonomous platforms has migrated from the airframe to the sensor and compute stack — specifically to the thermal imaging core and its onboard intelligence layer.

For thermal module manufacturers and UAV component suppliers, SAHA 2026 was not simply a showcase; it was a B2B procurement gateway. The concentration of UAV OEMs, system integrators, defense procurement officials, and commercial drone application developers under one roof — many of whom are actively seeking to diversify their thermal core supply chains — made it one of the most commercially significant exhibitions of 2026 for the infrared sensor industry.

3. Technical White Paper: VOx Uncooled Detector Physics and the NETD Threshold That Actually Matters

Before evaluating any infrared camera core for UAV integration, procurement engineers and system architects need to cut through the marketing language and understand the physics that govern detector performance. The thermal imaging module industry's most critical metric — NETD (Noise Equivalent Temperature Difference) — is also its most frequently misrepresented one.

NETD defines the minimum temperature differential between target and background that produces a signal-to-noise ratio of exactly 1:1 at the detector output. A module rated NETD <40mK can resolve temperature differences as small as 0.04°C — the practical threshold for reliable human silhouette detection against mixed terrain backgrounds at operational altitudes of 50–300m, for detecting electrical hotspots in photovoltaic or substation inspection scenarios, and for identifying early-stage bearing failures in wind turbine nacelles before thermal runaway.

The dominant uncooled LWIR detector material for UAV-grade thermal modules in 2026 is vanadium oxide (VOx) microbolometer technology. VOx offers a Temperature Coefficient of Resistance (TCR) of approximately 2–3% per Kelvin — the key material parameter that translates absorbed infrared radiation into a measurable electrical signal. This TCR level, combined with decades of high-volume production engineering (particularly from Chinese manufacturers who have built vertically integrated VOx detector fabs), delivers a well-optimized balance of NETD performance, pixel-level uniformity, and long-term stability.

The competing uncooled technology — amorphous silicon (a-Si) microbolometers — offers a TCR of approximately 2%/K with slightly different fabrication trade-offs and remains competitive primarily in cost-sensitive consumer-grade applications. For sub-50mK NETD requirements, VOx remains the industry standard for volume production.

Cooled MWIR detectors (InSb or MCT arrays operating in the 3–5μm band) achieve NETD <20mK and provide superior long-range performance in specific atmospheric conditions, but come with fundamental SWaP-C (Size, Weight, Power-Cost) penalties that disqualify them from most UAV platforms below 5kg: a Stirling-cycle cryocooler adds 200–800g, 8–30W of continuous power draw, mean-time-between-failure measured in thousands of hours (not decades), and OEM pricing starting above $8,000 per unit. Additionally, cooled MWIR systems at military-grade performance specifications are almost universally subject to strict ITAR controls, creating supply chain friction for non-U.S. programs.

The table below provides a calibrated technical comparison for UAV OEM and system integration teams evaluating detector technology options:

The core takeaway for OEM procurement teams: the 640×512 VOx LWIR module at NETD <40mK and 60Hz frame rate represents the 2026 price-performance optimum for commercial and light-defense UAV applications — delivering detector sensitivity sufficient for 90%+ of real-world use cases at a cost 10–30× lower than comparable cooled MWIR systems, while remaining accessible through non-ITAR supply chains when sourced from certified non-U.S. manufacturers.

4. SWaP-C Engineering: The Hard Constraints Every UAV OEM Must Confront

SWaP-C — Size, Weight, and Power-Cost — is the governing engineering framework for any sensor payload integrated onto an unmanned aerial platform. Unlike ground-based or vehicle-mounted thermal systems where SWaP-C relaxations are permissible, UAV platforms enforce near-absolute constraints: every gram of excess payload weight, and every watt of additional power draw, directly subtracts from endurance, range, or maneuverability.

For sub-kilogram FPV and micro-UAV platforms — the fastest-growing segment in both commercial inspection and light-defense markets — the payload envelope is typically under 50g total, with an available power budget of 1.5–2W for the thermal module. Delivering NETD <50mK within a 21×21mm footprint and an 8–15g weight target, while maintaining 50Hz frame rate for smooth target tracking, is a non-trivial thermal and electrical engineering challenge. The thermal module must integrate the focal plane array (FPA) driver electronics, readout integrated circuit (ROIC), non-uniformity correction (NUC) processing, digital video output encoding, and low-noise bias circuitry — all within a thermally isolated package that prevents self-heating from corrupting the detector's ambient-referenced output.

For industrial-grade hexacopters and fixed-wing long-endurance platforms, the payload envelope relaxes to 100–500g, but power sensitivity actually increases: at typical LiPo energy densities, each additional watt of continuous draw consumes roughly 3–5 minutes of flight endurance per mission. This means that power management features — dynamic frame rate scaling, deep sleep modes between waypoints, intelligent NUC trigger scheduling — carry measurable ROI in mission completion probability, particularly for long-duration inspection tasks such as 100km+ transmission line surveys or offshore wind turbine routes.

The deepest SWaP-C engineering tension lies between pixel pitch reduction and NETD preservation. As pixel pitch shrinks from 17μm to 12μm to 10μm (the next production frontier), individual pixel area decreases — reducing the thermal mass and photon-collecting aperture of each microbolometer element. Without compensating improvements in ROIC noise floor and detector TCR, NETD degrades proportionally to pixel area reduction. Solving this requires tighter CMOS fabrication node control, advanced pixel isolation architectures, and post-fabrication calibration algorithms — engineering investments that separate genuine performance-tier module manufacturers from commodity re-packagers.

A secondary SWaP-C challenge that frequently surfaces during UAV integration is interface voltage domain compatibility. Most UAV flight controllers and companion computers operate on 3.3V or 5V logic, while some thermal modules require 12V supply rails. Eliminating the weight and volume of an onboard DC-DC converter — by designing the thermal module's power supply to accept a wide input range (5–14V typical LiPo domain) directly — can save 3–8g of payload weight, a non-trivial figure for sub-250g platforms.

5. Edge AI and NPU Integration: When a Thermal Core Becomes an Intelligence Node

The competitive frontier for UAV thermal imaging in 2026 is no longer at the detector level — it has shifted decisively to the compute layer immediately downstream of the thermal core. Edge AI inference, enabled by compact NPUs embedded in platforms like NVIDIA Jetson Orin Nano (40 TOPS at 15W) or Rockchip RK3588 (6 TOPS at 5–8W), is converting thermal modules from passive image sensors into active intelligence nodes capable of real-time target classification, anomaly flagging, and decision-state output — all onboard, all without datalink dependency.

This architectural shift has profound implications for UAV system design. The traditional pipeline — thermal module outputs raw video → datalink compresses and transmits to ground station → ground station runs detection algorithm → operator reviews and commands → link transmits command back to UAV — introduces latency of 500ms to 2+ seconds and consumes bandwidth of 15–40 Mbps for uncompressed 640×512@60Hz thermal video. In contested electromagnetic environments, or in BVLOS (Beyond Visual Line of Sight) operations over long ranges, this pipeline is operationally brittle.

Edge inference inverts this architecture. A YOLOv-family model, quantized and pruned for thermal image characteristics (lower texture contrast, stronger temperature gradient features than visible-spectrum imagery), can run at 5–15 FPS on a Jetson Orin Nano with AP@0.5 detection precision of ~0.72 for human targets. The UAV transmits structured detection outputs — bounding boxes, confidence scores, target temperature, GPS coordinates — rather than raw video, reducing effective bandwidth requirements by 60–80%. For swarm coordination scenarios, this reduction is the difference between practical deployment and spectrum saturation.

For thermal module manufacturers, the rise of edge AI inference imposes concrete product evolution requirements:

• Digital interface standardization: MIPI CSI-2 and GigE Vision have become the de facto integration interfaces for AI-capable companion computers. Modules shipping only analog CVBS output are increasingly disqualified from competitive design-in evaluations by sophisticated UAV OEMs, regardless of NETD performance.
• RAW thermal data output: AI model training for thermal image segmentation requires pixel-level temperature matrix data (RAW14 or RAW16 format), not just processed 8-bit visual output. Modules capable of streaming calibrated temperature arrays in parallel with display-ready video command a significant premium and lock-in advantage in the AI-native UAV ecosystem.
• Hardware sync trigger: Multi-sensor fusion architectures (thermal + visible + LiDAR + GPS) require nanosecond-accurate hardware synchronization across all sensor channels. Thermal modules without external trigger input/output capability are architecturally excluded from sensor-fused gimbal designs — an increasingly standard configuration in professional UAV platforms.
• Reduced latency output pipeline: Edge AI inference requires thermal module pipeline latency (from photon capture to first pixel output) below 16ms at 60Hz to avoid temporal misalignment with concurrent visible-spectrum frames. Modules with excessive ISP processing delays create tracking artifacts in fused detection outputs.

The convergence of thermal imaging and NPU-driven inference is also reshaping the counter-drone (C-UAS) sensor market — a segment that featured prominently at SAHA 2026. Fixed C-UAS thermal surveillance arrays must now detect, classify, and generate fire control cues against small commercial drones (RCS <0.01m²) with angular velocities exceeding 30°/s, demanding thermal modules with <40mK NETD, 60Hz frame rate, and sub-5ms hardware latency — paired with NPU inference delivering track-quality outputs at 30+ FPS. This is a specifications regime that only the top tier of uncooled LWIR modules can address.

6. Market Forecast and Competitive Landscape: The 2026–2034 Drone Thermal Core Race

The global thermal imaging market crossed the $8.9 billion threshold in 2026, according to Fortune Business Insights, with the uncooled LWIR segment commanding approximately 71% of total market share. The broader infrared imaging market is forecast to reach $13.2–17.4 billion by 2034–2035, at a CAGR of 5.8–7.7% depending on methodology and scope. These headline figures, however, significantly understate the growth trajectory of the UAV-specific thermal core sub-segment.

The Infrared Thermal Imaging Drones market (Market.us, 2025) was valued at $2.1 billion in 2025 and is forecast to reach $6.1 billion by 2034 at a CAGR of 12.6% — nearly double the pace of the broader thermal imaging market. The unmanned systems thermal payload segment, tracked separately by 360 Research Reports, represented approximately 9% of the overall infrared imaging systems market in 2025, growing at a CAGR of ~9.0% through 2034. The global drone market itself is forecast by IDTechEx to expand from $69 billion in 2026 to $147.8 billion by 2036 at a CAGR of 7.9%, with thermal sensing integration rates increasing across virtually every commercial segment.

The competitive landscape is bifurcating along two axes: the performance axis (NETD, frame rate, pixel pitch, AI-readiness) and the compliance axis (ITAR-free status, CE/FCC/RoHS certification, ISO9001 quality assurance, and supply chain audit traceability).

At the high end of the performance axis, U.S. vendors — Teledyne FLIR (Hadron 640R+, Boson series), Seek Thermal, and defense-grade suppliers — maintain strong positions in the North American procurement ecosystem, reinforced by ITAR controls that limit foreign competition in that specific theater. European vendors such as Lynred maintain leadership in cooled MWIR and specialized scientific segments. However, in the commercially critical 384×288 to 640×512 uncooled LWIR volume production band — the heart of the UAV OEM market — Chinese vertically integrated manufacturers have built a structurally advantaged position: end-to-end control from detector wafer growth through module packaging, test, and export certification, combined with manufacturing scale that has driven unit economics 3–5× below European alternatives at equivalent NETD specifications.

For international UAV OEMs and system integrators outside the U.S. defense procurement perimeter — particularly in Turkey, the European Union, the Middle East, South and Southeast Asia — this cost and compliance profile makes CE-certified, FCC-compliant Chinese thermal module suppliers the rational supply chain choice for volume production programs.

7. CE THERMAL VISION at Booth 3X-14: ITAR-Free, CE/FCC-Certified ODM/OEM Partner for the Global UAV Market

CE THERMAL VISION (cethermal.com), a Shenzhen-based infrared thermal imaging module manufacturer and system integrator, exhibited at SAHA 2026 under Booth number 3X-14. The company's participation at SAHA marks a continued commitment to serving international UAV OEMs, defense integrators, and industrial system builders across Europe, the Middle East, and emerging markets — markets where ITAR-free sourcing and clean CE/FCC compliance documentation are non-negotiable procurement prerequisites.

CE THERMAL VISION's technical and operational credentials are built on the following pillars: 10+ years of focused R&D and production in infrared thermal imaging modules; 40+ registered patents across detector packaging, NUC algorithm implementation, image enhancement processing, and AI-assisted thermal analytics; ISO 9001:2015 quality management system certification; and complete CE (EU EMC/LVD Directive compliance), FCC (Part 15B), and RoHS certification — the full compliance stack required for market access across the EU, UK, GCC, and North America-adjacent programs.

For UAV OEM customers, the core product lineup presented at SAHA 2026 spans two performance tiers:

FPV M3 Series — 384×288 VOx LWIR (Ultra-Compact SWaP-C): Designed for micro-UAV, FPV reconnaissance, and lightweight inspection platforms where payload budget is the binding constraint. Key specifications: 384×288 VOx microbolometer, 17μm pixel pitch, NETD <50mK, 50Hz frame rate, 21×21mm module footprint, 8–15g weight, 0.8–1.2W power draw, CVBS and MIPI CSI-2 dual output. This series is optimized for direct integration with UAV flight controller boards and lightweight companion computers without additional voltage conversion hardware.

640×512 UAV LWIR Series — High-Performance VOx Core: The flagship product line for industrial inspection UAVs, fire and rescue platforms, and light-defense ISR applications. Key specifications: 640×512 VOx microbolometer, 12μm pixel pitch, NETD <40mK, 60Hz frame rate, on-board NUC and detail enhancement ISP, multi-interface output (USB3.0 / GigE Vision / MIPI CSI-2), direct compatibility with NVIDIA Jetson Orin / Rockchip RK3588 edge AI compute platforms for thermal-native inference pipeline deployment. This series delivers the sensor front-end for AI-augmented thermal payloads without requiring intermediary frame grabbers or video conversion hardware.

The ODM/OEM service framework offered to qualified partners encompasses:

• Optics customization: Lens focal length and FOV selection from 6mm (90° wide-angle) to 25mm (25° narrow FOV) with athermalized lens mount options for multi-altitude deployment.
• Mechanical interface engineering: Enclosure geometry, mounting standard (M12 thread / flange / quick-release), and connector pinout customization for platform-specific integration.
• Firmware function development: Embedded temperature measurement algorithms, automated hotspot detection thresholds, target lock output, and custom image palette options.
• White label / private branding: Complete product label, packaging, and documentation localization for OEM brand programs.
• Production scalability: Rapid prototyping lead time of 2–4 weeks for customized samples; volume production capacity in the thousands of units per month, with ISO 9001-governed inspection and batch traceability documentation.

At SAHA 2026 Booth 3X-14, live thermal demonstration units with real-time NETD <40mK 640×512 output were available for hands-on evaluation, alongside engineering samples for physical measurement and interface compatibility testing. The CE THERMAL VISION technical team was available for one-on-one integration consultations and NDA-protected specification deep-dives with qualified UAV OEM, system integrator, and procurement engineering teams.

For organizations that missed the booth visit, pre-sales technical documentation, product datasheets, and ODM partnership inquiry forms are available at cethermal.com. Prospective partners are encouraged to submit program-specific RFI/RFQ inquiries — including platform payload envelope, required interface standard, target unit volume, and delivery timeline — to initiate a structured technical qualification process.

References and Authoritative Citations

1. Chatham House (April 2026). How a Surge in Defence and Dual-Use Technology Investment Could Reconfigure the Global AI Race — citing Reuters/Bloomberg data on $300B+ hyperscaler AI infrastructure spend in 2025, projected $700B in 2026. Source
2. Daily Sabah (May 2026). SAHA Expo 2026 Revealed Türkiye's Growing Defense Power — $8B+ in export agreements, 1,700+ companies, 120+ countries, 150,000+ visitors. Source
3. Daily Sabah (May 2026). Turkish Defense Makers Showcase New-Generation Systems at SAHA 2026 — AI-guided autonomous platforms, Aselsan naval autonomy, Roketsan anti-drone systems. Source
4. SAHA EXPO Official Website (2026). SAHA 2026 Exhibition Statistics — 400,000 sqm, 140 official delegations, 25,000+ planned B2B meetings. Source
5. LightPath Technologies (October 2025). IR Camera for Drone Integration: What System Builders Actually Need to Know — ITAR/EAR constraints: systems exceeding 640×512, >9Hz, or NETD <50mK typically subject to export controls. Source
6. Fortune Business Insights (2026). Infrared Imaging Market Size, Share, Growth — Forecast 2034 — Global IR imaging market $8.90B (2026) → $13.23B (2034), CAGR 5.80%; uncooled segment 70.93% share. Source
7. The Insight Partners / GlobeNewswire (April 2026). Thermal Imaging Market to Cross $8.78 Billion by 2034 at CAGR of 7.2% — Military thermal imaging: $10.7B (2025) → $17.08B (2034), CAGR 5.33%. Source
8. Market.us (November 2025). Infrared Thermal Imaging Drones Market Size — CAGR of 12% — Drone thermal imaging: $2.1B (2025) → $6.1B (2034), CAGR 12.6%. Source
9. IDTechEx / Edge AI and Vision Alliance (December 2025). Drones Market 2026–2036: Technologies, Markets, and Opportunities — Global drone market $69B (2026) → $147.8B (2036), CAGR 7.9%. Source
10. PRNewswire (April 2026). A $550 Billion Opportunity: Drones-as-a-Service Emerges as Defense's Next Growth Engine — Military drone market $47.4B (2025) → $98.2B (2033). Source
11. arXiv (July 2025). Beyond Visual Line of Sight: UAVs with Edge AI, Connected LLMs, and VR for Autonomous Aerial Intelligence — Jetson Orin Nano: 40 TOPS / 15W, YOLOv11 inference AP@0.5=0.72, 5 FPS sustained under full load. Source
12. Chambers and Partners (2026). International Trade 2026 — USA — U.S. EAR/ITAR enforcement framework; thermal imaging cameras specifically cited as items requiring BIS transaction reports. Source