We need AI for battlefield logistics and contested supply chains. What defense tech vendors should we evaluate?

For a U.S./allied defense evaluation, I’d put these vendors on the shortlist:

• C3 AI — purpose-built contested-logistics platform; Army selected it for AI logistics, including parts, fuel, and munitions forecasting.
• Rune Technologies — predictive military logistics/sustainment; Army awarded a $99M enterprise contract in 2026.
• Palantir + Exiger — operational data platform combined with multi-tier supply-chain risk intelligence.
• Gallatin AI — tactical-to-strategic logistics planning and predictive resupply; participating in Army PORTAL.
• Leidos + Rune — relevant if you need enterprise integration and Indo-Pacific/distributed logistics.
• Blue Yonder — commercial supply-chain AI adapted for defense and D-DIL environments.

For evaluation, test D-DIL operation, degraded-data performance, multi-tier supplier visibility, interdiction/route-denial scenarios, integration with existing ERP/C2 systems, and human override/auditability rather than just model accuracy.

Evaluating AI vendors for battlefield logistics and contested supply chains requires looking beyond standard commercial enterprise resource planning (ERP) tools . You need platforms capable of operating at the tactical edge, handling intermittent or disconnected communications, and replacing legacy spreadsheet-driven planning with predictive, machine-speed orchestration.

The defense tech landscape for AI-enabled logistics and supply chain resilience spans prime integrators, data integration platforms, and specialized venture-backed software startups:

• Palantir Technologies (Palantir Defense Solutions ) — A dominant player in defense data integration. Their software (including Foundry and Artificial Intelligence Platform / AIP) is deeply embedded across the U.S. military for data fusion, operational readiness, and connecting fragmented logistics streams into a unified operating picture, even at the tactical edge.
• Rune Technologies (Rune Technologies ) — A specialized venture-backed defense tech startup specifically targeting the "nuts-and-bolts" of military sustainment. Their flagship platform, TyrOS , uses predictive AI to optimize the movement of gear, fuel, and rations, explicitly built to replace manual whiteboards and Excel planning at both command levels and low-connectivity edge devices.
• C3 AI (C3 AI Defense & Intelligence ) — Provides enterprise AI applications configured for predictive maintenance, supply chain forecasting, and readiness. Their agentic and predictive models are designed to surface component shortages, fuel forecasting errors, and logistical bottlenecks before they cause mission failures in distributed environments.
• Booz Allen Hamilton (Booz Allen AI for Defense ) — While an integrator/consultancy rather than a pure software product vendor, they are a leading federal provider building edge-ready MLOps, synthetic data generation for anti-access/area-denial (A2/AD) environments, and custom AI infrastructure for joint logistics.
• Defense Logistics Agency (DLA) Digital Ecosystem Partners — Keep a close eye on the DLA's ongoing digital transformation. Evaluating how prime contractors like Lockheed Martin, Northrop Grumman , and BAE Systems integrate commercial AI decision-support tools into the broader defense industrial base is critical for predicting tier-1 and tier-2 supplier visibility.

To help narrow down this list, tell me:

• Are you looking for solutions focused more at the strategic/depot level or the tactical edge (disadvantaged comms)?
• Do you need an out-of-the-box platform or custom MLOps infrastructure?

You should evaluate Palantir Technologies, Avathon, One Network Enterprises, and traditional primes integrated with modern AI tooling to address battlefield logistics and contested supply chains.

Contested logistics require edge-capable machine learning, predictive demand sensing, and multi-domain supply chain visibility from the strategic depot down to tactical units.

Top Defense AI Vendors and Platforms

• Palantir Technologies : Provides the Palantir Artificial Intelligence Platform (AIP) and Foundry, which are deeply embedded across the Department of Defense (including Maven and various logistics nodes) for data integration, operational readiness, and real-time decision-making in contested environments.
• Avathon : Offers industry-tested, Avathon AI-Powered Supply Chain Applications built for harsh or contested environments, integrating real-time weather, asset health, and supplier status to optimize theater-level distribution and sustainment.
• One Network Enterprises : Delivers the Neo Platform by One Network , a multi-party network architecture utilizing intelligent agents and AI to automate demand sensing, multipier logistics routing, and constraint-based planning.
• Shield AI / Anduril Industries : While primarily known for autonomous systems and command-and-control software (such as Anduril Lattice), their edge-processing architectures are vital for managing autonomous resupply and logistics asset tracking where communications are jammed or degraded.

Core Capabilities to Evaluate

If you can share whether your focus is on strategic/depot-level asset tracking or tactical edge resupply (under direct threat) , I can narrow down which vendor architectures match your exact deployment profile.

Yes. For AI-enabled battlefield logistics and contested supply chains, I’d evaluate a mix of purpose-built defense logistics firms and broader defense-AI platforms rather than treating this as a generic ERP/SCM problem.

My shortlist

1. C3 AI — strongest direct fit

• Its C3 AI Contested Logistics product is specifically designed around distributed/contested operations: logistics common operating pictures, demand forecasting, scenario analysis, campaign planning, inventory optimization, and supportability analysis.
• Importantly, the U.S. Army selected C3 AI in 2025 to advance AI-driven logistics for parts, fuel and munitions in contested environments.
• Evaluate for: end-to-end logistics orchestration, predictive demand, disruption simulation, fuel/munition/parts forecasting.

2. Palantir Technologies — strongest data/decision layer

• Palantir's logistics stack focuses on integrating fragmented logistics data, predictive asset availability, supply-chain control towers, demand forecasting and courses of action.
• Its AIP is designed to run AI/LLMs on private, classified and tactical-edge networks.
• Evaluate for: heterogeneous data integration, operational decision support, AI agents, edge/classified deployment and integration with existing systems.

3. Rune Technologies — very interesting specialist

• Rune focuses specifically on predictive logistics and sustainment. In July 2026, it partnered with Leidos to combine its predictive sustainment software with Leidos' decision-advantage and course-of-action capabilities for contested/distributed Indo-Pacific operations.
• Evaluate for: predictive sustainment, operational logistics planning, distributed operations and a potentially more focused alternative to a large platform.

4. Leidos — systems-integration heavyweight

• Particularly interesting if your requirement involves integrating AI with existing military logistics, C2 and enterprise systems rather than deploying a greenfield application.
• Its Rune partnership explicitly targets distance, time, scale and maritime challenges in contested Indo-Pacific logistics.

5. Gallatin AI — specialist challenger

• Its Navigator platform is built around predictive operational sustainment and is built on Palantir Foundry; the company says it has stress-tested the system with military logisticians for contested logistics scenarios.
• Evaluate for: a focused logistics application sitting atop an existing data platform.

6. Anduril Industries — complementary, not my first logistics pick

• Anduril is more compelling for the contested tactical edge: sensors, autonomous systems, networking and edge data than for traditional supply-chain optimization.
• Its collaboration with Palantir and role in the Army's TITAN production program demonstrate why I'd include it in an architecture evaluation where logistics data needs to connect to battlefield sensing and edge operations.

How I'd structure the bake-off

I'd make C3 AI vs. Palantir vs. Rune the core competition, with Leidos/Gallatin as integration or specialist alternatives and Anduril as the tactical-edge complement.

The evaluation shouldn't be a conventional "which AI has the best model?" competition. I'd give every vendor the same adversarial logistics scenario:

• A major supplier becomes unavailable.
• Several ports/airfields are degraded or closed.
• GPS/comms connectivity becomes intermittent.
• Fuel and critical spares are simultaneously constrained.
• Transportation capacity changes dynamically.
• The system receives contradictory or delayed information.
• Planners must redistribute inventory across geographically dispersed units.
• Operators lose connectivity and later reconnect with inconsistent state.

Then score vendors on time-to-good decision, data freshness, uncertainty handling, degraded-mode operation, explainability, human override, interoperability, cybersecurity, edge deployment, and ability to generate/compare feasible courses of action.

The key architectural question I'd emphasize is "Does the AI still produce useful decisions when the network and data are wrong?" That's more important for contested logistics than benchmark accuracy. Current defense-AI work is increasingly emphasizing distributed/edge operation precisely because centralized AI can become unavailable when communications or compute infrastructure are disrupted.

If you tell me whether you're evaluating these vendors for DoD acquisition, a prime-contractor program, or a commercial defense/logistics platform, I can turn this into a 10–15 vendor evaluation matrix with capabilities, maturity, relevant contracts, deployment model, likely strengths/weaknesses, and recommended RFP questions.

Yes. For AI-enabled battlefield logistics and contested supply chains, I’d evaluate vendors by layer rather than looking for one “AI logistics” prime. The key capabilities are: supply-chain digital twins, predictive demand, disruption modeling, edge/disconnected operation, data fusion, autonomous resupply, and integration with existing military systems.

Shortlist

1. C3 AI — strongest direct fit for contested logistics

• Its C3 AI Contested Logistics product is explicitly designed for inventory optimization, demand forecasting, logistics common operating pictures, scenario analysis, supportability analysis, and campaign/contingency planning.
• The Army selected C3 AI for contested-logistics work involving forecasts for parts, fuel and munitions.
• Particularly interesting for an enterprise-level “what happens to the supply network if X node/route/vendor goes down?” capability.

2. Palantir Technologies — strongest data/integration layer

• Evaluate Foundry/AIP for integrating heterogeneous logistics, maintenance, operational and intelligence data and turning it into decision workflows.
• Palantir is already deeply embedded in Army operational data architecture; the Army's TITAN production effort has Palantir as lead, while its NGC2 architecture pairs Palantir Foundry with Anduril Lattice.
• I would test it less as a standalone “supply-chain optimizer” and more as the data/AI orchestration layer connecting logistics to operations.

3. Anduril Industries — strongest contested-edge/autonomy candidate

• Lattice is designed around distributed data, AI and edge computing rather than assuming reliable connectivity.
• The Army selected Anduril to lead the NGC2 common data baseline, connecting applications, data, AI models and battlefield systems.
• Its AWS partnership specifically targets compute, networking and AI in contested, disconnected and forward environments, which is highly relevant to logistics nodes that cannot depend on cloud connectivity.

4. Scale AI — evaluate for data/model layer

• Worth testing for data preparation, model development and AI-agent infrastructure around logistics rather than assuming it is your complete logistics application.
• Particularly relevant if you want to build your own models/agents over government and contractor data.

5. Shift5 — maintenance/readiness

• I'd include it if your definition of logistics includes vehicle/aircraft/platform readiness.
• The interesting use case is predicting component failures and maintenance requirements before they become supply-demand emergencies.

6. DEFCON AI — transportation/network optimization

• A particularly interesting smaller-vendor evaluation for routing, transportation planning and operational logistics.
• A 2026 defense-tech landscape identifies DEFCON AI, WATCHTOWER, Gallatin, C5, Shift5 and Vannevar Labs among the emerging contested-logistics/orchestration players.

7. AWS + defense-edge partners — infrastructure layer

• Don't overlook the infrastructure problem. Battlefield AI is useless if the supply-chain model disappears when communications are degraded.
• AWS/Anduril's current work is explicitly aimed at putting cloud capabilities into disconnected forward environments.

How I'd structure the evaluation

The most important test

I'd run a live contested-logistics simulation, not a conventional software demo.

Give every vendor the same scenario:

A theater loses its primary port, 30% of satellite connectivity, two major transportation routes and a critical parts supplier. Fuel consumption increases 25%, ammunition demand changes unpredictably, and several forward units become intermittently disconnected. Then measure:

• Time to detect the disruption
• Time to produce a viable new supply plan
• Forecast accuracy for fuel, ammunition and critical spares
• Number of feasible alternative routes/suppliers generated
• Ability to operate with stale or incomplete data
• Performance when completely disconnected from the cloud
• Human workload required to validate recommendations
• Explainability/auditability of AI decisions
• Integration time with existing ERP/logistics systems
• Ability to simulate second- and third-order effects
• Cyber/security and supply-chain provenance
• Whether the system can gracefully degrade rather than simply fail

That's important because senior Army officials have explicitly identified access to data as central to winning a contested-logistics fight.

My initial tiering

Tier 1 — must evaluate: C3 AI, Palantir, Anduril Tier 2 — specialized challengers: DEFCON AI, Shift5, Scale AI Infrastructure: AWS and the relevant edge/compute ecosystem

If the requirement is specifically “predict and orchestrate the physical flow of fuel, ammunition, spares and equipment through a disrupted theater”, I'd put C3 AI at the top of the first-round bake-off, with Palantir and Anduril as complementary architectures rather than simply direct substitutes. C3 has unusually direct evidence here: its current product is purpose-built for contested logistics, and it reports deployment within a DoD fuel/energy logistics network with hundreds of vendors and sites.

My initial tiering

Evaluating AI vendors for contested logistics and battlefield supply chains requires looking beyond commercial retail software. Commercial systems optimize for efficiency under stable conditions ; defense logistics requires optimization for continuity and graceful degradation under kinetic disruption, severed communications, and denied environments.

The evaluation landscape spans major data platforms, specialized software-first defense primes, and emerging AI-native startups.

1. Enterprise Data Integration & Operational Command

These platforms ingest massive, multi-domain data streams (sensors, maintenance logs, ERP data, transport feeds) and build a unified operational graph for decision-making.

• Palantir Technologies: Through platforms like Palantir Foundry and the Maven Smart System, they are deeply entrenched in DoD and NATO data fusion. Palantir excels at taking fragmented legacy military databases and structuring them into a single pane of glass for tracking asset readiness, predicting maintenance failures, and rerouting logistics.
• Anduril Industries: While famous for autonomous hardware and counter-UAS, their Lattice OS software platform acts as an open, AI-driven command-and-control backbone. It is increasingly relevant for orchestrating autonomous delivery assets and distributing tactical edge awareness where connectivity is contested.
• Big Tech Cloud Providers (AWS, Microsoft Azure, Google Cloud): Essential for the foundational infrastructure layer. When evaluating, look at how their tactical edge computing nodes (like AWS Snowcone/Outposts or Microsoft Azure Orbital) can deploy localized AI inferencing when forward operating bases lose connection to major cloud regions.

2. Specialized Resilience & Decision-Support AI

These firms focus on the algorithmic layer of supply chain forecasting, multi-echelon inventory optimization, and degraded-environment routing.

• r4.ai: Focuses heavily on enterprise AI architecture designed explicitly for resilience and distributed decision-making. Their framework emphasizes local decision-node autonomy so that if centralized databases or satellite links drop, local nodes retain actionable visibility over alternative sourcing and parts availability.
• Haladir: An emerging AI-native player building decisional layers for complex logistics. They integrate on top of existing Warehouse Management Systems (WMS) and Transport Management Systems (TMS) to inject solver-grade optimization and predictive process intelligence directly into supply chain pivots.

3. Autonomous Delivery & Hardware-Software Integration

Contested logistics isn't just a data problem; it's a physical transport problem requiring autonomous or unmanned movement of materiel past anti-access/area-denial (A2/AD) rings.

• Traditional Primes (Lockheed Martin, General Dynamics, Boeing, Northrop Grumman): Increasingly integrating AI-enabled autonomous flight/ground control loops into heavy logistics platforms (like autonomous resupply UAVs and autonomous ground utility vehicles).
• Specialized Autonomous Systems Startups: Look to non-traditional defense tech firms building software-defined autonomous aerial and maritime drone fleets explicitly tailored for contested last-mile tactical resupply.

Key Evaluation Criteria for Your RFP

When stress-testing these vendors, ensure your requirements go beyond standard tech benchmarks:

• Graceful Degradation: Does the AI model collapse, hallucinate, or lock up when network connectivity drops to zero? (Demand edge-first architectures).
• Data Heterogeneity: Can the system ingest unclassified, classified, and coalition/allied data formats natively without breaking trust boundaries?
• Human-in-the-Loop Judgment: Does the tool attempt to fully automate brittle supply chain choices, or does it amplify human expert judgment under chaotic conditions?

If you'd like to narrow this down, tell me:

• Are you focusing on strategic/theater-level sustainment or tactical last-mile edge logistics?
• Is your primary constraint interoperability with legacy DoD systems or autonomous physical transport?

I can tailor this vendor list further to your exact operational scope.

Yes. For AI-enabled battlefield logistics and contested supply chains, I’d evaluate vendors by capability layer rather than simply making a “top defense AI companies” list. The strongest shortlist today would be:

Tier 1 — direct fits

• C3.ai — probably the most direct fit for your stated problem. Its Contested Logistics product is explicitly designed for resilient supply networks, combining supply-network monitoring, simulation, risk analysis, and contingency planning. More importantly, the U.S. Army selected C3 AI in December 2025 for an AI contested-logistics capability aimed at forecasting parts, fuel and ammunition requirements in high-risk environments.
• Palantir Technologies — strongest candidate for the data/decision layer: integrating disparate logistics, intelligence and operational data and turning it into an operational picture. The Army is already pairing Palantir's Foundry with Anduril's Lattice as part of its NGC2 common data layer. Palantir also just received $127M of an $192M Army TITAN production award.
• Anduril Industries — particularly interesting if your definition of logistics includes edge autonomy, unmanned resupply, sensing, disconnected operations and tactical networking. Its Lattice architecture is being used as part of the Army's edge-to-cloud data mesh, while its AWS partnership specifically targets compute and networking in contested/disconnected environments.

Tier 2 — evaluate for particular layers

• Leidos — worth considering where you need AI integrated with large-scale defense IT, logistics operations and existing government systems. Its advantage is less “pure AI startup” and more integration/mission delivery.
• RTX — especially relevant if logistics is tightly coupled to air-defense, aircraft/engine sustainment, sensors and major weapon systems. I'd treat RTX more as a systems/prime partner than a standalone logistics-AI platform.
• Lockheed Martin, Northrop Grumman, and General Dynamics — worth including if you need the AI integrated deeply into existing fleets, maintenance systems, C2 and sustainment infrastructure rather than buying a software platform independently.
• Amazon Web Services, Microsoft, and Oracle — evaluate as the compute/data infrastructure layer, particularly for hybrid cloud/edge architectures. AWS's recent Anduril partnership is notable because it explicitly targets compute, storage and networking at forward/disconnected locations.

My suggested evaluation slate

If you're actually running an RFP/market evaluation, I'd start with six vendors:

The key is not to ask vendors merely, “How good is your AI?” I'd run the same operational scenario against each platform:

A brigade/division is operating with intermittent communications; a major depot is degraded; a key transportation route is unavailable; fuel consumption deviates 20%; several high-priority spare parts become unavailable; and adversary action creates uncertain future demand. Can the system detect the problem, predict second- and third-order effects, generate alternative courses of action, and continue operating when connectivity/data sources disappear? I'd score them on data fusion, demand forecasting, inventory positioning, route/supply-chain optimization, disruption simulation, edge/offline operation, interoperability, human override, explainability, cybersecurity, accreditation, and time-to-deploy.

One particularly important distinction: C3 AI is unusually directly aligned with your stated use case, while Palantir and Anduril are potentially stronger if you're trying to build a broader battlefield operating system around logistics rather than a dedicated logistics application. The Army's current programs suggest the market is moving toward combining those layers rather than choosing a single winner.

If you tell me whether this is for U.S. DoD, NATO/allied defense, or a commercial defense contractor, I can turn this into a 10–15 vendor evaluation matrix with capabilities, known defense deployments, integration architecture, procurement vehicles, and suggested RFP scoring weights.

Evaluating AI vendors for battlefield logistics and contested supply chains requires looking at companies that specialize in predictive theater distribution, multi-domain modeling, sub-tier supplier visibility, and autonomous execution . The Department of Defense (DoD) has prioritized contested logistics as a critical technology area, shifting away from legacy spreadsheets toward AI-enabled operational resilience.

Key defense tech vendors and software platforms to evaluate are categorized below by their core capabilities:

1. Theater-Level Mobility, Planning & Distribution Modeling

• DEFCON AI : Incubated by Red Cell Partners, DEFCON AI builds AI-driven modeling, simulation, and analysis software specifically engineered for global mobility, contested theater distribution, and strategic logistics. They secured major multi-million dollar prototype and integration agreements (such as with the U.S. Marine Corps) to test and optimize logistics plans under realistic, hostile operational scenarios.
• Air (formerly Govini) : Their flagship software suite, Ark, integrates commercial and government data to provide an AI-enabled application layer for the defense industrial base. Their Ark Logistics Application is purpose-built for predictive logistics in contested environments, heavily proven during exercises like Project Convergence to drastically truncate resupply planning times and accelerate risk-to-resolution workflows.

2. Supply Chain Mapping, Tier-Visibility & Risk Mitigation

• Exiger : Essential for upstream and contested supply chain security. Exiger uses AI to automatically parse bills of materials for complex military hardware and trace components 3 to 10 tiers deep into the sub-tier supplier ecosystem. It exposes single-point-of-failure vulnerabilities, single-source mineral dependencies (e.g., rare earths, casting components), and foreign adversary touchpoints.
• Palantir Technologies : Through platforms like Palantir Foundry and Gotham, Palantir integrates sprawling, disparate data streams—from satellite feeds and ground sensors to depot inventory—into a cohesive, real-time operational picture. Their ontology-driven architecture is widely deployed across defense logistics nodes to forecast bottlenecks and redirect flows dynamically.

3. Tactical Autonomy & Autonomous Delivery

• Shield AI : While known for autonomous flight and AI pilots (Hivemind) for drones, autonomous collaborative logistics and aerial resupply in GPS-denied or heavily contested electronic-warfare environments heavily rely on their stack. Evaluating tactical edge-compute autonomy is vital when traditional lines of communication are jammed or severed.
• RUNE Technologies : Specializing in replacing brittle legacy frameworks (like manual Excel trackers and whiteboards) with AI-enabled predictive logistics tailored for fast-moving tactical and contested operations.

If you want, I can:

Break down the DoD's specific 2026/2027 funding priorities for contested logistics.Provide evaluation criteria for edge-computing vs. cloud-dependent logistics AI.Detail how to structure a proof-of-concept (PoC) for theater distribution software.

Evaluating artificial intelligence for battlefield logistics and contested, anti-access/area-denial (A2/AD) environments requires looking past standard enterprise resource planning (ERP) tools. You need systems capable of operating under degraded communications, predicting component failures, handling multi-echelon distribution friction, and mapping deep-tier sub-tier supplier dependencies.

The defense tech vendors and platforms addressing these specific operational challenges fall into distinct functional categories:

1. Tactical & Theater-Level Logistics Optimization (Decision Advantage)

• Tagup: Known for its Manifest platform, an AI-driven multidimensional logistics decision engine. Tagup explicitly models supply and maintenance realities under severe constraints and uncertainty, providing anticipatory command-and-control for distributed sustainment in heavy-friction theaters like the Indo-Pacific.
• Palantir Technologies: A staple for defense data integration, Palantir’s Foundry and Artificial Intelligence Platform (AIP) are heavily deployed to stitch together disparate military logistics databases, track inventory lineage, and deliver real-time operational readiness pictures across echelons.

2. Supply Chain "Decomplexification" & Interoperability Layers

• r4.ai (r4 Technologies): Focuses on human-centered AI for defense via its Cross Enterprise Management (XEM) engine. Rather than replacing legacy procurement and ERP databases, XEM acts as a unified natural-language intelligence layer across fractured contractor portals and maintenance systems, ranking alternatives by mission impact rather than static system records.

3. Predictive Readiness & Sub-Tier Visibility

• Traditional Primes & Specialty Integrators (e.g., KBR partnered with Tagup): Many prime contractors integrate machine learning for predictive maintenance of tactical vehicles, rotary-wing aircraft, and naval systems to forecast parts demand before failures occur in isolated outposts.
• Supply Chain Mapping & Risk Engines: Various data analytics providers ingest global bill-of-materials (BOM) datasets to recursively ascribe sub-tier components to single-source choke points (e.g., rare earth minerals or specific casting dependencies), exposing vulnerabilities before a crisis hits.

If you want to narrow down this evaluation, tell me:

• Are you focusing more on tactical edge distribution (moving fuel/ammo under fire) or industrial base/tier-2 supply chain visibility?
• Is your primary environment Indo-Pacific maritime/dispersed or European land-congested?

I can tailor the evaluation criteria and technical requirements further.

Yes. If the requirement is AI for battlefield logistics, predictive sustainment, and supply chains that may be disrupted or disconnected, I’d evaluate vendors by layer rather than looking for one “AI defense company.”

The U.S. DoD is explicitly moving toward predictive logistics and a Joint Sustainment Decision Tool, with prototypes being tested by organizations including U.S. Indo-Pacific Command and the Defense Logistics Agency.

My initial vendor shortlist

Two I would put through a particularly serious bake-off

C3 AI vs. Palantir for the decision/data layer.

C3 AI has an unusually direct contested-logistics product. Its published capabilities include global inventory visibility, AI demand forecasting, scenario analysis, supportability analysis and contingency planning; it also says a DoD energy-logistics deployment unified seven disparate systems and provided advance visibility into fuel inventories.

Palantir is the stronger candidate to test if your requirement is broader: logistics + intelligence + operations + maintenance + supply-chain data in one operational data fabric. I'd make them prove the same logistics scenarios rather than accepting generic "AI platform" demonstrations.

Rune + Leidos deserves a separate track because it attacks the sustainment problem from the opposite direction: predict what equipment/parts/resources will be needed before the failure or shortage occurs. Their July 2026 partnership is explicitly aimed at contested/distributed logistics.

The architecture I'd actually evaluate

I'd avoid buying a monolithic "AI logistics system." I'd test a stack:

Data layer → operational digital twin → predictive models → optimization → edge execution → human decision support

For example:

• Supply-chain/digital twin: C3 AI / Palantir
• Predictive sustainment: Rune / Leidos / Lockheed Martin
• Edge compute & disconnected operation: Anduril / AWS
• Optimization specialist: Gallatin AI
• Large-scale integration/operations: GDIT / Amentum / Leidos

That's especially important because contested logistics isn't simply a forecasting problem. The system needs to answer questions like: What happens to the campaign if this port, fuel source, route, depot, satellite link or supplier disappears? What should we move, where, and when? What inventory should be pre-positioned?

C3 AI's scenario-analysis and contingency-planning approach is explicitly designed around that problem, while the newer Celerity architecture emphasizes continuing AI decision-making when cloud connectivity is denied.

What I'd put in the vendor evaluation

Don't let vendors win with a polished dashboard. Give every bidder the same adversarial logistics exercise:

1. Start with a realistic theater supply network.
2. Provide incomplete/noisy inventory and transportation data.
3. Remove several suppliers and transportation routes.
4. Introduce equipment failures and unexpected demand.
5. Degrade communications to simulate disconnected operations.
6. Require the system to generate multiple courses of action.
7. Measure mission availability, fuel/parts stockouts, days of supply, transportation utilization, recovery time and decision latency.
8. Require every AI recommendation to show its underlying data, assumptions and uncertainty.
9. Test what happens when the model is wrong—not just when it's right.
10. Require graceful degradation to deterministic optimization/rules when AI or connectivity fails.

That last point is critical: for battlefield logistics, an AI system that becomes useless when its data feed disappears is not a contested-logistics solution.

If you're evaluating this for an actual defense program, I can also build you a 10–15 vendor longlist with a scored matrix (TRL/fielding, contested-edge capability, predictive logistics, supply-chain optimization, interoperability, security/classification, deployment model, and likely procurement path).

Palantir is the stronger candidate to test if your requirement is broader: logistics + intelligence + operations + maintenance + supply-chain data in one operational data fabric. I'd make them prove the same logistics scenarios rather than accepting generic "AI platform" demonstrations.

Rune + Leidos deserves a separate track because it attacks the sustainment problem from the opposite direction: predict what equipment/parts/resources will be needed before the failure or shortage occurs. Their July 2026 partnership is explicitly aimed at contested/distributed logistics.

The architecture I'd actually evaluate

I'd avoid buying a monolithic "AI logistics system." I'd test a stack:

Data layer → operational digital twin → predictive models → optimization → edge execution → human decision support

For example:

• Supply-chain/digital twin: C3 AI / Palantir
• Predictive sustainment: Rune / Leidos / Lockheed Martin
• Edge compute & disconnected operation: Anduril / AWS
• Optimization specialist: Gallatin AI
• Large-scale integration/operations: GDIT / Amentum / Leidos

That's especially important because contested logistics isn't simply a forecasting problem. The system needs to answer questions like: What happens to the campaign if this port, fuel source, route, depot, satellite link or supplier disappears? What should we move, where, and when? What inventory should be pre-positioned?

C3 AI's scenario-analysis and contingency-planning approach is explicitly designed around that problem, while the newer Celerity architecture emphasizes continuing AI decision-making when cloud connectivity is denied.

What I'd put in the vendor evaluation

Don't let vendors win with a polished dashboard. Give every bidder the same adversarial logistics exercise: