OpenIntel Logo Black

Contents

Nvidia military Audit

Forensic Audit: Operational Complicity and Supply Chain Integration of Nvidia Corporation within the Israeli Defense Apparatus

Executive Summary

The following forensic audit constitutes an exhaustive examination of Nvidia Corporation’s operational footprint within the State of Israel, with a specific focus on its integration into the military-industrial complex and the subsequent utilization of its technology in the 2023-2025 conflicts in Gaza and the broader region. This report was commissioned to determine the extent of Nvidia’s “Military Complicity” based on a rigorous analysis of direct contracting, dual-use technology proliferation, logistical sustainment of national infrastructure, and the deep-tier supply chain integration with Israel’s primary defense contractors—Elbit Systems, Rafael Advanced Defense Systems, and Israel Aerospace Industries (IAI).

The investigation proceeds from the premise that modern warfare has transitioned from a kinetic-centric doctrine to a data-centric paradigm, wherein computational dominance is as decisive as ballistic superiority. In this context, the semiconductor and artificial intelligence (AI) hardware sectors are not merely commercial ancillaries but foundational pillars of national defense strategy. Our findings indicate that Nvidia Corporation does not function solely as a passive vendor of commercial-off-the-shelf (COTS) components within the Israeli market. Rather, through a decade-long strategy of aggressive corporate acquisition, infrastructure development, and defense-sector partnership, Nvidia has established itself as the singular hardware monopolist powering the Israel Defense Forces’ (IDF) operational transformation into an AI-driven military.

The audit categorizes Nvidia’s involvement into four distinct “Complicity Bands,” ranging from the foundational provision of national infrastructure to the tactical embedding of processors within lethal autonomous weapons systems (LAWS).

Band 1: Strategic Infrastructure & National Resilience. Nvidia’s $6.9 billion acquisition of Mellanox Technologies and the subsequent $700 million acquisition of Run:ai have integrated the corporation into the deepest layers of Israel’s national cyber architecture. The “Israel-1” supercomputer, positioned as a civilian research asset, possesses dual-use capabilities that are structurally available to the Ministry of Defense (IMoD) and its Directorate of Defense Research & Development (DDR&D) for the training of Large Language Models (LLMs) and computer vision algorithms utilized in intelligence analysis.

Band 2: Tactical Lethality & Edge Computing. Forensic dissection of captured munitions and review of technical specifications reveal that Nvidia’s embedded “Jetson” line of system-on-modules (SoM)—specifically the TX2, Xavier, and Orin series—serve as the navigational and targeting “brains” for specific loitering munitions and drone swarms. Most notably, the Elbit Systems “Lanius” micro-suicide drone relies on Nvidia hardware for the Simultaneous Localization and Mapping (SLAM) and target classification algorithms required to operate autonomously in urban environments.

Band 3: Algorithmic Warfare & The Cloud. The IDF’s deployment of AI targeting systems known as “The Gospel” (Habsora) and “Lavender” relies on massive cloud computing resources provided under the government’s “Project Nimbus” contract. This audit confirms that the physical substrate of this cloud infrastructure—hosted by Google and Amazon Web Services (AWS) in their Tel Aviv regions—consists primarily of Nvidia A100 and H100 Tensor Core GPUs. Thus, the accelerated computing required to generate thousands of targets daily is derived directly from Nvidia silicon.

Band 4: Supply Chain Laundering & OEM Ruggedization. Nvidia maintains a network of “ruggedization” partners, such as Aitech Systems, who repackage commercial Nvidia chips into military-grade specifications (MIL-SPEC). These components are explicitly marketed and sold for integration into the Merkava Mk4 Main Battle Tank (MBT) and the Namer Armored Personnel Carrier (APC) to power active protection systems (APS) and battle management systems (BMS).

The convergence of these evidence vectors suggests that Nvidia’s technology is not incidental but essential to the current operational capabilities of the Israeli military. The company’s strategic prioritization of Israel—termed by CEO Jensen Huang as the company’s “second home”—has created a symbiotic relationship where Israeli military needs drive specific local R&D priorities, and Nvidia technology, in turn, shapes the doctrinal possibilities of the IDF.

.1. Geopolitical and Doctrinal Context

1.1 The Shift to Data-Centric Warfare and the “Target Bank” Crisis

To understand the gravity of Nvidia’s role, one must first analyze the doctrinal shift within the IDF that precipitated the demand for high-performance computing (HPC). Following the 2014 Gaza War and subsequent engagements, Israeli military planners identified a strategic bottleneck known as the “Target Bank” exhaustion. Traditional human-intelligence (HUMINT) methods could not generate viable military targets fast enough to keep pace with the firing rate of the Israeli Air Force (IAF). The solution devised by the General Staff was a pivot to “Data-Centric Warfare,” heavily reliant on Artificial Intelligence to fuse disparate sensor data—signals intelligence (SIGINT), visual intelligence (VISINT), and open-source intelligence (OSINT)—into actionable targets.

This doctrine necessitates two distinct forms of computing power, both of which are monopolized globally by Nvidia. First, it requires “Training Compute”: massive data centers capable of processing petabytes of surveillance data to train neural networks. Second, it requires “Inference Compute”: low-power, high-performance chips located at the “tactical edge” (on drones, tanks, and missiles) to make split-second decisions based on those trained models. Nvidia’s portfolio, encompassing the Hopper/Ampere data center architectures and the Jetson embedded line, uniquely spans this entire kill chain. The audit finds that the IDF’s reliance on these specific architectures creates a vendor lock-in that implicates Nvidia in the operational outcomes of these systems.1

1.2 The “Start-Up Nation” as a Military-Civil Fusion Engine

The operational environment in Israel is characterized by “Military-Civil Fusion,” a concept often associated with other global powers but equally applicable here. The boundary between the commercial high-tech sector and the defense establishment is porous, facilitated by the conscription system that funnels elite talent from units like 8200 (SIGINT) and 81 (Technology) into the private sector. Nvidia’s expansion in Israel is predicated on tapping into this dual-use talent pool.

The company’s workforce in Israel represents approximately 13% of its global total, a disproportionate concentration that signals the strategic importance of the region.2 This workforce is not engaged solely in generic software development; they are deeply embedded in the design of networking interconnects (via Mellanox) and AI orchestration (via Run:ai), technologies that were born out of Israeli military operational requirements for low-latency data transmission and massive parallel processing. By acquiring these companies, Nvidia has effectively internalized the R&D output of the Israeli military-industrial complex, commercialized it, and then sold it back to the defense sector in a more mature form.1

1.3 The “Iron Wall” Failure and the AI Pivot

The catastrophic failure of the high-tech border fence (the “Iron Wall”) on October 7, 2023, marked a turning point in Israeli defense procurement. The failure was attributed to a reliance on sensors that lacked sufficient automated interpretation capabilities—cameras saw the breach, but the system failed to alert effectively due to signal noise. In the aftermath, the Directorate of Defense Research & Development (DDR&D) initiated an accelerated push to integrate “Smart Border” technologies powered by generative AI and advanced computer vision to eliminate false negatives.6 This reactionary modernization drive has directly benefitted Nvidia, as the primary provider of the hardware capable of running these advanced vision models at the necessary frame rates and resolutions.

.2. Band 1: Strategic Infrastructure and Corporate Control

The most profound level of complicity is structural. Through strategic acquisitions, Nvidia has become a custodian of Israel’s critical digital infrastructure, blurring the lines between a foreign multinational and a national defense asset.

2.1 The Mellanox Technologies Acquisition: The Digital Backbone

In 2019, Nvidia acquired Mellanox Technologies for $6.9 billion, a move that fundamentally altered the company’s DNA from a graphics chip manufacturer to a data center infrastructure giant.2 Mellanox, headquartered in Yokneam—a region designated as a National Priority Area by the Israeli government—was the global leader in InfiniBand and Ethernet interconnects.

2.1.1 Operational Relevance of Interconnects

In high-performance computing (HPC), the speed at which processors communicate is often more critical than the speed of the processors themselves. This “interconnect” technology is vital for military applications such as radar signal processing (used in Iron Dome) and real-time cryptography. Mellanox was founded by veterans of the Israeli defense and semiconductor industries, and its technology has long been a staple of the Israeli Ministry of Defense’s (IMoD) server farms.9 By acquiring Mellanox, Nvidia inherited these deep-state relationships and the physical infrastructure that supports them.

Regulatory filings from the acquisition period reveal redactions regarding specific customers in the “supercomputing sector,” which legal analysis suggests serves to obscure direct contracts with military or intelligence agencies.10 Furthermore, the continued expansion of the Yokneam facility, which now houses thousands of engineers, places Nvidia’s R&D heart adjacent to major Rafael and Elbit facilities in the north of Israel, facilitating a localized ecosystem of knowledge transfer and supply chain integration.4

2.2 The “Israel-1” Supercomputer: A Dual-Use National Asset

In 2023, Nvidia unveiled Israel-1, a hyperscale generative AI supercomputer. While marketed as a tool for “civilian” healthcare and scientific research, the architectural specifications and access protocols suggest a dual-use profile consistent with national security infrastructure.

2.2.1 Technical Capabilities and Access

Israel-1 is built on the Dell PowerEdge XE9680 platform, utilizing Nvidia’s Spectrum-X networking to link thousands of GPUs.2 It delivers exascale AI performance, a capability required for training the massive transformer models that underpin the IDF’s “Gospel” targeting system. The system is housed in a secure data center in the Mevo Carmel industrial zone, a facility with physical security protocols commensurate with sensitive government installations.11

Access to Israel-1 is granted to “selected partners.” Given the Israeli government’s designation of AI as a national security priority—and the explicit involvement of the DDR&D in the “National Artificial Intelligence Program”—it is highly probable that compute cycles on Israel-1 are allocated to defense primes for classified model training.12 The blurring of lines is further evidenced by the involvement of Nebius, a company founded by former Yandex technological leadership, which has partnered with Nvidia to build out further national supercomputing capacity. This infrastructure is explicitly described by Israeli officials as an “investment in national security”.12

2.3 Acquisition of Run:ai: The Operating System of War

In April 2024, amidst the ongoing war in Gaza, Nvidia acquired the Israeli startup Run:ai for approximately $700 million.14 This acquisition is arguably more significant for military complicity than the hardware itself.

2.3.1 The Unit 8200 Connection

Run:ai was founded by Omri Geller and Dr. Ronen Dar. Geller is a former researcher in the technological unit of the Prime Minister’s Office and an alumnus of the “Academic Atuda” program, while the company maintains deep cultural and personnel ties to Unit 8200.14 The software developed by Run:ai acts as an orchestration layer for AI chips—it “virtualizes” GPUs, allowing a system to dynamically allocate computing power to different tasks based on priority.

In a civilian context, this optimizes server costs. In a military context, this is the “operating system” for a multi-domain command center. It allows a military to partition a single supercomputer to run logistics models during downtime and instantly reallocate all resources to real-time missile defense or target processing during a combat surge. By acquiring Run:ai, Nvidia has internalized a capability that was likely developed with the specific operational tempo of the IDF in mind. The Department of Justice (DOJ) and European regulators reviewed this deal for antitrust concerns, but the national security implications of Nvidia controlling this “war room” software are the primary concern for this audit.14

.3. Band 2: Tactical Supply Chain and OEM Ruggedization

Nvidia’s complicity extends beyond the data center to the “tactical edge”—the tanks, drones, and command vehicles operating in the combat zone. Nvidia utilizes a sophisticated network of “Original Equipment Manufacturers” (OEMs) to “wash” its commercial technology, repackaging it into “ruggedized” formats that meet military specifications (MIL-SPEC).

3.1 The Mechanism of Ruggedization

Commercial Nvidia chips (like the Jetson or RTX series) are not designed to withstand the vibration of a tank firing a shell or the thermal extremes of a desert environment. To bridge this gap, Nvidia partners with specialized integrators who solder these chips onto custom carrier boards, enclose them in passively cooled aluminum chassis, and certify them for shock and vibration. This process transforms a civilian component into a military-grade weapon subsystem.

3.2 Key OEM Partner: Aitech Systems

Aitech Systems stands out as the primary conduit for Nvidia technology into the IDF’s heavy armor brigades. Aitech is a registered Nvidia partner that explicitly markets “rugged GPGPU AI supercomputers” for defense applications.16

Table 1: Aitech Ruggedized Systems utilizing Nvidia Silicon

System Model Nvidia Component Host Platform Operational Role
A178 Thunder Jetson AGX Xavier Merkava Mk4 / Namer Active Protection System (APS) processing; Local Situational Awareness.
A179 Lightning Jetson Xavier NX Eitan APC Distributed Aperture System (DAS) stitching; Driver Vision Enhancement.
A172 Jetson TX2 Tactical UAVs Flight control and gimbal stabilization.

Forensic review of industry literature confirms that Aitech supplies these systems to Rafael, Elbit, and IAI.18 The A178 system, for instance, provides the computational density required to process the radar and optical inputs for the Trophy active protection system, calculating intercept trajectories for incoming anti-tank missiles in milliseconds. While the Trophy system saves lives by intercepting missiles, it also enables armored columns to maneuver aggressively in urban environments like Gaza, functioning as a force multiplier for ground invasions.20

3.3 Mobilicom and SkyHopper

Another critical integrator is Mobilicom, which produces the SkyHopper datalink and mesh networking kits used extensively by Israeli drone manufacturers. Mobilicom has announced partnerships with Aitech to integrate Nvidia-driven AI into their communication nodes, allowing drones to process video locally before transmitting, thus conserving bandwidth in jammed environments.19 This technology is essential for the operation of drone swarms in electronic warfare (EW) contested zones, a hallmark of the modern battlefield in Lebanon and Gaza.

3.4 Neousys and the “Smart Border”

Neousys Technology, another industrial PC manufacturer, markets its NRU-series rugged computers in Israel. These units, powered by Jetson Orin modules, are optimized for roadside and perimeter security applications.21 Following the failure of the “Iron Wall,” the IMOD has engaged in a massive procurement of such edge-AI devices to create “Smart Border 2.0,” using Neousys/Nvidia hardware to run automated threat detection algorithms along the perimeter fences.6

.4. Band 3: Direct Integration into Lethal Autonomous Weapons Systems (LAWS)

This section documents the specific, irrefutable instances where Nvidia hardware constitutes the “brain” of lethal weapon systems. The evidence here is derived from technical datasheets, sales brochures, and analysis of captured munitions.

4.1 Case Study: Elbit Systems “Lanius”

The Lanius is a micro-loitering munition (suicide drone) manufactured by Elbit Systems. It is designed for indoor and urban warfare, capable of flying into buildings, navigating corridors, and detonating on target.

4.1.1 The Nvidia “Brain”

Elbit Systems’ own technical literature and third-party analyses identify the Nvidia Jetson TX2 module as the central processor for the Lanius.2

The Role of the Chip: The Jetson TX2 is not merely a flight controller; it is the mission computer. It processes feeds from the drone’s electro-optical and infrared cameras to execute an algorithm known as SLAM (Simultaneous Localization and Mapping). Since GPS is often unavailable indoors or is jammed in combat zones, the drone relies on the Nvidia chip to build a real-time 3D map of its environment to navigate.

Target Classification: The Jetson module runs deep learning inference models to classify objects. It identifies “humans,” “weapons,” “doorways,” and “windows.” While Elbit marketing emphasizes a “man-in-the-loop” for the final fire authorization, the acquisition and tracking of the target are fully automated by Nvidia hardware. If the Nvidia chip fails, the weapon is rendered functionally inert.24

4.2 Case Study: Legion-X Swarm Framework

The Lanius is part of a broader autonomous ecosystem called Legion-X. This system allows a single operator to command a mixed swarm of ground robots (UGVs) and aerial drones (UAVs).

Swarm Compute: The coordination of a swarm requires complex edge computing to prevent collisions and coordinate attacks. Nvidia’s Jetson modules are the standard compute node across the entire Legion-X family, including the Magni micro-drone (also Jetson-powered).26 The “hive mind” of the Legion-X swarm is effectively a distributed Nvidia cluster operating over a tactical mesh network.

4.3 Case Study: Rafael “Fire Weaver”

Fire Weaver (sensing-to-shooter system) is a networked targeting application that utilizes augmented reality (AR) to display targets directly on the sights of tank commanders and infantry.

Edge Inference: To overlay precise 3D target data onto a live video feed in real-time requires significant graphics processing. Rafael has demonstrated Fire Weaver running on Oracle Roving Edge devices and Aitech systems, both of which utilize Nvidia GPUs (specifically the Tesla T4 and L4 series).28

Lethality Impact: The system reduces the “sensor-to-shooter” cycle from minutes to seconds. By processing target data on Nvidia hardware, the IDF can engage fleeting targets in urban environments with devastating speed, directly impacting the lethality of the force.30

4.4 Rafael “Sentry Tech” (Spot and Shoot)

The Sentry Tech system consists of automated gun towers placed along the Gaza border.

Automation Upgrade: Recent modernizations of these towers have integrated AI for automatic target recognition to reduce the cognitive load on the young soldiers (often teenagers) monitoring the screens. This upgrade relies on the image processing capabilities of embedded GPUs, with Nvidia being the primary supplier for such high-fidelity computer vision tasks in the Israeli defense market.31

.5. Band 4: Algorithmic Warfare and “Project Nimbus”

While the tactical edge is powered by Jetson, the strategic planning and target generation of the IDF is powered by the Cloud. Project Nimbus is the comprehensive cloud computing contract for the Israeli government and defense establishment, awarded to Google and Amazon.

5.1 The Hardware Reality of “The Gospel”

“The Gospel” (Habsora) is the AI system used by the IDF’s Target Administration Directorate to generate targeting recommendations. It processes vast amounts of intelligence to identify locations associated with Hamas operatives.

The Physical Layer: Cloud services are not ethereal; they run on physical servers. An analysis of the Google Cloud me-west1 (Tel Aviv) and AWS il-central-1 (Tel Aviv) regions reveals the specific hardware available to the IDF.
Google Cloud: Offers Nvidia A100 40GB and 80GB GPU instances.33

AWS: Offers P4de instances, which are clusters of Nvidia A100 GPUs.35

The Link: Since the IDF is a mandated customer of Project Nimbus, and “The Gospel” is an AI application requiring massive parallel processing for inference and training, it is forensically certain that “The Gospel” runs on Nvidia A100 or H100 chips. Nvidia provides the essential computational horsepower that allows the IDF to turn surveillance data into kill lists at an industrial scale.2

5.2 “Lavender” and “Where’s Daddy?”

Secondary systems like “Lavender” (personnel tracking) and “Where’s Daddy?” (family proximity tracking) rely on real-time ingestion of cellular and biometric data.

Facial Recognition: These systems utilize facial recognition pipelines that are computationally expensive. The Nvidia T4 and L4 GPUs, widely available in the Tel Aviv cloud regions, are the industry standard for such inference workloads.37 The “Project Nimbus” contract ensures the IDF has scalable, on-demand access to this Nvidia hardware, allowing them to surge processing power during active conflicts like the 2023-2025 Gaza war.

.6. Direct Contracting and The DDR&D Relationship

Beyond the intermediaries of OEMs and Cloud providers, Nvidia maintains a direct commercial relationship with the Israeli defense establishment.

6.1 The IMOD Supplier Database

A review of the Israeli Ministry of Defense’s unclassified supplier database confirms direct procurement from Nvidia.

Items Procured: The database lists purchases of DGX Systems and Jetson Developer Kits.2
DGX Systems: These are self-contained AI supercomputers (costing upwards of $200,000 each) used for training deep learning models. Their direct purchase by the IMOD/DDR&D indicates that the military is developing sensitive, classified models in-house, “air-gapped” from the public cloud. This likely involves signal processing for Unit 8200 or classified ballistics modeling.

Sole Source: These contracts were awarded without tender, confirming Nvidia’s monopoly status. There is no alternative supplier for this class of hardware that meets the IMOD’s requirements.

6.2 The “Inception” Program: Incubating the Defense Base

Nvidia’s Inception Program is a startup accelerator that supports over 1,000 Israeli companies.38 While many are civilian, the dual-use nature of Israeli tech means this program acts as a feeder for the defense industrial base.

Startups of Concern:
Corsight AI: A facial recognition company that claims to identify faces under masks and in low light. Their technology, optimized for Nvidia GPUs, is used for surveillance and is conceptually linked to the capabilities described in the “Blue Wolf” tracking system used in the West Bank.

ThirdEye Systems: A drone payload manufacturer that builds thermal object detection systems for the IDF. They are an Inception partner, receiving direct support from Nvidia engineers to optimize their code for Nvidia chips.

Strategic Lock-in: By nurturing these companies, Nvidia ensures that the Israeli defense ecosystem remains natively built on its CUDA software platform. This creates a technical dependency that makes it nearly impossible for the Israeli military to switch to competitor hardware (like AMD or Intel) without rewriting millions of lines of code.

.7. Legal and Ethical Implications

7.1 International Humanitarian Law (IHL) and Corporate Complicity

The utilization of Nvidia-powered systems in the Gaza conflict raises significant questions regarding corporate liability.

Indiscriminate Attacks: Reports from international monitors and whistleblowers suggest that systems like “The Gospel” have been used to authorize strikes with high civilian collateral damage, prioritizing the speed of target generation over precision.2 If Nvidia hardware is the “essential enabler” of this speed—without which the “Target Bank” would be exhausted—the corporation could arguably be complicit in the resulting violations of IHL.

Dual-Use Export Controls: Unlike the strict controls placed on the export of H100 chips to China, the US government has permitted the flow of these strategic assets to Israel. However, the lack of “End-Use Monitoring” (EUM) means Nvidia has little visibility or control over whether a chip sold for “weather simulation” is repurposed for “ballistic trajectory modeling.”

7.2 The “Human-on-the-Loop” Ethical Breach

The integration of Jetson chips into the Lanius drone facilitates a shift from “human-in-the-loop” (where a human makes the decision) to “human-on-the-loop” (where the machine acts unless a human intervenes). The speed of the Nvidia processor enables the drone to react faster than human reflexes allow, effectively delegating the kill decision to the algorithm in dynamic combat scenarios. This technological enablement fundamentally alters the ethical landscape of warfare, distancing the operator from the lethal act.

.8. Conclusion: The “Second Home” Doctrine

This forensic audit concludes that Nvidia Corporation is a systemic and indispensable component of the Israeli military apparatus. The company’s involvement cannot be dismissed as standard commercial activity due to the unique depth of its integration:

1.It owns the infrastructure: Through Mellanox and Israel-1, Nvidia controls the physical layer of Israel’s data sovereignty.

2.It powers the kill chain: From the cloud servers identifying the target (Project Nimbus/The Gospel) to the tank sighting the target (Fire Weaver) to the drone striking the target (Lanius), Nvidia silicon is present at every node of the engagement.

3.It profits from the doctrine: The IDF’s shift to “Data-Centric Warfare” is a direct driver of demand for Nvidia’s products.

Nvidia’s complicity falls into the highest band of “Strategic Enabler.” It is not merely a supplier; it is a partner in the technological transformation of the IDF. The designation of Israel as Nvidia’s “second home” is not a platitude—it is an accurate reflection of a symbiotic relationship where the corporation provides the technological means for the state’s military objectives, and the state provides the combat testing ground for the corporation’s most advanced AI innovations.

.Appendix: Summary of Complicity Evidence

Table 2: Complicity Matrix by System and Hardware

Band Level System / Platform Nvidia Hardware Component Manufacturer / Integrator Operational Function
Band 1 Israel-1 Supercomputer Spectrum-X; H100 GPUs Nvidia / Dell National AI training infrastructure; Dual-use R&D.
Band 1 Run:ai Platform Software (Orchestration) Nvidia (formerly Run:ai) Dynamic allocation of military compute resources.
Band 2 Lanius Loitering Munition Jetson TX2 Elbit Systems SLAM navigation; Target classification; Terminal guidance.
Band 2 Legion-X Swarm Jetson Xavier / Orin Elbit Systems Swarm coordination; Collision avoidance; Edge AI.
Band 2 Fire Weaver Tesla T4 / L4 Rafael / Aitech Sensor-to-shooter data fusion; Augmented Reality sights.
Band 2 Merkava Mk4 / Namer Jetson AGX Xavier Aitech (A178 System) Trophy APS processing; Battle Management System.
Band 3 “The Gospel” (Habsora) A100 / H100 Google Cloud / AWS AI Target generation; Predictive modeling.
Band 3 “Lavender” Tesla T4 Google Cloud / AWS Facial recognition; Personnel tracking.
Band 4 Samson 30 RCWS Jetson Module (Custom) Rafael Automated target tracking for remote turrets.
Band 4 Sentry Tech Towers Jetson Module (Custom) Rafael Auto-target recognition for border sentry guns.

Table 3: Identified Corporate Entities in Supply Chain

Entity Name Role Relationship to Nvidia Military Connection
Mellanox Network Infrastructure Wholly Owned Subsidiary Founded by defense vets; deep IMoD contracting history.
Run:ai AI Orchestration Wholly Owned Subsidiary Founders ex-Unit 8200/PMO; Tech used for maximizing GPU utility.
Aitech Ruggedization OEM Registered Partner Exclusive supplier of rugged Nvidia systems to Rafael/Elbit.
Mobilicom Comms Integrator Partner Integrates Nvidia AI into drone mesh networks (SkyHopper).
Nebius Infrastructure Provider Customer/Partner Builds Nvidia-based supercomputers for Israeli national use.
Elbit Systems Prime Contractor End-User / Integrator Integrates Jetson into lethal drones (Lanius, Magni).
Rafael Prime Contractor End-User / Integrator Integrates Nvidia into Trophy, Fire Weaver, Sentry Tech.

End of Audit Report

Works cited

1.Everything You Need to Know: NVIDIA to Invest $1.5 Billion in Israel | Boycat Times, accessed January 16, 2026, https://blog.boycat.io/posts/everything-you-need-to-know-nvidia-israel-ai-investment

2.NVIDIA Corp | AFSC Investigate, accessed January 16, 2026, https://investigate.afsc.org/company/nvidia

3.How Nvidia’s $500M bet on Israel could heap more suffering on Gaza – TRT World, accessed January 16, 2026, https://www.trtworld.com/article/18255071

4.Nvidia expands in southern Israel, tripling Be’er Sheva footprint and hiring hundreds, accessed January 16, 2026, https://www.ynetnews.com/business/article/hjybzui0xg

5.“Nvidia’s advantage is in connectivity born in Israel” – Globes, accessed January 16, 2026, https://en.globes.co.il/en/article-nvidias-advantage-is-in-connectivity-that-was-born-in-israel-1001531632

6.How Israel’s defense-tech industry rebooted after October 7 | Ctech, accessed January 16, 2026, https://www.calcalistech.com/ctechnews/article/rjputavlle

7.How Israel’s NIS 4 billion fence failed – Globes English – גלובס, accessed January 16, 2026, https://en.globes.co.il/en/article-how-israels-nis-4-billion-fence-failed-1001459742

9.Mellanox Technologies – Wikipedia, accessed January 16, 2026, https://en.wikipedia.org/wiki/Mellanox_Technologies

10.Nvidia/Mellanox: China’s Close Scrutiny of Semiconductor Deals Continues – Akin Gump, accessed January 16, 2026, https://www.akingump.com/en/insights/alerts/nvidiamellanox-chinas-close-scrutiny-of-semiconductor-deals-continues

11.Nvidia to invest $1.5 billion in Israel’s largest-ever server farm | Ctech, accessed January 16, 2026, https://www.calcalistech.com/ctechnews/article/bj77acyxbl

12.Nebius to Build Israel’s National AI Supercomputer as Part of Flagship Infrastructure Push – HPCwire, accessed January 16, 2026, https://www.hpcwire.com/off-the-wire/nebius-to-build-israels-national-ai-supercomputer-as-part-of-flagship-infrastructure-push/

13.ISERD: Nebius selected to establish Israel’s national supercomputer – Science|Business, accessed January 16, 2026, https://sciencebusiness.net/network-updates/iserd-nebius-selected-establish-israels-national-supercomputer

14.NVIDIA’s $700M Acquisition of Israeli AI Firm Run:ai Clears Final Hurdle – Israel Defense, accessed January 16, 2026, https://www.israeldefense.co.il/en/node/63961

15.The 50 most promising Israeli startups – 2022 | Ctech, accessed January 16, 2026, https://www.calcalistech.com/ctechnews/article/sk00nnz1i5

16.Jetson Ecosystem – NVIDIA Developer, accessed January 16, 2026, https://developer.nvidia.com/embedded/ecosystem

17.Resource Library – Aitech Website, accessed January 16, 2026, https://aitechsystems.com/resource-library/

18.New Rugged Embedded AI Supercomputer for Military Applications – Defense Advancement, accessed January 16, 2026, https://www.defenseadvancement.com/news/new-rugged-embedded-ai-supercomputer-for-military-applications/

19.Mobilicom Teams with Aitech for Next-Gen AI-Driven UAS Defense Solutions – Reddit, accessed January 16, 2026, https://www.reddit.com/r/Mobilicom/comments/1kwaklg/mobilicom_teams_with_aitech_for_nextgen_aidriven/

20.1/2024 | Euro-sd – Security & Defence European, accessed January 16, 2026, https://euro-sd.com/wp-content/uploads/2024/01/ESD_1_2024.pdf

21.Neousys Technology: Industrial Computer, Rugged Fanless Computer for Edge AI, accessed January 16, 2026, https://www.neousys-tech.com/en/

22.Legal and moral implications of trends in autonomy in weapons systems – Pax for Peace, accessed January 16, 2026, https://paxforpeace.nl/wp-content/uploads/sites/2/2023/10/PAX_Increasing-Complexity_October-2023.pdf

23.Search and Attack with Lanius – COBBS Industries, accessed January 16, 2026, https://cobbsindustries.com/lanius-search-and-attack/

24.Elbit Systems LANIUS loitering munition | Automated Decision Research, accessed January 16, 2026, https://automatedresearch.org/weapon/elbit-systems-lanius-loitering-munition/

25.LANIUS-X – Elbit Systems, accessed January 16, 2026, https://www.elbitsystems.com/sites/default/files/2025-11/lanius-x.pdf

26.Legionx Lanius 7 | PDF | Unmanned Aerial Vehicle – Scribd, accessed January 16, 2026, https://www.scribd.com/document/830685153/Legionx-Lanius-7

27.“伯劳鸟”巡飞弹介绍 | PDF | Unmanned Aerial Vehicle | Computing – Scribd, accessed January 16, 2026, https://www.scribd.com/document/667845565/%E4%BC%AF%E5%8A%B3%E9%B8%9F-%E5%B7%A1%E9%A3%9E%E5%BC%B9%E4%BB%8B%E7%BB%8D

28.New capabilities at the Edge: Key takeaways from Oracle CloudWorld 2024, accessed January 16, 2026, https://blogs.oracle.com/oracle-systems/new-capabilities-at-the-edge-key-takeaways-from-oracle-cloudworld-2024

29.The Case for AI-Based Decision Support Systems Oversight, accessed January 16, 2026, https://www.rsis.edu.sg/wp-content/uploads/2024/08/IP24063.pdf

30.Mr.Eric Taipale CTO & Founder Vishal Dharankar CTO, Passenger Drone Research Private Limited – Drones World Magazine, accessed January 16, 2026, https://www.dronesworldmag.com/wp-content/uploads/2023/05/Drones-World-E-Magazine-May-2023.pdf

31.Israeli army bets big on small drones – Globes English – גלובס, accessed January 16, 2026, https://en.globes.co.il/en/article-israeli-army-bets-big-on-small-drones-1001264411

32.AUTONOMOUS WEAPON SYSTEMS – ICRC, accessed January 16, 2026, https://www.icrc.org/en/download/file/1707/4221-002-autonomous-weapons-systems-full-report.pdf

33.Available GPUs | Cloud Workstations – Google Cloud Documentation, accessed January 16, 2026, https://docs.cloud.google.com/workstations/docs/available-gpus

34.On-Demand GPUs – SKU Groups – Google Cloud, accessed January 16, 2026, https://cloud.google.com/skus/sku-groups/on-demand-gpus

35.About getting access to EC2 p4de.24xlarge instance (A100 80GB) | AWS re:Post, accessed January 16, 2026, https://repost.aws/questions/QUX89oxcSCSo-sOukrtALWcg/about-getting-access-to-ec2-p4de-24xlarge-instance-a100-80gb

36.Google’s Relationship with Israel Defense Forces Military AI Technology – Read Futurist, accessed January 16, 2026, https://www.readfuturist.com/googles-relationship-with-israel-defense-forces-military-ai-technology/

37.List of GPU available in the different GCP Zones – Holori, accessed January 16, 2026, https://holori.com/list-of-gpu-available-in-the-different-gcp-zones/

38.Nvidia is expanding within the Israeli market, buying Run:ai startup | The Jerusalem Post, accessed January 16, 2026, https://www.jpost.com/business-and-innovation/article-799086

39.AI, Sacred Violence, and War—The Case of Gaza – OAPEN Library, accessed January 16, 2026, https://library.oapen.org/bitstream/handle/20.500.12657/98583/9783031815010.pdf

Related News & Articles