Of the 50 companies in the Israeli Defence-AI Deployment Index, 13 have a named public source reporting that their system has been used by a military in actual operations. The other 37 do not. That distribution is the finding of this index, not a gap in its research, and it is the reason the index exists: defence AI deployment evidence is far scarcer than the sector's public presence suggests.
Existing landscape maps of Israeli defence technology count between 60 and 334 companies. They are organised by product category and annotated with funding stage. That is an investor's frame. It tells a reader that a company closed a Series B. It does not tell a procurement officer whether the product has left the demonstration tent.
The question the index was built to answer
A procurement officer evaluating Israeli suppliers asks one thing that directories do not answer: has this actually been fielded, and who says so?
The index applies a single filter. For each company, what is the strongest piece of publicly published evidence that its system has been procured or used, and where can a reader open that evidence themselves. Every grade carries a link. Where the only available source is the company's own press release, the entry says so.
Four grades result. L3 · Operational means a named public source reports the system has been used by a military in actual operations; training exercises do not qualify. L2 · Contract means a publicly announced procurement contract or order exists, but no reported operational use. L1 · Trial covers public trials, pilots, military evaluations or exhibition demonstrations only. L0 · Claimed means the company claims a defence application and no independent public source corroborates it. The full grading rules are published alongside the index.
How the 50 distribute
| Grade | Companies | Share |
|---|---|---|
| L3 · Operational | 13 | 26% |
| L2 · Contract | 22 | 44% |
| L1 · Trial | 6 | 12% |
| L0 · Claimed | 9 | 18% |
The largest band is L2. Forty-four per cent of the field has a publicly announced contract or order and nothing published about the system in use. That is not a criticism of those companies. It is a description of what defence contracting publishes and what it withholds.
The 13 companies at L3 are the exception, and their sourcing is instructive. Globes reported the IDF using Axon Vision's AI against drones. Janes reported the IDF using UVision's mini loitering munition. Breaking Defense reported the US Army expanding its use of SMARTSHOOTER's fire-control system. In each case an identifiable outlet, not the vendor, described operational use. The complete list of the 13 sets out each source.
What the L2 band actually tells a buyer
An L2 grade often means, in practice, "a defence ministry somewhere signed something." Contract announcements routinely omit the customer, the quantity and the configuration. A press release describing a "multimillion-dollar order from a European NATO member" is a real fact, verifiable and dated, and it still leaves a buyer unable to identify who bought it or what they did with it afterwards.
This is a limit of public sourcing rather than an oversight in the research, and it is treated in more depth in the undisclosed-customer problem. The practical consequence for a buyer is that L2 establishes commercial traction and nothing about performance.
What L0 does and does not mean
Nine companies sit at L0. The grade records that no independent public source corroborates a claimed defence application. It does not record that the claim is false, that the technology does not work, or that the company is overstating itself.
Several L0 entries are substantial businesses with real products in adjacent markets. Bluewhite builds autonomous agricultural machinery. Dream Security is a cyber company reported by the Times of Israel at a $3 billion valuation. What is missing in each case is a published, independent account of defence use. A single credible report of a military order would move any of them immediately, which is the point of grading rather than ranking.
Where the field is crowded and where it is empty
Counter-UAS is the densest category in the index, with eleven entries spread across every grade. ISR and sensor fusion, autonomous air systems and simulation and training follow. The concentration is legible: these are the areas where Israeli operational experience, export demand and available capital overlap.
Three categories came back almost bare. Active protection, targeting and fire control, and military logistics and sustainment each yield exactly one qualifying company under the index's inclusion test. Israeli AI in those areas remains a primes' business. The index cites the UK's Project THESEUS sustainment programme, where the awards went to Rafael and to Israel Aerospace Industries through a UK partner rather than to startups. Because the index excludes the large primes by design, the emptiness of those categories is a finding about where startups have not gone, not a claim that Israel lacks the capability.
Limits of this reading
Fifty entries is a working index, not a census. The sector holds several hundred companies. The sample is weighted toward firms with English-language coverage, which biases against smaller companies publishing only in Hebrew, and therefore probably understates the L2 and L3 bands rather than overstating them.
The index contains no technical performance data and is not a capability assessment. It records what has been published, by whom, and when. A company at L0 may hold a classified contract. A company at L3 may have been reported inaccurately. Neither possibility is resolvable from public sources, and the index does not pretend otherwise.
Frequently asked questions
How many Israeli defence-AI companies have documented operational military use?
Thirteen of the 50 companies in the index have a named public source reporting that a military used the system in actual operations. Training exercises and demonstrations do not qualify for that grade. Each of the 13 entries links to the specific report.
Does a low grade mean the technology does not work?
No. The grades record the strength of publicly available evidence, not capability. An L0 grade means no independent public source corroborates a defence application. Classified contracts, undisclosed customers and Hebrew-only coverage all produce low grades for capable companies.
Why are Elbit, Rafael and IAI not in the index?
The index covers companies whose core product is AI or autonomy applied to defence. For the large Israeli primes, AI is a component across a broad portfolio rather than the product itself, so they fall outside the inclusion test.
Sources
- idf.ai, The Israeli Defense-AI Deployment Index v1.0, 24 August 2026
- Globes, IDF using Axon Vision's AI to combat drones
- Janes, IDF using UVision mini loitering munition
- Breaking Defense, Israel's SMARTSHOOTER says US Army expanding use of high-tech fire control system, June 2025
- The Times of Israel, Israeli cyber unicorn valued at $3 billion aims to help governments own and control their AI
Independent publication of idf.ai. Not affiliated with, endorsed by, or connected to the Israel Defense Forces, the Israeli Ministry of Defense, or any government body. Compiled entirely from publicly published sources. No classified, restricted or non-public information. Listed companies may dispute any entry: send the published source that contradicts it and the entry will be amended or removed.