Start a trial

Operational notes Partnerships

Thales and Destinus: flipping the cost of shooting down a drone

7 min read

Close-up of a stack of thin metal plates piled one on top of another against a plain grey background
Layered defence only holds if each layer meets the one beneath it: that is where an industrial agreement becomes architecture.

On 20 July 2026 Thales (€22.1 billion in 2025 sales) announced from Paris a cooperation agreement with Destinus, a five-year-old maker of counter-drone interceptors and long-range effectors. The release says the agreement was signed “at Eurosatory 2026”, a show that closed on 19 June: more than a month passed between signature and announcement, with no explanation.

Who Destinus is

Founded in March 2021 in Payerne, Switzerland, by Mikhail Kokorich — a Russian-born physicist, founder of the space company Momentus — Destinus moved from hydrogen hypersonic aircraft to weapons after the invasion of Ukraine, and in November 2024 shifted its registered office to Hengelo, in the Netherlands, over Swiss arms-export rules. It reports around 750 people in four countries, close to €400 million raised; in May 2026 it opened a €200 million pre-IPO round, targeting over €5 billion and the Amsterdam exchange. Not the first large group to take a slice of it: on 13 April 2026 it announced with Rheinmetall the Rheinmetall Destinus Strike Systems JV for cruise missiles and rocket artillery — 51% to the German group, 49% to Destinus, formation due in the second half of 2026, subject to approvals.

On the founder, both versions belong in the record. In July 2021 the SEC sued Kokorich over misleading statements about Momentus’s technology, about the national-security risks associated with him and about his own immigration status; on 25 November 2024 a US federal court entered a final consent judgment — a $2 million civil penalty and a five-year bar from serving as an officer or director of a public company — with no admission or denial. On the other side: Russian citizenship renounced in January 2024 in disagreement with Moscow, systems supplied to Ukraine.

The technology: the ledger before the ballistics

The problem the Hornet promises to solve is economic, not technical: a single-use attack drone costs tens of thousands of euros, the missile that downs it hundreds of thousands, and whoever attacks in mass wins on the ledger even losing everything. The answer is an intermediate layer: not a missile, but a small fixed-wing interceptor from a container. For the Hornet Block 1 Destinus states a range of “75+ km”, a 1.5 kg payload, initial radar guidance and an AI-assisted terminal seeker on electro-optical and infrared sensors, engaging even with GPS jammed.

What it does not publish matters as much: no speed, endurance, engagement probability or cost per round — and without cost per round the Hornet that “reverses the cost asymmetry” stays a sales argument. There are no performance figures verified by a third party; there is a demonstration run with the Spanish Navy: the launch of a Hornet Block 1 from a container fitted on the frigate F-81 Santa María, on 18 June 2026.

The known terms: what the release says, and what it does not

The vocabulary should be read literally: the parties have “signed a cooperation agreement” and “identified five areas”. They are: key subsystems supplied by Thales, “such as for example military warheads, aligned with NATO standards”; joint production of the Hornet at a site qualified for pyrotechnical operations; interoperability with Thales’s radars and counter-UAS systems; long-range effectors deployable from tactical transport aircraft; ground-based interception for future NATO integrated air and missile defence architectures. Areas identified, not orders: no number appears in the text.

Almost nothing else is public: value, duration, exclusivity, where the industrial site is, whether there is a launch customer, how intellectual property is split on the jointly built Hornet, whether Thales takes a stake. One quotation appears, from Destinus CTO Tim Moser: “Layered air defence only works when systems interoperate by design, not as an afterthought”, with engagement decisions staying “with the operator”. And the release speaks of “reinforcing their collaboration”: an earlier collaboration neither company has ever described.

What it means

Destinus gets three things venture capital cannot buy: warheads, the final link in the chain and a European bottleneck; a site qualified for pyrotechnical work; integration with radars already in service in Europe. The price is structural: a minority shareholder in a JV controlled by Rheinmetall and a supplier-partner of Thales, it is unclear how much is still a company and how much a portfolio of other people’s stakes. Thales gets a fast, cheap effector layer beneath its own radars without building it. In Europe the buyer of defence start-ups is no longer only American: it is the national champions — the same logic as the Anduril and Rheinmetall alliance, scaled back a year after launch.

Drones, counter-drone and integrated air defence are among the first five common defence projects proposed by the Commission on 3 July 2026: the money exists. For Italy the link is close: Thales is one of three shareholders of Eurosam, with MBDA France and MBDA Italy, the consortium behind the SAMP/T NG the Italian Army has begun receiving; the counter-drone layer assembled with Destinus sits beneath a system Italy already fields. Yet no Italian company is in this deal: whoever buys for defence will find these products in tenders, not in the national supply chain.

There are three risks, contractual before technological. Lock-in on consumables: an interceptor is a munition, and whoever integrates the launcher with a radar is tied for years to the reload supplier. Dual-use and export: the same family protects a military base and a civilian airport, but the regime changes — warheads and interceptors sit on the EU Common Military List and, in Italy, under Law 185/1990. Governance of the emerging supplier: a registered office moved between countries, a founder subject to a US order on corporate roles, a listing in preparation; what effect that bar has in Amsterdam is a question for the lawyers, not for a press release.

The operational lesson

  1. “Areas of cooperation identified” is not an order: ask for units, timing, unit price, penalties.
  2. On consumables the constraint is the reload, not the purchase: in the tender, a second source and portability of the radar integration.
  3. Demand independent evidence — engagement probability and cost per round — not the brochure range.
  4. Check the emerging supplier’s governance — ownership, jurisdiction of the registered office, orders against directors — on a re-checking cadence.
  5. On the AI on board, put it in the contract: who decides engagement, which logs are kept, who reconstructs the decision. The AI Act (Regulation (EU) 2024/1689) does not apply to exclusively military, defence or national security use (Article 2(3)): the guarantee does not come from the rule.

An announcement describes an intention, a contract creates an obligation, and the buyer pays for the difference. For an interceptor as for an AI system, you must be able to reconstruct how a decision was made, on infrastructure you control. That is the criterion behind every architecture we build to stay verifiable: dedicated, closed models, disconnected from the open web, on-premise at the client or on a dedicated cloud for the single client, with a dedicated VPN and a data centre in Italy, in premises we staff.

Do you have to assess a supplier arriving with “areas of cooperation” with a large group, or an “AI-assisted” function with no logs? Let’s talk in thirty minutes.

Sources