Array Labs and Mitsubishi Electric: the Asian channel is there, the exclusivity is not

On 24 July 2026 Mitsubishi Electric invested $10 million in Array Labs, a Californian startup that builds satellite radar; on 28 July it made the investment public, the same day Array Labs announced it had closed, with that investment as its anchor, an oversubscribed round totalling $21 million. The two companies state their aim is to jointly commercialise a space-based maritime and aircraft tracking service for defence customers in the Asia-Pacific. It is a useful case because it shows a pattern now common in defence: the large group does not buy the startup, it takes a stake, and becomes at once its customer and its channel into an entire regional market.

Who Array Labs is

Founded in 2022 — Y Combinator’s Summer 2022 batch — by Andrew Peterson, an aerospace engineer with a background in satellite constellation design and the Vera Rubin Observatory, and Jose Isaac Robledo, with a finance career that includes Morgan Stanley and JLL, Array Labs is based in Redwood City, California. Around thirty people, for a company that two years ago had not yet put anything into orbit.

Its first capital — a $5 million seed round in October 2022 — funded the prototype; a $20 million Series A in January 2026 funded radar production ahead of launch. The 28 July round, led by Mitsubishi Electric alongside Catapult Ventures, Kompas and Y Combinator Management, brings — the release states — total funding raised to $41 million. Ahead of any publicly named commercial customer, the company states it already holds contracts with DARPA, the US Navy and the US Air Force, alongside backing from SOCOM: from day one, the target market has been defence.

The technology: a radar that does not work alone

A conventional radar satellite transmits and receives from the same spacecraft: it illuminates a point on the ground and reads the echo back, one vantage point at a time, one pass every few days. Array Labs puts several satellites into orbit that observe the same point at the same instant, flying in formation: while one transmits, the others receive the same echo from different angles — a setup radar engineers call a multistatic configuration. The advantage is not only redundancy, if one satellite fails the others keep watching, but geometry: several simultaneous vantage points on the same target make it possible to compute both position and velocity without waiting for the same satellite to pass overhead again days later.

This is the capability defence circles call AMTI, Airborne Moving Target Indication: tracking a moving object from orbit — an aircraft, a ship — day or night, through cloud cover, because a radar does not depend on light or clear skies. From the same data, Array Labs also states it derives digital elevation models of the terrain down to 10 centimetres of resolution. The stated plan is to launch the first satellites of the constellation between 2027 and 2028: today the technology is validated on the ground and in operational trials with the US armed forces, not yet flying in constellation form.

The deal, and what it does not say

Mitsubishi Electric brings its capacity to build and launch satellites — its Defense & Space Systems division was awarded a contract by Japan’s Ministry of Defense on 6 February 2026 for the country’s next-generation military satellite communications system — and a network of relationships with Asia-Pacific governments that a Californian startup does not have on its own. Array Labs brings its AMTI technology and elevation data. The stated aim of both companies: to jointly commercialise space-based maritime and aircraft tracking services for defence and security customers in the region.

What the two releases do not say matters just as much. Neither mentions exclusivity: nothing states that Mitsubishi Electric is the only channel through which Array Labs may sell in the region. But it is, today, the only channel named. Until other regional agreements surface, anyone wanting to buy the service in Japan or nearby countries would, in practice, go through it. Nor is any operational availability date given: the satellites still have to be launched. What was signed on 28 July is an investment and an industrial intention, not a supply contract with an end customer.

What it means

For Array Labs the deal is worth more than the capital itself: it opens access to the Japanese and Asian defence market that no Californian startup reaches alone, for reasons of institutional relationships as much as of asset qualification. For Mitsubishi Electric it is the opposite: buying, for a few million dollars, a formation-flying radar technology that would have taken years of in-house development to reach the same point, at a moment when space-based maritime surveillance across the Indo-Pacific has become a stated priority for governments in the region.

For buyers of similar capability in Europe — defence ministries, coastguards, port operators — the case matters beyond Array Labs: increasingly, a small supplier’s technology reaches the market bundled inside a large group’s offer, with that group acting at once as customer, investor and distributor. Assessing “the product” is not enough: buyers also need to assess who sells it in their own territory, and what happens if the large group changes priorities. The same principle applies to another constellation presented as sovereign: who really controls the distinctive component matters more than the flag on the programme.

The risks, in order of how concrete they are

First, dual use: a service that tracks aircraft and ships from orbit is dual-use by definition; export-control rules — American ones on the origin of the technology, Japanese ones on its onward transfer — apply to both the satellite and the data it produces, and can change by political decision overnight, as China has already shown by blocking dual-use exports to 14 European companies. Second, concentration: a thirty-person company that depends, for its most promising market, on a single investor-customer-distributor carries a lock-in risk that no clause truly removes. Third, timing: two to three years still separate the announcement from a satellite in orbit; anyone drafting a tender on the strength of a press release is planning against a promise, not an available capability.

The operational lesson

Anyone assessing whether to buy — today or in two years — a service born of a deal like this one should put in writing, before signing anything: the real date of operational availability, not the announcement date; who, in their own territory, holds the contractual right to sell and update the system, and what happens to that right if the large group transfers it; who processes the data generated on their own territory, and where, and which dual-use export rules apply upstream and downstream. These are the same questions that, with different names attached, apply to every deal between an emerging player and a major operator: separating what the release promises from what the contract actually delivers.

Anyone buying a system that handles tracking or security data asks the same question about the software that analyses it: where the models run, who can access them, who updates them. That is why a dedicated AI system for this kind of analysis must be delivered in two modes, never just one: on-premise in the client’s own environment, or on CSIDIA’s dedicated cloud — an environment reserved for the single client, access over a dedicated VPN, a data centre in Italy, premises staffed directly by us. Method matters more than the supplier’s name.

Do you need to assess a technology supply agreement that depends on a third-party commercial channel, or draft the clauses that protect you if that channel changes? Let’s talk it through in thirty minutes.