A £6M software contract is now the most interesting line in a British defence announcement. Arondite, a London company founded in December 2023, told Tectonic Defense that its deal to put the Cobalt platform across the UK Commando Force is valued at roughly £6M ($8M). The Ministry of Defence describes it only as "multi-million."

Two larger numbers sit beside it. The Commando Force gets a £100M capability uplift through the Defence Investment Plan, which commits £298bn over four years. Arondite separately plans to spend £35M, or NOK 430 million, building its first office outside Britain, in Oslo, over five years.

Read the ratio and the priority becomes obvious. Money still buys hulls, airframes and missiles. Influence is shifting to the software that decides what those things do. A firm that expects to double to 150 staff by next summer is being handed the coordination layer for Britain's amphibious strike force, with joint Norwegian testing promised and a central role in Babcock's ARMOR Force framework.

The case for the layer

Cobalt is mission orchestration software. It pulls sensor feeds, drone tracks, vehicle positions and written reports into a single operational picture, then lets a commander task the connected assets from one screen. Arondite says more than 50 integrations are live, with new ones arriving weekly.

"If the UK Commando Force, or any other team we work with, buys a new sensor, data source or effector, we make sure it can be integrated into Cobalt," co-founder and chief executive Will Blyth told Tectonic Defense.

That claim is the whole commercial argument. Hardware fleets are heterogeneous by nature: Royal Marines buy from whoever passes trials, and the Joint Expeditionary Force adds nine more nations with their own supply chains. Someone has to make the parts talk. The company that owns that translation layer sees every new purchase as extra demand, while the platform vendor sees a competitor each time.

The British Army has already tested the premise at scale. On Exercise Rhino Titan, over six weeks, 16 Air Assault Brigade ran Cobalt across more than 2,000 soldiers and 400 networked edge devices. Arondite has held that brigade contract since March 2026, delivered under RAPSTONE, the Army's rapid-insertion framework.

American evidence points the same way. At Ivy Mass, the Army's next-generation command-and-control mesh carried 40 applications, 36 data feeds, 37 effectors and nine hardware form factors, all communicating across the mesh. The number of tactical edge computers doubled past 130 nodes. Operation Jailbreak, the hackathon that preceded it, existed largely to force vendors to publish their interfaces.

Even the gun-makers are hedging. Rheinmetall published APIs for its Battlesuite middleware on GitHub on 9 September, opening the Onboard and Tactical components to outside developers. A vehicle manufacturer decided that being the integration surface was worth more than keeping it closed.

Primes tolerate the layer because it lowers their own bid risk. Babcock's ARMOR Force framework puts Cobalt at the centre of a multi-vendor maritime ecosystem that includes the US shipbuilder HII, and Rheinmetall's open APIs invite the same arrangement on land. The integrator absorbs the compatibility work that would otherwise land on the platform's own engineering team. Nobody in that arrangement has to declare a winner yet.

The cost structure explains the enthusiasm. A platform programme spends years and billions before the first unit ships; an integration layer deploys in months across units that already exist. Arondite's own trajectory shows the compounding: a 2023 founding, a $10M seed led by Index Ventures in May 2025, an Army brigade, a commando force, a Norwegian office, and a maritime framework, all inside three years.

The case for the platform

There is a competing reading, and it is not weak. The most successful defence software layer of this decade belongs to a hardware company. Anduril's Lattice runs the US Army's NGC2 mesh, the same one that hit 40 applications at Ivy Mass. Lattice is open enough to integrate third parties and closed enough that Anduril captures the value. The layer did not displace the platform; the platform company built the layer and kept it.

Rheinmetall's GitHub release reads the same way from the other side. Publishing APIs invites developers into an architecture the company still defines, and where its own hardware sits at the centre. Openness is a distribution strategy, not a surrender of control.

Then there is the contract itself. The Ministry of Defence has not released acceptance criteria, a measured improvement in decision speed, or the value of the award. IN Defence noted that Arondite has not published the interface standards, communications protocols or data-assurance arrangements behind the commando deployment. Norway gets joint testing; no purchase commitment accompanies it.

Those gaps matter because integration businesses can look like products while behaving like consultancies. Every customer arrives with a different estate, different security rules and different legacy software. A Baltic Defense Initiative profile of the company framed the central risk plainly: the next records to watch are repeat purchases and customer-described milestones that would distinguish a reusable platform from a series of isolated integration projects.

Switching costs cut both ways here. If Cobalt genuinely absorbs 50-plus systems from 50-plus vendors, the cost of replacing it is high. If it is open by design, a competitor can plug into the same interfaces and bid the next contract lower. Openness is the sales pitch and the commoditisation risk in one sentence.

🎯
Where the value actually sits

Software layers scale faster than platforms but are easier to re-compete at contract renewal.

The winner is whoever makes integration work repeat instead of bespoke — Arondite has the contracts, not yet the proof.

Watch repeat orders and measured decision-cycle times, not announcement counts.

What the announcement does not say

The public record is thinner than the press release suggests. Four of the five most decision-relevant data points are missing, and the one that exists comes from the company rather than the buyer.

ItemDisclosedMissing
Contract value ✔ £6M, company-stated ✗ MoD figure
Decision-speed gain ◐ Qualitative claims ✗ Any measured baseline
Acceptance criteria ✗ None published ✗ Milestone schedule
Integration standards ◐ "50+ integrations" ✗ Protocols, data assurance
Norwegian commitment ✔ Joint testing planned ✗ Purchase decision

Sources: GOV.UK/Ministry of Defence (12 Nov 2026); Tectonic Defense (13 Nov 2026); IN Defence (13 Nov 2026)

None of that is unusual for a first award, and it is not evidence of failure. It does mean the £6M should be read as an option premium rather than a verdict. Britain is buying the right to find out whether orchestration software holds up in Arctic conditions, at a price small enough that the political cost of being wrong is contained.

Arondite is behaving like a company that expects the answer to be yes. It hired Dilan Koç, previously chief technology officer of NATO's Task Force Maven, to lead deployments with allied customers, and put Eirik Friise, twelve years in the Norwegian Armed Forces, in charge of Oslo. Headcount is due to double to 150 by summer 2027. As we wrote in November about Leonardo's HiveMind acquisition, defence buyers are paying for autonomy and getting integration — the difference between owning the capability and owning the layer that commands it is where these deals keep landing.

The two cases will not be settled by argument. They will be settled by whether the second and third commando orders arrive without a bespoke integration project attached, and whether anyone outside Arondite ever publishes a decision-cycle number. Until then, the £6M is the cheapest test in the announcement — and the only one that touches how the force actually fights.

UK commandos get AI-enabled edge from Arondite as the company expands to NATO's northern flank
The official Ministry of Defence release confirming the Commando Force deal, the £35M Norway investment, the £100M force uplift and the £298bn Defence Investment Plan.
The only primary-source document in the set. It confirms the relationship and the money around it, and leaves the contract value unstated.
Arondite takes a multi-million pound Royal Navy contract and expands to Norway
Carries the only figure attached to the award — £6M, as stated by the company — plus Blyth on the 50-plus integrations already live in Cobalt.
Where the contract number comes from. Treat it as a company figure until the MoD publishes its own.
Arondite wins Royal Marines mission software deal
The clearest account of what has not been disclosed: interface standards, data assurance, acceptance criteria and measured performance gains.
Useful mostly for its negatives. The list of missing data is the strongest argument against reading the award as proof.
Babcock assembles an industry ecosystem for the Hybrid Navy
Seven partners join the ARMOR Force framework, with Cobalt named as the platform that coheres the systems on ARMOR Force vessels.
The maritime half of the story, and the reason the commando award is not an isolated pilot.
Accelerating multi-domain interoperability through open software architectures
US Army account of the NGC2 mesh at Ivy Mass: 40 applications, 36 data feeds, 37 effectors across nine hardware form factors.
The best available evidence that orchestration layers work at formation scale — and that a platform company can own one.