This Week in Dual-Use

NEWS

Ukraine deploys drones carrying drones

A year ago I wrote about Ukraine’s remote drone capabilities, following the success of Operation Spiderweb, a covert drone attack carried out deep inside Russia.

Now Ukraine is achieving deep strike success using drones carrying other drones, known as marsupials. They are being deployed by Ukraine against Russia in different setups - UAVs carrying UGVs, and USVs carrying UAVs.

Until now, range was the binding constraint of deep strike by drones. A drone could reach only as far as its fuel and control link allowed, and Russia responded by pulling its most valuable assets further east, beyond the reach of Ukrainian FPVs.

Now reach is cheap(ish) and defending against it is still expensive. For the moment that gives an attacker the upper hand (see Iran’s drone and missile campaign against Gulf states). At least until air defence becomes cheaper.

But this is also a window on to the future of warfare. 

UK unveils VANAHEIM and ANVIL

The MoD published details of Projects VANAHEIM (portable counter-drone for light infantry) and ANVIL (drones that detect and identify targets using computer-vision models retrainable against new threats within 24 hours). Scientists have been working with the British Army and industry partners to test new technologies and develop AI-enabled capabilities.

For those of us in the defence tech sector these seem like obvious projects - of course you need C-UAS at platoon level, and the ability to update software every 24 hours. But it’s worth reflecting on the pace of change within the UK military, which is often (unfairly) accused of being slow to adopt new technology.

These are capabilities that are still being refined in Ukraine right now, so it’s promising to see the lessons from the conflict already mapping across to capability requirements in the West.

The 24 hour retrainable model is particularly telling. The MoD has recognised the weekly iteration speed that Ukraine achieves in the field, and is seeking to match it.

The Pentagon's autonomy doctrine gap

General David Petraeus (Ret.) has weighed in on the fact that less than 2% of the Pentagon’s $54B FY27 autonomous warfare request is allocated to training and doctrine. He points out that the kit is being bought faster than rules are developed to govern it.

Petraeus is right. What exists on the battlefield today is large numbers of individually piloted machines, each requiring an operator. The marsupial raids against Russia are impressive, but they are not autonomous.

So for the moment manpower is still important. And manpower has to be trained, which is the line item Petraeus has correctly identified as neglected.

A factory can double drone output in a year, but a training pipeline cannot double the supply of competent operators in the same period. And the swarm narrative is probably oversold (at least for the moment), so the bottleneck will be trained operators.

Eventually operators will be replaced by software, something I wrote about recently. But for the moment humans must learn to pilot the kit.

FUNDRAISING

DEFENCE

Agon, a British developer of AI training infrastructure and synthetic battlefields for autonomous systems, emerged with $30M in funding from Lakestar, Lux Capital and others.

Monava, a Swedish-Finnish developer of passive acoustic drone detection, raised an undisclosed amount led by Gungnir Capital.

SPACE

K2 Space, an American developer of satellites for commercial and national-security payloads, raised a $500M Series D co-led by Kleiner Perkins and ICONIQ at a $6.8B valuation.

Stars Harmony, a Chinese company building a space situational awareness constellation, raised a $14.8M round.

QUANTUM, CYBER & AI

Moonshot AI, a Chinese AI lab and developer of the Kimi chatbot and open-weight AI models, raised a $3.5B round at a $35B valuation

Horizon3, an American company that probes networks for vulnerabilities in a controlled, authorised manner, raised a $250M Series E co-led by NightDragon and NEA at a $2B valuation.

ZuriQ, a Swiss developer of trapped-ion quantum processors that arrange ions on two-dimensional chips to support larger, more scalable quantum computers, raised a $25.5M seed led by Quantonation.

GOING PUBLIC

Space-Eyes, an American developer of AI agents that orchestrate situational awareness and control, backed by Eric Trump, is going public in a $638M SPAC deal with McKinley Acquisition Corp.