This Week in Dual-Use
NEWS
In counter-drone news, Anduril has partnered with OpenAI to deploy AI for national security missions. The partnership will focus on improving Anduril’s CUAS capability and seeks to understand whether OpenAI’s model can synthesise data faster than a human operator, improving response time to threats. The models will be trained on Anduril’s library of data on CUAS threats and operations.
This is an interesting development. OpenAI had previously stated that its products wouldn’t be employed by the military, and much of the current press coverage focuses on the concerns of using AI for warfare. But I think that misses the point. On the face of it, one might assume that this is about improving Anduril’s CUAS capability against Russian drones, since its products have been deployed in Ukraine and that is presumably the source of its data library. But the Anduril press announcement makes no mention of Ukraine. Instead it focuses on the challenge of keeping up with Chinese technological development.
What to take from this? Although Anduril products have seen some deployment in Ukraine, I suspect that their price point makes them too expensive for Ukrainian forces, who need to buy drones in the thousands. Although any drone manufacturer worth their salt must have their kit ‘battle tested in Ukraine’, the theatre may no longer be Anduril’s primary strategic focus. I also think the company has sensed the shift in political winds and is pivoting to the Asia. The co-founder of Anduril, Trae Stephens, had previously been tipped for the role of Deputy Secretary of Defense, so is clearly close to the new administration. It is also a recognition that China, not Russia, is the true adversary of the US in both military technology and AI. This is not news to national security and defence organisations, who will be glad to hear that the global leader in AI has unwound its policy of not working with defence suppliers. It is unlikely that Chinese companies have such reservations. LINK
In quantum computing news, Google released a new quantum chip, Willow, which has demonstrated some exciting results in quantum error correction. Thanks to an improved processor, Google's team found that increasing the number of hardware qubits dedicated to an error-corrected logical qubit led to an exponential increase in performance. The work proves that quantum error correction is likely to support the execution of complex algorithms which might require hours to execute.
Why are people excited about this? Error correction is one of the biggest challenges in quantum computing, as quantum systems are incredibly sensitive to noise and errors. This development means that long-term stability for quantum systems is feasible. In short, we can take confidence that useful quantum computing is likely to be a reality, and is not too far away.
This is not just a technological milestone but also a national security milestone. As in AI and military technology, adversarial nations are developing quantum computers and the risk is that they achieve quantum supremacy before the West. They could then use it for cyberattacks, espionage or even to undermine economic systems. So this is a race. But unlike Cold War arms races, the players here are not exclusively nation states. And the question isn’t just who gets quantum first, but how ownership and influence over this technology will be divided between governments and commercial entities - and what that means for everyone else. Who will set the rules for this new frontier, and how will they balance security, innovation, and access? LINK
In AI news, Anduril and Palantir have partnered to improve US government data readiness and processing government data at scale. The plan is to use Anduril’s software (Lattice) and deployable compute (Menace) to harvest data from government systems, then pipe this into Palantir’s AI Platform (AIP), which will prepare it for AI training. This will apparently be done at all levels of classification. They hope to expand the partnership to other industry players.
This is another example of how AI isn't just a technology but an enabler for completely rethinking processes at scale. Governments have vast amounts of data - some of it real-time, some deeply classified, and much of it siloed or inaccessible. The challenge isn’t just processing it but turning it into something actionable, and that’s what this partnership is trying to solve.
It’s also a reminder of the geopolitical AI race. If the US can integrate AI effectively into national security, it sets a benchmark that adversarial nations will scramble to match. The bigger question: how do you manage trust, accountability and transparency when systems like this become black boxes? LINK
In space news, China’s new Long March 12 rocket made a successful inaugural flight, placing two experimental satellites into orbit and testing uprated, higher-thrust engines that will allow a larger Chinese launcher to send astronauts to the Moon. China's Long March rocket family has been flying since China launched its first satellite into orbit in 1970. This new Long March rocket is part of a bid to develop a less expensive way of deploying thousands of small satellites to rival SpaceX's Starlink network. China is currently working on a super-heavy fully reusable rocket which appears to be a clone of SpaceX's Starship.
China’s ambitions in space continue to accelerate, and the parallels with SpaceX’s playbook are striking. The Long March 12 rocket isn’t just a step forward in payload capacity or cost efficiency - it’s a statement of intent to compete directly in commercial satellite deployment and human spaceflight. The move toward reusable rockets underscores how closely China is tracking, and arguably emulating, the disruptive innovations that SpaceX has pioneered.
But this isn’t just about technical competition. China’s satellite deployment ambitions tie directly into broader strategic goals - extending influence over global communications infrastructure, building an alternative to the Western-dominated internet, and supporting military capabilities. I am interested to see whether these efforts will spur a new space race, and how geopolitical tensions in space might play out. LINK
FUNDRAISING
Nscale, a UK-based developer of infrastructure for AI, which launched from stealth in May 2024, has raised $155m. LINK
D-Fend Solutions, an Israel-based developer of non-kinetic counter-drone technology, has raised $31m. LINK
SpinLaunch, a California-based developer of low-cost kinetic space launch systems, has raised £11m. LINK
Spexi, a Canada-based developer of a decentralized network for capturing high-resolution drone imagery, has raised $11.5m. LINK
PLD Space, a Spain-based developer of private satellite launchers, has raised €16m. LINK
GovSignals, a New York-based developer of an AI platform which streamlines the process of finding and responding to government contract opportunities, has raised a $5.5m Seed round. LINK
Pixxel, an India and Los Angeles-based developer of hyperspectral imaging satellites, raised a $24m Series B extension. LINK
Netacea, a UK-based developer of counter-bot cyber security, raised a £4m round. LINK
Fibrecoat, a Germany-based developer of aluminium-coated fibres designed to protect against radiation, heat and electromagnetic interference, raised €20m. LINK
System Two Security, a California-based company which automates the process of identifying and analysing potential security breaches, has raised $7m. LINK
Axiado, a California-based developer of AI-driven hardware security processors, has raised a $60m Series C. LINK
Tuskira, a California-based company which consolidates data from over 150 security tools to identify and address cybersecurity threats, has raised $28.5m. LINK