Leveraging AI democratisation for drone development
Plus The Headlines, Money Matters, and Recommended Reading (Week 5 2025: 27 Jan - 2 Feb)
Welcome to today’s Beyond Line of Sight edition.
I want to acknowledge last week's air crashes in Washington D.C. and Philadelphia. Thoughts go out to the victims and their families. Even with the most sophisticated and mature systems and regulations, tragedies can happen. This is a reminder of the high stakes manned-unmanned airspace integration. One of the aircraft involved in these crashes could have been a large drone. The result would have been just as tragic.
Developments last week in AI has significant implications on the drone industry.
Recently, DeepSeek, a Chinese startup, released R1, a model competing with OpenAI's ChatGPT. DeepSeek trained R1 reportedly for only $6 million. By comparison, OpenAI spent more than a hundred million to train ChatGPT-4.
The $6 million figure is likely deflated, but it has stuck in the media. Last Monday, Nasdaq dropped 3.1% in a day, with Nvidia being its biggest loser, down 17.4%.
Apart from the stock market sell-offs, the world is questioning incumbent business models. This is because DeepSeek's R1 model:
Costs 27x less than ChatGPT-4o1 model for similar performance
Is trained on less powerful GPUs (due to U.S. sanctions on China)
Is open source (and can run offline)
It's not without its eyebrow raises and praises. OpenAI claims DeepSeek has plagiarised its training data. Some postulate CCP subsidised costs or DeepSeek are outright lying. Some have also called this the "Sputnik moment" for AI.
Whatever the case, this has put pressure on OpenAI and incumbents. OpenAI responded by releasing its o3-mini model for free and will speed up future product releases. Microsoft already made the DeepSeek R1 model available on its Azure AI foundry. This is interesting, since Microsoft has invested $13 billion in OpenAI.
What does this DeepSeek saga have to do with drones?
AI is fast democratising and competition is increasing. Like drones, AI is also on the way to becoming a commodity.
When companies, investors, or stakeholders talk about AI for drones, it's usually:
Target /object detection, route planning, GNSS denied operations, etc.
These are useful, but all about drone operations.
Drones are getting commoditised, and the market for drone manufacturers is already crowded. There is a massive opportunity in using AI for drone development.
Why?
Outside of meetings, all engineers spend most of their time on these activities:
CAD
writing code
generating or analysing data
writing reports and presentations
finding reports and research
AI is a productivity tool on steroids. And there are AI-based tools for all these activities. These tools can increase productivity two-fold, three-fold, or even more.
Leveraging AI, engineers can spend more time solving problems and solve problems faster.
Drone manufacturers can improve profit margins by reducing product development costs and time.
AI is not a magic pill and has its limitations. But companies need to be ready to leverage the fast-improving AI models and capabilities.
Drone companies need to re-design their engineering processes with AI tools in mind. Today, this is an afterthought for most.
The takeaway:
AI is fast democratising. Drone manufacturers and OEMs should re-think their engineering approach to leverage AI democratisation. Those that do adopt it early will become more profitable, competitive, and successful.
The Headlines
Defence and Security
Monday last week, Russian forces launched a significant drone attack across multiple regions in Ukraine, targeting critical infrastructure. Ukraine's air force reported shooting down 57 out of 104 drones.
Ukraine's military has developed long-range drones capable of striking targets up to 1,240 miles away, allowing them to reach deep into Russian territory. A Ukrainian Leopard 1A5 tank, enhanced with additional armour, endured at least eight direct hits from Russian FPV drones before being immobilized.
From the U.S., Lockheed Martin has successfully demonstrated that its F-35 fighter jet can control AI-enabled drones during flight, and the U.S. Marines conducted a live-fire exercise in Hawaii using the Marine Air Defense Integrated System (MADIS). The U.S. Army is seeking industry input for a new spy aircraft capable of launching drones to enhance ISR capabilities and have initiated the production of 3D-printed tactical drones to enhance special operations.
President Trump has signed an executive order to develop a nationwide air defense system, dubbed the "Iron Dome for America".
From the Middle East, Iran's Islamic Revolution Guards Corps unveiled a new 3,100 kg drone named "Gaza" during a large-scale military exercise.
Landing Zones Canada has successfully tested the Eagle Advanced Payload Delivery System (APDS), the world's first balloon-launched stealth military glider.
The U.S. Air Force's XQ-67A drone is undergoing significant upgrades under the "Demon Ape" program to enhance its modular design and capabilities.
NATO countries are urgently developing drones that can operate in the Arctic's extreme cold, as rising geopolitical tensions and increased Russian and Chinese activities in the region expose current technological limitations.
Commercial
ANRA Technologies has successfully completed Project CAELUS in Scotland, focusing on integrating drones into the National Health System (NHS) Scotland for medical deliveries, enabling complex BVLOS operations.
Swiss company Dufour Aerospace has successfully completed its first flight test campaign of the Aero2 drone.
China has developed the CH-YH1000, an unmanned transport aircraft designed to carry up to 1,000 kilograms of cargo.
Product Launch
Roark Aerospace has introduced over-the-counter drone detection systems starting at £4,500 ($5,715).
Amprius Technologies unveiled its new SiCore™ cell, a silicon-anode battery designed for drones and EVs with an energy density of 370 Wh/kg, while BEI have developed one with energy density of 410 Wh/kg.
Terra Drone Corporation has introduced its first domestically developed indoor inspection drone, the "Terra Xross 1," in both Japan and the U.S.
Legal
Matternet has received approval from Saudi Arabia's General Authority of Civil Aviation to operate its M2 drone in the country, marking the first such authorization granted.
The White House has clarified that the controversial New Jersey drone sightings in late 2024 were FAA authorised for research and recreational purposes.
A 56-year-old from Culver City has agreed to plead guilty to recklessly operating a drone that collided with a firefighting aircraft during the Palisades Fire on January 9, 2025.
Partnerships
Manna Aero and Wolt have initiated a pilot program in Espoo, Finland, to test drone deliveries during the challenging Nordic winter.
Japanese drone manufacturer ACSL has partnered with Exertis Almo to expand its distribution network across North America.
Skyway, a California-based company specializing in vertiport development and operations, has partnered with MatrixSpace, a firm focused on AI-enabled sensing technology, to enhance air traffic management and unmanned aircraft detection systems.
Money Matters
Contracts
ANELLO Photonics: secured a Phase I Small Business Innovation Research (SBIR) contract with the U.S. Navy to demonstrate its advanced optical gyroscope and sensor fusion technologies for navigation in GPS-denied environments.
Investment and M&A
Piaggio Aerospace and Baykar have signed a preliminary contract for the transfer of Piaggio's business complexes. The deal was authorised just before end of 2024 and is expected to close in the spring.
Vizgard, a UK-based AI software developer, has secured £1.5 million in funding to enhance its FortifAI platform.
Indian drone manufacturer Garuda Aerospace has announced a ₹100 crore investment to develop Andhra Pradesh as India's first 'Drone City'.
Recommended Reading This Week
ABOUT THE AUTHOR
Adit Shah is a UK-based business and strategy consultant specialising in aerospace, defence, and space sectors. If you are interested in Adit’s expertise, please get in touch via LinkedIn.