Manufacturing for the Mission: What it Takes to Build Hardware for DoD
Neros’s Journey to Building the First Blue UAS-Approved FPV Drone
First, I just wanted to make a quick announcement: registration is now open for the second annual National Security Hackathon, taking place April 26–27 in San Francisco! This isn’t your average hackathon. Hackers will be tackling real-world national security problem sets provided by DoD stakeholders alongside mission-driven peers, with direct input from government stakeholders. The event is hosted by Stanford DEFCON, Stanford Gordian Knot Center, and Shield Capital. Other sponsors include Scale AI, IQT, NATO Innovation Fund, Vannevar Labs, Anthropic, Cerebras, Microsoft, and more! Register here if you’re interested in joining, and please forward to anyone else who may be interested in participating.
OK, on to the article…
Bringing manufacturing back to America is all the rage right now. Earlier this month, the Trump administration announced a swath of "reciprocal tariffs” (which have since been postponed for 90 days) designed in part to encourage manufacturing to return to the United States. Today, the American defense industrial base (DIB) and other critical industries are overly reliant on raw materials, components, and other manufactured goods that come from the People’s Republic of China (PRC) and other foreign nations, which leaves the US vulnerable to supply disruptions in a future conflict.1
I figured this was the perfect moment to dive into what it is required to manufacture critical systems in the US, and what it means to build and deploy hardware for national security missions. Just like software, delivering hardware to the Department of Defense (DoD) is far more complex than in the commercial world. I asked Gleb Shevchuk, an engineer at American-made FPV drone2 startup Neros, to co-author this piece with me to outline what it takes to develop American-made hardware for the DoD. In just 18 months, Neros designed, manufactured, and deployed the only FPV drone on the Blue UAS list, and now ships thousands of drones each month to active conflicts.
Current State of US Supply Chains and Manufacturing
As I’ve written about in the past, it’s not entirely fair to say that the US does not manufacture things anymore. In fact, the US manufacturers more today than it ever did in the pre-globalization era. However, it is true that many fewer Americans today actually work in manufacturing, making it a much less visible industry to the average American. Further, manufacturing is a much smaller share of the US’s GDP today than in the past, and while the US is still the second largest manufacturer in the world, the US controls a much smaller share of overall manufacturing than it used to.
Regardless, there is no doubt that US supply chains are overly reliant on nations like the PRC. Many parts of the DIB rely on supply chains with key vulnerabilities in the PRC, posing a threat to US national security. A Govini report highlights that many US military systems rely on microelectronics from the PRC, and another report reveals, “China is the single or sole supplier for a number of specialty chemicals used in munitions and missiles. In many cases, there is no other source or drop-in replacement material and even in cases where that option exists, the time and cost to test and qualify the new material can be prohibitive – especially for larger systems (hundreds of millions of dollars each).”
Source: “Numbers Matter: Defense Acquisition, US Production Capacity, and Deterring China,” Govini
The startup community became acutely aware of US DIB supply chain vulnerabilities in late 2024 when the Chinese government hit US drone manufacturer Skydio with sanctions, prohibiting Skydio from purchasing the Chinese-made batteries it needs for its drones. Skydio predicted that it would take more than half a year to identify a viable alternative battery source.
Unmanned aerial systems (UAS) and other “precision mass” will be crucial technologies in the future of warfare. Today, the PRC holds a commanding lead in both UAS development and the production of key drone components. Chinese company DJI dominates the FPV market, with ~70% global drone market share, and even non-Chinese FPV drone makers rely on Chinese components. One Ukrainian military official suggested that Ukraine was buying as much as 60% of all DJI Mavic quadcopters. It is essentially impossible to procure certain UAS components, such as motors, from non-Chinese sources, leaving American UAS supply chains incredibly vulnerable to Chinese trade policies.
Of course the DoD is aware of the challenges associated with UAS supply chains. The NDAA3 mandates that UAS companies eliminate certain Chinese components from their supply chains to remain eligible for DoD contracts. The Defense Innovation Unit (DIU) manages the Blue UAS list, a list of vetted commercial drone platforms that meet the DoD’s cybersecurity, supply chain, and operational standards. Designed to accelerate the adoption of secure, reliable small unmanned aerial systems, drones on the Blue UAS List do not require a DoD exception to policy to procure or operate as they have undergone a cyber-security evaluation, an NDAA compliance check, and were issued the necessary administrative documentation.
It is not easy for a company to get on the Blue UAS list. One of the hardest requirements to get on the list is related to supply chains. Specifically, drones on the Blue UAS list must not include the following critical components if they are manufactured in the PRC, Russia, Iran, or North Korea: flight controllers, radio communication systems, data transmission devices, cameras, gimbals, ground control systems, operating software, network connectivity, and data storage.
So far, ten companies have made the Blue UAS list. Neros builds the only FPV drone on the Blue UAS list, the Archer, which is ten times cheaper than the next cheapest option on the list.
So, what does it actually take to build a BlueUAS certified drone that works on the battlefield? What does it take to build a compliant supply chain, harden a system for use in a conflict zone, and design products to meet DoD standards? Over the course of this article, we’ll outline the process Neros went through to get on the Blue UAS list, describing how Neros 1) managed their supply chain, and 2) hardened and tested their systems.
Supply Chain & Component Selection
A drone shares many of the same components as your phone. Like a phone, it has a computer, a radio, and a user interface – even though it has propellers, you could probably run Doom on it. However, unlike a phone, it also has to pull as much as ten kilowatts of power, fly for tens of miles, and survive through the conditions of war. To make that possible, we rely on dozens of intricate, electromechanical assemblies, and building the supply chain to support that has taken a massive amount of work.
Over the last eighteen months, Neros went through about seven block upgrades of our drone and four block upgrades of our ground control system, which means that we’ve had to repeat this process many, many times.
During the first month or so of Neros, like a lot of others in the FPV drone world, started by using off-the-shelf components, many of which were built around cheap, widely available Chinese electronics. But it quickly became obvious that if we wanted to meet the NDAA's compliance standards, we’d have to rip most of those Chinese made components out and start from scratch. The NDAA’s supply chain requirements primarily affect critical electronics like microcontrollers, radio chips, and other advanced integrated circuits. And unfortunately, the vast majority of those in the commercial drone space, including popular components like ExpressLRS radios, are built with Chinese chips. ESP32 chips, for example, are particularly popular in the FPV space because they’re inexpensive and come with built-in Wi-Fi and Bluetooth, but they’re exclusively made in China.
Now, if you had told me two years ago that we would be designing custom radios, motor controllers, flight controllers, antennas, I would have called you crazy. But that was the only solution: we had to design everything from scratch. We found suppliers outside of China for some components (motors, mechanical parts, etc), but redesigned a number of systems in-house. With each block upgrade, we tried to inhouse a few drone components, starting with the flight control board and ending with the video transmission module, and we put a big focus on using non-Chinese integrated circuits to drive each submodule. On the electronics side, we now produce about seven-core PCBs and assemble them in the United States with non-Chinese critical components.
Notably, the NDAA and Blue UAS list do not require that all components come from outside of China. Some components, like motors, are almost impossible to source outside of China. Right now, there are no U.S. companies manufacturing motors at the size and scale needed for small drones. These parts require precision engineering, expensive tooling, and manual, repetitive labor. The raw materials, particularly copper windings and silicon steel needed to make stators,4 are also difficult to source in the U.S. at the price point and in the form factors we need. For most US drone companies, motors still remain a tough bottleneck.
Additionally, mechanical components like the drone frame aren't regulated under the NDAA. China dominates the market for the kind of carbon fiber used to make drone frames for one simple reason: US-made carbon fiber is 10-20x more expensive. While the US excels at making aerospace-grade layup composites5 for planes and rockets, that expertise doesn’t translate well to the needs of small drone manufacturers. For the time being, there’s little incentive to source carbon fiber outside of China – it’s more expensive, lower quality, and not restricted by the NDAA. However, the US drone industry needs to onshore carbon fiber manufacturing or transition to structures that are not made of carbon fiber (bent sheet metal, injection molded plastic, etc). Otherwise, during a potential future conflict with the PRC, we’ll be stuck making drones out of cardboard.
One of the biggest trade-offs we’ve had to make is increasing the cost of our system. When we started, the general consensus among experts we talked to was that a US-made drone would cost about ten times as much as a Chinese-made drone (if you believed Twitter, that number was closer to 100-1000x). Though we still aren’t matching Chinese prices, we’ve managed to significantly decrease the cost of our system, mostly by in-housing our electronics stack.
Additionally, it has been difficult to find replacement components that meet our quality standards. Chinese suppliers are still the gold standard for drone motors, cameras, and other major components. To close that gap, we’ve had to work closely with vendors and do extensive testing, comparing specs, tuning components, and ensuring performance is up to par.
One benefit Neros has compared to other drones on the Blue UAS list is that our drones are much cheaper than other drones on the list (almost ten times cheaper than the next cheapest option), and are manufactured at a larger scale than other US-made drones. As such, Neros benefits from some economies of scale, and we are able to save by ordering large volumes of materials and components. If you're only building a small number of drones, everything becomes more expensive (especially custom-designed electronics). Without scale, you can end up paying 20x more per board just because of how pricing works at fabrication houses for small purchase orders. That’s where a lot of startups struggle. At Neros, we’ve leaned hard into volume to drive down those costs and make the whole approach viable.
Hardening and Testing
It’s difficult to design any product, let alone a drone, to survive the rigors of modern warfare. From meeting MIL-SPEC durability standards to surviving in electronic warfare (EW) environments, the process forces you to rethink everything: your architecture, your supply chain, even your assumptions about what a drone should look like. Hardware systems need to reliably work even after being dropped out of an airplane, deployed in the middle of a rainstorm or sandstorm, or jammed with enemy electronic warfare devices, and that takes a lot of testing.
Physical Hardening
MIL-SPEC and MIL-STD are DoD standards that dictate the performance of DoD hardware. They represent a massive collection of standards, ranging from waterproofing to EMI shielding6 to human usability standards and long-term storage. For a drone, that can mean everything from meeting an IP rating7 for dust and water resistance to proving that a system will still function after sitting in a warehouse for a decade.
Complying with these standards isn’t cheap. Military hardening means putting every subsystem into a waterproof, dustproof, impact-resistant enclosure and subjecting it to dozens of tests. That adds weight, complexity, and cost, and requires the use of expensive manufacturing techniques like injection molding and casting. Non-MIL-SPEC UASs like those typically used in Ukraine are incredibly simple: just four screws, a couple plates of carbon, and some exposed electronics and solder. Unlike MIL-SPEC hardened systems, they’re cheap, light, and easy to replace and repair.
It’s also important to recognize the mismatch between traditional military hardening expectations and the new reality of low-cost, attritable drones. MIL-SPEC standards were developed for large, multi billion-dollar weapon systems that are too important and expensive to lose. FPV drones, in contrast, cost less than $5000 and don’t need to last 10 years. They don’t even need to survive their mission. That shift in mindset hasn’t caught up across the board, and it’s part of the reason why DoD procurement is still slow and expensive. The acquisitions process is designed to procure a fundamentally different kind of system.
Electronic Warfare Hardening
Electronic warfare (EW) is a fancy word for a simple idea: if your enemy is broadcasting on a frequency, mess with their signal. The two main ways to do that are to either broadcast “louder”, or to spoof their transmissions. Now, before talking about what it takes to EW-harden a system against this kind of interference, I’d like to reflect on how we’ve seen these two concepts play out in Ukraine.
In everyday life, our devices transmit on pre-defined frequencies called ISM bands.8 These are specific frequencies set aside for things like WIFI or Bluetooth, and our devices are regulated by bodies like the FCC9 to ensure that they don’t transmit on unauthorized frequencies. But, at the end of the day, there are only a few bands that commercial drones can legally communicate on. This becomes a problem in a warzone because adversaries can easily jam signals in those bands. Famously, other American drone companies have struggled to operate in Ukraine for this reason, while Ukrainians have innovated on modifying radios to transmit on those unused frequencies. This is not a phase – every future war will be fought on non-ISM frequencies, and the US is still playing catch up.
Another important lesson we learned from Ukraine is that GPS is obsolete, and it comes down to signal strength. GPS satellites are, by nature, very far away. Once a GPS signal reaches a system on the ground, it has a strength of around -130 dBm, which, to the unacquainted, is very, very quiet. Now, the second you pick up a jammer off your favorite Chinese e-commerce marketplace (both Temu and Alibaba sell cheap jammers online) and start broadcasting your own 35 dBm signal, your GPS receiver now has to distinguish a very weak satellite signal from a very strong jamming signal. This might not be so bad if jammers were complicated, multi-million dollar systems, as might have been the case fifty years ago. Nowadays, however, you can make one for less than five dollars. GPS is dead, and yet the U.S. military remains heavily reliant on GPS-based navigation.
At Neros, we’ve tried to follow what works in Ukraine: don’t use ISM bands, and don’t use GPS. On the command and control (C2) side, we build cheap, frequency-hopping radios that operate across a wide range of frequencies. If you can hop quicker, it becomes harder for a jammer to match your signal. If you can transmit on a wider bandwidth, it becomes harder for them to shout louder than you.
On the GPS side, there are a number of solutions in the works, but the most promising ones are based on visual navigation. One recent example of vis-nav comes from Maxar, and while many such systems are still at the prototype stage, they all follow roughly the same idea: pull a bunch of satellite data and make an algorithm to match what a drone’s camera sees with what a satellite sees. This technology works great for localizing a drone while you still have control of it, but, ultimately, the farther away a drone is from an operator and the more obstacles that are in the way, the easier it is for an adversary to jam your system.
Overall, the best way to test EW resilience is, unfortunately, the most obvious: deploy into real-world contested environments. We’ve done field testing in Ukraine and other operational settings. But in the US, you can’t legally test jamming capabilities outside of government authorized test facilities, which can take months to access.
To work around this, we simulate jamming conductively by connecting transmitters directly to receivers via cables and injecting noise directly into the signal chain. This allows us to model packet loss, test signal degradation, and evaluate how well our systems recover, all without emitting RF energy into the air. It’s not perfect, but it’s better than nothing. Simulators are also useful for testing autonomy software, but ultimately, nothing beats real-world testing. When it comes to EW, only real-world experience tells the truth.
Integration and Modularity
When building hardware systems for DoD, it is also important to ensure that your systems are interoperable with others. For example, at Neros, we made sure our radios and flight controllers could speak the same protocol as other systems. We also integrated with ATAK, the government’s standard tactical Android platform.
Additionally, the DoD mandates that hardware is developed using MOSA (Modular Open Systems Architecture) principles. MOSA aims to improve interoperability, accelerate innovation, and reduce costs by enabling plug-and-play integration of components from different vendors – avoiding vendor lock-in and making it easier to upgrade systems over time. So, Neros designed our system to have easily swappable and upgradable batteries, radios, and payloads.
Conclusion
One of the biggest challenges of building hardware for DoD is just the sheer number of edge cases hardware encounters on the battlefield. You can do 50 flights and still miss a critical failure mode. That’s why we’ve learned to test constantly and start simple. Build the core system that solves your customer’s biggest pain points, then layer in complexity.
Neros now flight-tests every drone we ship before they leave our facility, conducting over a thousand flight tests each month to ensure reliability and performance. That level of quality assurance wasn’t something we fully appreciated early on. We tested consistently, but not at the intensity we do now. With thousands of drones shipped so far, we’ve learned that it’s not just about building prototypes, it’s about designing for scale and making sure the production systems actually work in the field.
That daily test cadence is part of why our reliability is strong, especially compared to some Chinese drones, which can have failure rates as high as 30-40%. They're built at massive scale, and when you’re pumping out drones at high volume, little issues can slip through and multiply quickly.
Engineering-wise, the hardest part of building these drones for DoD customers has been the time and complexity. Across our entire system (our drones and ground stations) we’ve built a dozen different circuit boards, each running its own firmware, each needing its own test suite and integration. That’s a massive lift. But it’s the only way to create something that’s not only NDAA compliant, but also performs well enough to be fielded at scale.
In the early days, our team didn’t fully understand the production pipeline. But now, eighteen months in, we’re shipping at scale, which required investing in PLM (Product Lifecycle Management), MES (Manufacturing Execution Systems), and other supply chain tooling very early on. These systems help us keep our production consistent and scalable, even as we grow the team and the complexity of the product.
Unlike consumer electronics, we don’t have hundreds of thousands of consumers who can help us surface bugs. We have a handful of mission-critical customers. When something breaks, it’s not just an inconvenience, it could be life or death. So we’ve built a culture of responsiveness: we’re ready to hop on a plane, make urgent design changes, or swap out systems overnight.
Other startups looking to build hardware for DoD customers need to take the time to understand the supply chain constraints and system hardening required by their customers as early as possible. Of course, startups don’t need to meet all of the DoD’s stringent requirements when building initial prototypes, but they need to understand how the DoD’s unique supply chain and system hardening requirements will impact the cost, performance, and manufacturability of their systems, work those constraints into early product designs, and start the process of hardening systems and supply chains early on. Additionally, it is critical that startups test their systems as often and as realistically as possible to ensure that the systems actually meet the needs of warfighters. We’re still early in this process, and if anyone reading this blog wants to share their experience, we would greatly appreciate it.
Building American-made defense hardware is hard, but it’s not impossible. As the US looks to reassert its manufacturing strength and reduce reliance on adversarial supply chains, companies like Neros show what’s possible with the right focus, urgency, and willingness to rethink old assumptions. There’s still a long road ahead, but the blueprint is finally starting to take shape.
As always, please reach out if you or anyone you know is building at the intersection of national security and commercial technologies. And please let me know your thoughts! I’m always happy to chat about the world of NatSec + tech + startups, as I know this is a quickly changing space as the administration rolls out its new initiatives.
For an excellent overview of US Supply Chain Vulnerabilities, particularly as they relate to China and Russia, see the “US Supply Chain Vulnerabilities” report released by the US-China Economic Security Review Commision.
FPV = First Person View. FPV drones are small, cheap, remote-operated copter-powered UASs.
NDAA = National Defense Authorization Act
The stator is the stationary part of a rotary system, found in electric generators, electric motors, sirens, mud motors, or biological rotors
A Lay-Up process is a moulding process for composite materials, in which the final product is obtained by overlapping a specific number of different layers, usually made of continuous polymeric or ceramic fibres and a thermoset polymeric liquid matrix.
EMI = Electromagnetic interference, which can disrupt electronic devices, equipment, and systems that are used in critical applications.
IP = Ingress Protection. An IP rating is a two-digit code that defines the level of protection an electronic device has against the penetration of solid objects, dust, and water.
ISM = Industrial, Scientific and Medical
FCC = Federal Communications Commission
Note: The opinions and views expressed in this article are solely our own and do not reflect the views, policies, or position of our employers or any other organization or individual with which we are affiliated.
 |
|
Re: American-made motors and actuators for UAVs and similar systems, would look into https://corvexsystems.com/. New startup attempting to tackle exactly that.