You felt it yourself: two front-line men gave you their time, their pain points, and an open door, and it felt almost too good. This report stress-tests that feeling against the real structure of Ukraine's drone market, so you walk into the next call knowing exactly what their warmth is worth and what it is not.
Picture this: before you spend real money building prototypes and flying to a war zone, you spend twenty minutes here. You learn that the enthusiasm is genuine but nearly free for them to give, that Ukraine cannot even afford the commodity drones it already makes, and that the one business model you might have assumed (ship AI drones, bill the ministry) is the exact model Ukraine is now legislating against. Better to know now.
The first draft leaned heavily on the longer, more detailed Serhii interview. This box consolidates what the Dimas (brigade) call changes, so you do not have to re-read the whole report. The body below has been corrected to match.
1. Dimas is the stronger buyer signal, and it softens the gloomiest finding. Serhii offered to test and give feedback (near-zero cost to him). Dimas went further: he said outright that if a prototype works he can buy it "by money of my brigade." That is the single best pro-demand data point in either call, a commander asserting direct discretionary purchasing power. It does not cancel the codification and localization wall in Section 03, but it means at least one contact is a potential real buyer, not only a free tester. Action: pin down the size and rules of that brigade budget next call.
2. Both contacts are warm personal relationships, now confirmed. Serhii came through Sayed (his CrossFit friend since 2014); Dimas is your own boxing coach from Kiev, a year of training together and a genuine friendship. So neither man is talking to you because the market sent him, both are doing it because they know and trust you. That cuts two ways: it confirms the core finding (the warm reception is relationship-driven, not demand-driven, do not read it as product-market fit), but it is also a real asset in a relationship-driven defense market, and it means the feedback will be candid, Dimas was already blunt on the calls (anti-AI, realistic on hit rates), not just being polite.
3. Dimas's numbers diverge from Serhii's, do not average them. Dimas reported a 4 to 6 of 10 hit rate, well above Serhii's 1 to 2 of 10 and the public ~20 percent. Different unit, different conditions, possibly optimistic. Treat it as a question to probe, not a settled fact.
4. Dimas reinforces the wedge, not changes it. His top pain (fiber cable quality + battery quality) is exactly the component-QC, hardware-adjacent opening flagged in the verdict, and he, unlike Serhii, names battery quality explicitly. Net effect on the verdict: the "is there a payable buyer" question moves from weak toward mixed-to-positive because of the brigade budget; the "localize or license" conclusion is unchanged.
Your instinct was that demand is huge and suppliers cannot keep up. For the cheap commodity drones, the opposite is true: supply already outruns the money.
Ukraine went from about 7 drone makers before 2022 to roughly 500 by 2025, producing around 2.2 million UAVs in 2024 and targeting over 4.5 million in 2025. Monthly FPV capacity rose roughly tenfold. The result is a glut of cheap drones the budget cannot absorb. Only 14 percent of producers expect full capacity utilization (OSW). Kyiv is now opening an export framework specifically to soak up the surplus. So the commodity market is the opposite of a desperate, underserved buyer.
The picture inverts for advanced capability. Here there is a real, unmet shortage, and it is constrained by components and quality, not by a missing supplier:
This is the encouraging part. Both men described exactly the Tier 2 pain the public data shows, which means they are credible, not pitching you a story.
Dimas (brigade): top problems are fiber-optic cable quality and battery quality; prefers fiber because radio jamming defeats frequency hopping. → matches the documented fiber-spool quality crisis (3DTech: defective coils with invisible internal refractions that snap mid-flight).
Serhii (unmanned battalion): signal loss is the number one killer; ~20 percent hit rate. → matches RUSI/IEEE (about 10,000 drones lost per month, mostly to jamming) and CSIS (FPV success 10 to 15 percent typical, 30 to 50 percent for the best operators).
One divergence worth noting: Dimas reported a 4 to 6 of 10 hit rate, well above Serhii's 1 to 2 of 10 and the public ~20 percent. That is a different unit, different conditions, and possibly an optimistic read, do not average the two; it is a question to probe, not a settled fact.
Three forces, none of which mean "you are our only hope." Stacked together they fully explain the warm reception.
1. A unit conversation is nearly free for them. Ukraine pushed buying power down to about 700 units with direct budgets (DOT-Chain Defence, Brave1 Market). But the same units are also habitual solicitors of donated equipment (Prytula, Come Back Alive, diaspora-funded production lines) and the designated field-testers in a bottom-up R&D model where a maker ships a prototype and gets feedback "within the same week." A unit that says "bring 5 to 10 and we will test them and give feedback" risks nothing: no money, no contract, a few flights, and comments they would generate anyway. The asymmetry is the whole point. You invest R&D, travel and hardware; they invest an afternoon.
"The more infantry you kill, the more drones you get to kill more infantry. This is becoming kind of a self-reinforcing cycle." First Deputy PM Mykhailo Fedorov, on the Army of Drones Bonus
2. Serhii literally described the gamified procurement engine, and it checks out. His "Yabali" points system is the real Army of Drones Bonus: units upload video-verified kills, earn ePoints, and redeem them on the Brave1 Market for equipment. Confirmed value (first-hand from Serhii): a killed enemy ≈ 12 points. The drone-operator 25, tank 40, and Vampire-heavy-bomber ~43-point figures are project-research enrichments, not in Serhii's account, treat them as illustrative. Over 400 units (around 95 percent of drone formations) participate. One correction worth carrying: points buy equipment, not cash; Serhii's "1 point ≈ 5,000" refers to "greenhouse" points-currency slang (not dollars and not a confirmed UAH/USD rate), so there is no official cash equivalence, treat any conversion as folk knowledge.
3. For both men, the proximate reason is personal trust, not market pull. Serhii came through Sayed, your friend of 10-plus years and his CrossFit buddy since 2014. Dimas is your own boxing coach from Kiev, a year of training together and a real friendship. In a market drowning in pitches, those warm relationships are why you got the time at all. That is a genuine asset, and the candor that comes with friendship (Dimas was bluntly negative on AI targeting) makes their feedback worth more, but it is still a relationship signal, not a demand signal.
The Dimas counter-signal, in fairness. Dimas did not just express interest, he said plainly that if a prototype proves useful he can buy it "by money of my brigade." That is the strongest pro-demand signal across either call: a commander asserting direct, discretionary purchasing power. It does not erase the barriers in the next section (a brigade buy still favors codified, locally sourced kit), but it does mean at least one of your contacts is a potential actual buyer, not only a free tester. Verify the scope and ceiling of that brigade budget on the next call.
Front-line enthusiasm is cheap to win and, on its own, close to worthless. Converting it into a paid contract is where foreigners hit a wall.
First, the track record is grim. The marquee Western systems mostly failed in-theatre: Anduril's Ghost and Altius were called an "abject failure" and dropped by Ukrainian forces in 2024; Skydio's CEO admitted his drone was "not a very successful platform on the front lines"; the Switchblade 300 at 60,000 to 80,000 dollars lost out to a 400 to 500 dollar FPV, a 150-to-1 cost gap. The lesson the market drew: disposability beats sophistication, and exquisite foreign hardware that cannot survive jamming at 150x the price gets quietly retired.
Second, even a beloved prototype cannot be bought through official channels until it clears a gauntlet:
Third, what Ukraine actually wants from foreign partners is now explicit: capital, components, technology transfer, and co-production lines inside Ukraine, not finished imports. The April 2026 "Drone Deals" framework is built on reciprocity and "mutual technological exchange." The firms that converted interest into contracts (Quantum Systems, Orqa, Terra Drone, Skyeton, Summa Defence) all did it by localizing and feeding Ukrainian component suppliers. China's May 2025 cutoff of DJI and components to the West and Ukraine is precisely what is driving the "make it in Ukraine" push, and it is exactly where foreign capital and parts are welcome.
Partly. The obvious AI play is crowded and capital-heavy. The genuinely open wedges are hardware-adjacent, and that actually fits an IoT shop better than an "AI targeting" shop.
The headline niche, autonomous last-mile lock-on (the drone finishes the strike after jamming cuts the link), already has well-funded owners: Auterion (50 million dollar Pentagon deal, 33,000 kits, lock-on from about 1 km), The Fourth Law's TFL-1 (mass-produced on Vyriy drones, about 80 percent hit in the final 500 m), Swarmer (largest Ukrainian defense raise, filing to IPO), plus Saker and ZIR. The Ministry also gates the Delta battlefield training data a newcomer would need. CSIS notes last-mile guidance lifts strike success from 10 to 20 percent up to 70 to 80 percent, so the prize is real and that is exactly why it is contested.
And both your contacts independently confirmed the known limits of this tech, which again marks them as credible:
Serhii: the auto-aim systems they tested only worked at 10 to 50 m from the target, not 500 m, and took 3 to 5 minutes to lock. Matches the public record: lock-on detection is only 300 m to 1 km, you are still jammed before lock, and camera quality caps performance.
Both men reject AI choosing human targets near their own troops. Matches Ukrainian policy: human-in-the-loop is retained; CSIS says full autonomy "is not yet present on the battlefield"; Russia reportedly recalled Lancet auto-targeting after field tests.
Where openings plausibly remain, and notice these lean IoT and computer-vision-for-quality, not autonomous weapons:
Why: There is no payable demand for one more commodity FPV, the budget is the bottleneck and factories sit idle. There IS real, unmet demand in Tier 2 (fiber quality, anti-jam comms, last-mile reliability), but it is quality and component constrained, and the obvious AI slice is already owned. Your realistic target is a narrow, hardware-adjacent wedge, with computer-vision fiber-spool QC the most defensible fit for an AI/IoT shop.
Counter: Demand can shift fast in a 3 to 6 month tech cycle, and the EU's ~6 billion euro drone money plus unit-level budgets mean cash is loosening. If you bring a genuinely novel capability at a disposable price point, the crowded-incumbent argument weakens quickly.
Why: "We will test it and give feedback" is the cheapest sentence in this market and the units say it constantly. The real driver here was a trusted personal intro. Treat the calls as free, high-quality market research and relationship building, which is genuinely valuable, not as a pipeline.
Counter: A trusted intro is not nothing, and Dimas went further than enthusiasm, he claimed direct brigade-budget buying power. In a relationship-driven market, Sayed's vouching for Serhii plus a commander (Dimas) who says he can actually buy is a better starting position than a cold foreign vendor. If you convert it into a Brave1 listing and a local partner, the warmth becomes a real edge.
Why: Every foreign winner localized: Ukrainian entity, local component sourcing, co-production, a Brave1 Market listing, and codification. Shipping finished drones from abroad to bill the ministry is the exact model Ukraine is legislating against. So either commit to a localization path (partner, entity, Brave1) around a narrow novel wedge, or be honest that this is a mission and learning play, not a near-term sale, and bound your spend accordingly.
Counter: Localization is slow and capital-heavy ("2026 build, 2027 payoff", favoring big patient money), which may be too much for a small shop. A lighter alternative: do not sell drones at all, license your wedge (for example a QC vision module or an anti-jam comms layer) to an existing Ukrainian manufacturer who already has the codification, the factory and the Brave1 slot. You supply the brain, they supply the body and the paperwork.