The Procurement Question Most People Don't Ask

When an agency evaluates a drone platform, the spreadsheet almost always starts with unit price. Drone cost. Dock cost. Annual software licence. Training. Maybe a line for spares. The numbers get compared, the cheapest-per-capability vendor wins, and the procurement officer moves on to the next line item.

The question that gets asked much less often is: what does it cost us to change our mind in 2030?

This is the real TCO question for any agency buying drone capability on a five-to-ten-year horizon. Hardware refreshes. Payload requirements shift. A regulator changes the rules. A vendor gets acquired, deprecates a product line, or raises prices 40% on renewal because they've worked out you can't easily leave. The cost of the original purchase is small next to the cost of replatforming when circumstances change.

Most procurement teams have been burned by this before — usually in IT systems, sometimes in radios, occasionally in vehicles. Drones are new enough that the lesson hasn't fully transferred. It's worth thinking about before the purchase order goes in.

The Closed-Vertical Bet

Two vendors dominate the autonomous-drone category for public safety buyers today: Skydio and DJI. Both are excellent at what they do. Both run closed verticals.

Skydio Dock is the clearest example. You buy the drone (Skydio X10 or X2D), the dock, the flight software, the cloud platform, and the operator console all from Skydio. The onboard autonomy is genuinely leading-edge — Skydio's obstacle avoidance is arguably the best in the industry. The integration across the stack is smooth because one vendor controls every layer. If you want a turnkey autonomous drone in a box and you're happy with Skydio's choices, Skydio Dock delivers.

DJI's enterprise line (Matrice 30T, Mavic 3 Enterprise, Dock 2) follows the same pattern. Superb hardware, particularly the thermal stack on the M30T, tightly integrated with DJI's ground-station software, cloud platform, and payload ecosystem. For agencies outside procurement-restricted jurisdictions, the price-to-capability ratio is hard to beat.

The bet in both cases is the same: trade flexibility for integration quality. If the vendor survives, keeps pricing reasonable, keeps shipping features you need, and remains available in your jurisdiction, the closed vertical is probably the lowest-friction choice. That's a real bet, and for some agencies it's the right one. But it is a bet.

What Goes Wrong

The bet fails in four distinct ways, and public safety buyers have now seen all four in the wild.

Vendor deprecation. DJI drones are barred from US federal procurement under the American Security Drone Act, excluded from most state and local public-safety grants, and increasingly scrutinised in EU defence procurement and the UK Ministry of Defence supply chain. Agencies that bought deep into the DJI ecosystem a few years ago are now carrying fleets they can't replace like-for-like under current funding rules. Skydio exited the consumer market in 2023 to focus on enterprise and public safety — a reasonable business move, but a reminder that a vendor's product strategy is not under your control. If Skydio chooses tomorrow to sunset the X2D in favour of X10-only, every X2D in your fleet becomes a depreciating asset on a shortened timeline.

Pricing leverage. Once you've standardised on a vendor's batteries, docks, chargers, cloud, and training, the switching cost is high enough that the vendor can price renewals accordingly. First-party batteries that only work with first-party firmware. Annual software licences that renew at 20-30% above initial. Cloud storage tiers that meter per-flight-hour. None of this is unreasonable on its own; it's the cumulative effect of having no credible exit that changes the negotiation.

Feature gaps you can't close. Your operations team wants a specific thermal payload, or a gas sensor, or a LiDAR module, or a radio you can integrate with your existing comms kit. On a closed platform, you wait for the vendor to build it, or you don't get it. On an open platform, you or a third party integrate it on the existing payload bus. The gap between "we need this in Q3" and "the vendor's roadmap says 2027" is where capability debt accumulates.

Sovereignty changes. The EU is tightening rules on drone sovereignty, component provenance, and data residency. The US is tightening rules on Chinese-origin components under Blue UAS and NDAA. A single-vendor stack inherits that vendor's regulatory exposure. A drone assembled in Shenzhen with a US software platform has exposure on both ends. A stack you can reconfigure — different airframe, different ground station, different cloud — can be adjusted as rules shift.

The MAVLink Alternative

The open alternative is a stack where each layer is separable and standards-based.

MAVLink is an open messaging protocol maintained by the Dronecode Foundation. PX4 and ArduPilot are both open-source flight stacks that speak it. Together they form the dominant non-proprietary autopilot standard in the industry. Airframes from different vendors all speak the same protocol. Ground-station software written against MAVLink works with any MAVLink-compatible drone. Payload slots are physical and documented, not firmware-gated.

In practical terms: you can pick an airframe from Freefly for your 2026 fleet and swap in a different vendor for the 2031 refresh without rewriting your mission software, retraining your operators, or migrating your ground station. The flight-plan format, the telemetry messages, the command channel, and the safety constructs are all the same across vendors.

This is the same standardisation logic that underpins most serious infrastructure procurement. Agencies don't buy radios that only talk to one vendor's radios. They don't buy vehicles with proprietary fuel nozzles. Drones have not yet reached that level of standardisation universally, but MAVLink is the closest thing the industry has to it, and it's the protocol used by the majority of non-DJI industrial airframes on the market.

What Overwatch Is For

Overwatch is the software layer that rides on top of a MAVLink-compatible fleet. Mission planning, autonomous grid search, 24/7 fleet patrol with relay handoff, ground station, detection pipeline, safety supervision. One product, two operating modes, driven by fleet size.

Today, Overwatch runs on Freefly Astro Max (USA, NDAA / Blue UAS), Sky-Hopper (Israel), Quantum Systems (Germany, EU), and Parrot ANAFI UKR (EU / UK, via our AirSDK partnership). Whichever of these suits your jurisdiction, payload requirements, and budget, the software is the same. When the industrial MAVLink airframe that matters in 2028 ships — whoever makes it — we intend Overwatch to run on that too.

That is the trade we are offering procurement teams: commit to a protocol, not a vendor. Buy airframes on a procurement cycle; buy software on a capability cycle; let each evolve without forcing a replatform of the other.

The Honest Caveat

Open is not free. There are real trade-offs.

Integration cost is higher than a single-vendor stack in the first deployment. You are combining an airframe vendor, a payload vendor, a software vendor (us or someone else), and possibly a comms vendor — where a closed stack gives you one throat to choke. Polish at the seams is better on a vertically integrated platform because one company owns every interface. Decisions multiply: which airframe, which batteries, which payload, which ground station, which cloud — each a choice rather than a default.

For an agency that wants to buy a drone and fly it this afternoon, a closed vertical is probably faster to stand up. For an agency buying on a 5–10 year horizon, with a procurement team and a capital plan and regulators to answer to, the open stack is almost always the lower TCO outcome. Not because the closed vendors are bad — they're not — but because the cost of changing your mind is so much lower.

Overwatch is priced to match both models. Our Integrated Fleet tier starts at €15,000 per drone for a turnkey deployment with hardware, software, and onboarding included. Our Software-only tier is €3,000 per drone per year for agencies bringing their own MAVLink-compatible fleet. The software is the same in both cases; the decision is only how much of the stack you want us to supply.

If you're weighing a closed-vertical purchase against an open-protocol path, we're happy to walk through the TCO maths on your specific fleet and timeline. See our platform-selection guide for the hardware comparison, or Overwatch Ground Station for the software side. When you're ready, see pricing or get in touch for a demo against your own procurement scenario.