Four Drone Business Models, and Why Selling Airframes Is the Worst One

🌏 中文版

The industry map argued value concentrates in Layers 3 and 5; the Taiwan supply chain piece showed Taiwan sitting in Layer 2. This one supplies the reason: selling airframes is the worst of the four available business models.

Four models, four completely different financial shapes

How big is the gap? Per Pulse's commercial UAV manufacturing KPI benchmarks:

The durable margin is in software, not airframes. Hardware gross margin runs 35-55% and is under permanent price pressure. Autonomy software, fleet management, data pipelines, and Drone-as-a-Service subscriptions run 60-80% margin and recur.

The same benchmark offers a sharper test:

A manufacturer that ships 800 units but attaches autonomy software to only 200 of them has a worse business than a competitor shipping 500 units with 90% software attach. Unit count alone is a vanity metric.

Worth writing down, because Taiwan's industry narrative — "NT$12.9 billion of output," "123,000 airframes exported" — is built entirely on unit counts and output value, not revenue structure.

Model one: hardware — a ceiling DJI defines

Hardware is the most intuitive model and the hardest to defend. The problem isn't building it; it's that someone else sets the price.

The cycle history piece covered how 2016 ended: DJI's vertical integration and manufacturing automation let it profit at price points competitors couldn't survive, clearing out the US and European consumer field within a year. That cost structure still exists.

For Taiwanese manufacturers: the non-Chinese supply chain gave you orders, not pricing power. Certification settles whether you can sell, not how much you can charge. When qualified US/EU/Japan suppliers go from three to ten, the price war restarts inside the certified list.

Model two: service (DaaS) — customers want the outcome, not the aircraft

The core insight is simple: most enterprises don't want to buy drones, they want the job done. Buying aircraft, training pilots, handling regulation are all costs they'd rather not touch.

Market estimates put global DaaS at roughly $6–8 billion today, with most projections reaching $15–20 billion+ within the decade.

India's Garuda Aerospace is a concrete case of this path working. Per Unlisted Network's analysis, its FY24 revenue split was roughly 62% DaaS to 38% drone sales, with revenue growing from about ₹47 crore in FY23 to ₹117 crore in FY25.

The cash-side comparison is more interesting still: Garuda's cash conversion cycle ran about 351 days against defense-focused peer IdeaForge's roughly 597 days. Same industry, same country — the business model difference shows up directly in capital efficiency. The long receivable cycle on defense contracts is this industry's most underrated cost.

Model three: software subscription — turning one-time revenue into an annuity

The logic: the hardware sale is the beginning, and the real value is in the subscription years that follow.

Skydio is the clearest sample. Per Sacra's estimates, 2024 revenue was about $180 million (up ~80% from $100M+ in 2023), with software subscriptions already ~30% of total revenue as of 2023, a blended gross margin of ~38%, and about $1.2 billion in bookings with over half from defense.

That 38% blended margin says something important: even with software at 30% of revenue, hardware still drags the blend down. Which is why the KPI benchmark above insists on reporting hardware and software margin separately — blended, you can't see whether the flywheel is turning.

(Sacra's figures are a 2024 snapshot. By April 2026, Skydio announced a Series F at a $4.4 billion valuation and described a core business generating "hundreds of millions in annual revenue" — the flywheel is turning. But the company has not disclosed a margin split or growth rate since, so the 2024 figures remain the citable structural data.)

Software's strategic value isn't only margin, it's switching cost. Once a customer embeds drones into daily workflows, changing vendors means redoing processes, retraining people, and reintegrating systems. Hardware alone can't produce that stickiness.

Model four: data and analytics — highest margin, furthest from the airframe

Market analysis notes that companies like DroneDeploy and Pix4D sell SaaS subscriptions that convert flight data into usable conclusions — construction progress tracking, volumetric measurement, vegetation health indexing. They don't necessarily manufacture drones, yet they occupy the highest-margin position in the chain.

The barrier here isn't flight technology, it's vertical domain knowledge. Construction progress analytics requires understanding project scheduling; vegetation health requires agronomy. Which is why hardware makers can't casually absorb this layer.

Where Taiwan sits, and where it could move

Put Taiwan on this table and the position is unambiguous: almost entirely model one.

• Airframe and module manufacturing → model one
• 80% of output from public-sector and defense procurement → the longest-payment-cycle variant of model one
• DaaS, software subscription, data analytics → a few startups, all small

And Taiwan's domestic service market is held down by regulation — the BVLOS comparison showed Taiwan has no standardized beyond-visual-line-of-sight framework, so delivery and long-range inspection can't scale. Model two's ceiling in Taiwan is neither technology nor capital; it's regulation.

What's left? Two realistic paths:

Path one: move to model three, but sell to foreign airframe makers. Taiwan has the ICT and embedded software base; flight control software, fleet management, and edge AI modules can all be sold to run on someone else's hardware. This path doesn't require Taiwanese regulation to change — you're selling into someone else's jurisdiction.

Path two: inside model one, pick the highest-margin cells. Selling hardware isn't uniform. Layer 3's flight computers, jam-resistant comms modules, and EO/IR payloads carry far better margin and substitution resistance than Layer 2's motors and frames. That is exactly the logic behind the "three chips, two software" policy bet.

Both paths point the same direction: away from the airframe.

Three questions for judging any drone company

1. What share of revenue recurs? Per the KPI benchmark, leading manufacturers push this toward 30–50% of total, while legacy OEMs sit at 5–10%. That number determines whether the company is valued as hardware or software.
2. Are hardware and software margins reported separately? A rising blended margin can mean the software mix grew (good) or that hardware simply got cheaper to build (not good). Blended, you can't tell.
3. Is the customer base commercial or defense? The sales cycles and payment terms differ entirely — per the same benchmark, commercial deals run 3–9 months at 25–45% win rates; defense runs 12–36 months at 10–30%, with much longer receivables.

The bottom line

The money in this business is not in the thing that flies. Hardware is customer acquisition cost; software and services are where you recover it. The structure matches printers, razors, and plenty of other hardware-to-service industries — the difference is that drones are still early, and most players remain on step one.

Taiwan's current strength — fast, flexible airframe and module manufacturing — sits at the lowest-bargaining-power segment of that chain. Moving up means climbing to Layer 3 technically and converting one-time revenue into recurring revenue commercially. Those are two descriptions of the same move.

References

Business models and KPIs

• Pulse Industry KPIs — Commercial UAV & Drone Manufacturing sales KPI benchmarks (hardware vs. software margin ranges, recurring revenue mix, commercial vs. defense sales cycles)
• Smarter.com — Evaluating Drone Companies for Investment: Market, Business Models, and Trade-Offs
• Market Research Future — Drones Market Trends, Analysis, Revenue, and Forecast (DaaS and the analytics layer)

Cases

• Skydio — Strong Business, Bigger Mission, New Capital (2026 Series F)
• Sacra — Skydio revenue, funding & news (revenue estimates, software subscription share, gross margin, bookings mix)
• Unlisted Network — How Garuda Aerospace Quietly Became India's Agriculture Drone Leader (DaaS revenue share, cash conversion comparison)
• PR Newswire — Drones-as-a-Service Shifts from Emerging Tech to Revenue Engine (DaaS market sizing)

Tools and platforms

• DroneDeploy
• Pix4D

On this site

• The Drone Industry Map: Components, Regulatory Ceilings, and the Non-Chinese Supply Chain Rebuild
• Taiwan's Drone Supply Chain: Where the 267 Companies Are, and Which Layer They're Stuck On
• Drone Industry Cycles: How the 2016 Bubble Burst, and What's Different This Time
• BVLOS in Three Jurisdictions