I Think I Found Something

Happy Friday folks - 

Want to share something I've been thinking about throughout the week: Drone Infrastructure as a Service.

As I think through the Aerial Economy, I'm imagining the application layer.

We're seeing a few unique use cases for drones + aerial technologies, but we're not even in the first inning of this trend.

It's so early.

Software as a service did not win because of monthly billing.

It won because one instance served thousands of customers at once. Build it once, serve the next customer for almost nothing. Median software gross margin held at 80% in 2025 across 342 companies, per the Aleph and Benchmarkit benchmarks.

Multi-tenancy was the innovation. Subscription pricing was the receipt.

Then robots as a service copied the playbook.

In December 2024, Serve Robotics announced it could self-fund equipment investments, eliminating the near-term need for equipment financing.

Read that twice. Funding a depreciating fleet with equity, announced as a win.

Shares went from roughly 51.5 million at the end of 2024 to about 67.8 million by the third quarter of 2025. Today, roughly 2,000 robots across 44 cities, with 2026 revenue guidance near $26 million.

About $13,000 per robot per year. My math, not theirs.

One robot. One delivery. One customer at a time. The marginal cost of the next customer is another robot.

That is not software. That is leasing.

Here is where drones play their part.

A robot is a thing. A node is a place.

Fire district in the morning. Utility inspection at midday. Insurance carrier flying hail claims in the afternoon. Rancher checking stock water at dusk.

None of them can justify the node alone. Together, underwriting is easy.

The demand is growing - exponentially. 

Over 1,000 public safety agencies held the FAA Part 91 waivers needed for a Drone as First Responder program as of February 2026. Only 976 had been granted from 2018 through April 2025.

Permission and regulation stopped being the bottleneck. 

It's capital.

A docked Drone First Responder program runs $25,000 and climbs beyond $60,000. 

There are thousands of agencies with a hunting license and no rifle. 

Cell towers are the model.

One tenant returns about 3%. Nobody finances 3%. Add a second and a third, operating costs barely move, no new capital goes in, and the tower sells for 30 to 40 times cash flow.

Same steel. More tenants. 

American Tower does not manufacture towers. Somebody welds steel at a manufacturing margin. Somebody else owns the site and collects the rent.

Two businesses. Two multiples.

This is the layer I want. 

Now the challenging part - a carrier’s antenna is passive. 

Four carriers share a tower with zero contention. 

A drone can only be in one place at a time. 

So multi-tenancy has to come from capacity, not scheduling. More bays. More aircraft. 

I don’t have a clear read on that yet. 

Maybe it's a mobile drone factory. Maybe it's something else. Time will tell. 

Most drones today fly under Part 107. The aircraft stays within sight of a human. Beyond that, you need a waiver, one at a time.

Part 108 replaces that. A standing rule for beyond visual line of sight, up to 1,320 pounds, with operating area approvals instead of per-flight paperwork.

Here’s why it matters to a node. 

With visual line of sight, the droneport is a garage. 

Beyond visual line of sight, the node serves a radius. It becomes a hub. 

Part 108 reached OIRA on July 10, 2026. Review can run 90 days. Energizing a site takes about 18 months. Construction is 4 to 8 weeks. The rest is permits and the utility queue. 

If we start now, the waiting is free.

If we wait for the rule, start the day everyone else does, we’re late.

Having fun yet? 

I texted my manufacturing partner about the concept. His response:

"What prompted this thought?"

A few days later:

"The droneport as a service model is pretty bonkers. You're properly onto something!!"

I’m still flying blind, building in uncharted waters. But how does that saying go… 

If you build it, they will come.