Sebastian Asprella

Accelerating EU and protecting the skies from rogue drones

[Sebastian Asprella @SebaAsprella]

Europe can't build drone-detection radar fast enough.

One short-range radar can take two years to deliver, and everyone is queuing for the same handful of suppliers. The instinct is to fight for a place in line. Order earlier, pay more, lobby for priority.

Wrong move. If radar is the scarce thing, the win is getting more coverage out of every radar you can actually get.

That is what we build: a mesh where the node next door still sees the drone that vanished in a building's shadow, and coverage grows by adding cheap nodes instead of buying one scarce expensive box.

When supply is the bottleneck, the architecture that multiplies coverage per radar is worth more, not less.

[Sebastian Asprella @SebaAsprella]

@wiredcoach @sairahul1 🎯

[Sebastian Asprella @SebaAsprella]

@Shail_2302_ @garrytan I stack Claude Design on top, perfect trifecta

[Sebastian Asprella @SebaAsprella]

Sovereign sensing is not the same as redundant sensing.

A redundant sensing system has two of everything. Two radars, two operator stations, two data feeds. If one fails, the other carries the load. This is a resilience pattern.

A sovereign sensing system has different requirements. The supply chain for the sensors must be built and maintained inside the buyer's economic zone. The firmware and signal-processing stack should be legible to the buyer's engineers and auditable by the buyer's regulators. A failure-diagnosis path must be runnable by the buyer without an external party's data feed. Component replacement should be available without a customs broker. This is an autonomy pattern, not a resilience one.

The two are not the same. A redundant system supplied by a vendor whose firmware updates depend on overseas approval is not sovereign in any operationally meaningful sense. A single sensor whose entire stack is in-country, auditable, and field-repairable is more sovereign than five redundant boxes from outside the perimeter.

Europe is currently buying for redundancy and calling it sovereignty. The next test of the distinction will arrive the first time a buyer has to do a forensic on a sensor failure during a live incident and cannot get answers from the supplier in the time the incident allows.

The question worth asking, before the next procurement round, is which of the two is actually being bought. And whether the answer matches what was paid for.

[Sebastian Asprella @SebaAsprella]

@visegrad24 This is heartbreaking, rip Henry

[Sebastian Asprella @SebaAsprella]

You do not need obsidian to build your company OS

[Sebastian Asprella @SebaAsprella]

Europe is putting money behind its own defence sensing layer.

In the past 12 months: the EU's €1.07B EDF round (57 projects). The Action Plan on Drones and Counter-Drones, €250M. National-level capital starting to flow. Luxembourg launched a €150M national defence fund last month (€75M SNCI, €50M state, €25M FSIL; targets cybersecurity, space, advanced sensing, dual-use materials).

The Luxembourg fund is small in absolute terms. It matters because it is one of several national programmes activating in parallel across Europe, in the same window EU-level instruments are scaling.

The question that follows for operators and procurement teams: when the budget arrives, are the architectures being bought sovereign in any operationally meaningful sense, or just redundant? That distinction is harder to fix once a programme is committed than to think through before signature.

Capital availability is the precondition for sovereign sensing infrastructure. It is not the same as having one.

[Sebastian Asprella @SebaAsprella]

@ycombinator @LightconePod @koomen Would be great to see it live over a demo..

[Sebastian Asprella @SebaAsprella]

Preliminary results are not pilot-grade evidence.

Four radars meshed at a Luxembourg site this past Monday. Small drone detected combining tracks across the mesh. Two methodology calls that made the difference: a co-located baseline calibration before deployment (all radars side-by-side same direction, 30 seconds of capture, cross-checked for divergence so sensor variance is separated from scene geometry), and a static-clutter masking pass on the angular and range zones where ground metallic returns were creating spurious detections. After masking, real targets show cleanly. Both methods are currently manual. Both are obvious Mark III workstreams.

What this is. Field evidence that the four-radar mesh architecture detects, that the calibration method is repeatable, that the masking approach handles cluttered industrial environments without losing real targets.

What this is not. A pilot. A pilot has signed acceptance criteria, a defined duration, a customer's procurement team in the room, and a path-to-deployment clause if the metrics are met. None of those existed Monday. The field-test was an engineering proof-of-architecture, not a buyer-validated trial.

The discipline that matters at this stage is being clear which is which. Founders who let "preliminary results" inflate to "pilot evidence" in investor and customer conversations damage their credibility before the next milestone has even arrived. Founders who keep the narrow word for the narrow thing get to make the next claim cleanly.

What comes next: signed pilot acceptance criteria with two named lighthouse customers, then the actual paid pilot, then the deployment. None of those are this. This is the engineering work that makes the next step possible.

[Sebastian Asprella @SebaAsprella]

Drones, robots and autonomous systems do not scale safely just because the vehicle gets better.

They need an operating layer around them: sensing, identity, airspace awareness, reliability data, regulation, and trust.

Zipline built the logistics fabric around delivery.

Europe will need the sensing fabric around low-altitude autonomy.

[Ole Lehmann @itsolelehmann]

This is how your next Chipotle burrito, online package, and medical prescription will be delivered.

And it might be my favorite company in all of robotics right now.

Zipline has done over 2 million autonomous drone deliveries to date.

That's more than every other drone delivery company on Earth combined.

For reference, Alphabet's Wing has done about 500,000.

So obviously the first question is: how did a 10-year-old company outscale Google at autonomous drone delivery?

The answer is one of the smartest strategic bets I've seen in tech.

Most drone delivery companies started by trying to drop pizzas in American suburbs.

Because low stakes, easy customers, friendly weather, etc.

Zipline went the opposite direction.

They started in rural Rwanda, flying emergency blood to remote clinics where deliveries by truck could otherwise take 3+ hours.

It looked like a terrible idea in 2016.

But it turned out to be the smartest decision any drone company has ever made.

Because if you can build a system reliable enough to deliver blood to a dying patient in a monsoon in a 3rd world country...

Then delivering a Chipotle order to a suburban backyard becomes trivial.

The hardest use case became the moat.

Every single one of those flights made their system smarter, faster, and more reliable.

That compounded. Hard.

The result, 10 years in, is a lead that's basically structural:

> 2 million deliveries done
> 5,000 hospitals served across multiple countries

The first million deliveries took them 7 years.

The second million took less than 2.

US deliveries growing 15% week-over-week for 7 straight months.

Now they're coming for everything.

And the tech they've built to pull it off is genuinely science fiction.

Their newest drone, called the "Zip," hovers a few hundred feet up while a smaller "Droid" gets lowered on a tether.

The Droid uses tiny thrusters to position itself precisely over a target zone the size of a picnic table.

It drops the cargo, then gets winched back up into the Zip.

And the whole thing is so quiet that customers couldn't tell when their order had arrived.

So Zipline literally had to add an audible beep on approach.

The Zip then flies to one of Zipline's autonomous charging stations, which look like streetlamps with an arm and a disc.

As Zipline rolls these out across cities, the coverage compounds automatically.

And the system uses 97% less energy than a gas-powered delivery truck.

Which makes obvious sense when you think about it. A delivery truck weighs 3,000 pounds. The average package weighs under 5.

And the scale of what they're building toward is much bigger than just food delivery.

In November 2025, the US State Department signed a $150 million contract with Zipline to expand drone delivery across Africa.

It's the first pay-for-performance contract the State Department has ever signed, anywhere.

At full scale, the deal will triple the hospitals Zipline serves and give 130 million people instant access to blood and essential medicine.

In the US, partnerships are stacking up fast:

Walmart, Chipotle, sweetgreen, Panera, Cleveland Clinic, Mayo Clinic.

They're already in Dallas and Arkansas, expanding to Houston and Phoenix this year, with at least 4 more states by year end.

And a June 2025 Executive Order on drones gave them operating rights across all 50 states.

The regulatory groundwork for nationwide deployment is basically done.

Their ultimate vision:

Every package under 8 pounds (food, medicine, household items) delivered by silent autonomous drones in 3 minutes instead of 3 days.

A network of streetlamp-sized charging stations across every city, with drones flying between them based on real-time demand.

In rural areas, every remote clinic gets the same access to medicine as a hospital in a capital city.

In American suburbs, your Chipotle order arrives from the sky in 3 minutes.

In Zipline founder @Keller's own words:

"It's very obvious that whoever succeeds will be one of the largest companies on Earth. Bigger than UPS and FedEx combined."

And the insane part is that this isn't a 5 or 10-year prediction.

It's already happening at scale, in cities you can actually visit, today.

This is the infrastructure of the future being built right now, by a company most people in tech still couldn't tell you the name of.

Oh yeah, and they just closed an $800M Series H at a $7.6B valuation.

Gud tech.

[Sebastian Asprella @SebaAsprella]

@TaylorCSargent @AmcaInc Very few people operate like Jai. Outstanding founder operator and outstanding human being. Congrats to Jai and the Amca team

[Sebastian Asprella @SebaAsprella]

@hollybynight11 @Defence_Index Merci. Was reading the cost per hour for the jet is around $70k, so it’s about $100k+ per drone. Still unsustainable

[Sebastian Asprella @SebaAsprella]

@NXT4EU @MarvinTBaumann The European upside is so vast, that brings tremors to competitors

[Sebastian Asprella @SebaAsprella]

@MPrinParr A sensing layer across all critical infrastructure seems inevitable

[Sebastian Asprella @SebaAsprella]

@SwenRoschlau We already are, Swen 💪

[Sebastian Asprella @SebaAsprella]

@SwenRoschlau There is so much untapped upside in Europe

[Sebastian Asprella @SebaAsprella]

The job is not to do what you like.

The job is to become very good at what the mission requires.

Especially the parts you dislike.

When instinct is weak, facts become the operating system.

Facts beat ego. Facts expose gaps. Facts make discomfort executable.

Defence deeptech rewards execution under constraint.

[Sebastian Asprella @SebaAsprella]

@SwenRoschlau @business Inevitable capability needs

[Sebastian Asprella @SebaAsprella]

Entrepreneurs are social benefactors. We’re all communicating via this stack of tech, for example, thanks to entrepreneurs

[Palmer Luckey @PalmerLuckey]

I got a lot more than that, I don't think any reasonable person could read what I wrote and think I am pretending it was small. My whole point is that I did create a massive $1B+ amount of value on my own, without China, yet you insist I did nothing.

Your argument is so easy to make if you admit there are exceptions, but for some reason you are compelled to double and triple down on blatantly false statements. Consider that maybe the world isn't as black and white as you think.