A Teen Founder Wants to Make AI-Guided Missiles Cheaper

Henrijs Dandzbergs is 16, has already been accepted into Y Combinator twice, and is now building Polemos Industries, a startup developing cheaper, AI-guided missiles designed to resist electronic jamming. The company says its guidance software is ready and hopes to demonstrate a prototype by December. The technology is still unproven, but the idea raises some rather uncomfortable questions about the future of autonomous warfare, especially if sophisticated weapons become significantly cheaper and easier to manufacture.

WTF Index TERMINATOR
◄ Terminator 5 Idiocracy 0 ►

The story centers on cheaper AI-guided missiles and autonomous warfare, increasing the potential for lethal harm.

A Teen Founder Wants to Make AI-Guided Missiles Cheaper

At 16, most people are still trying to figure out what they want to do when they grow up.

Henrijs Dandzbergs seems to have skipped that part.

He’s already founded a company, got accepted into Y Combinator (twice, actually), and is now working on building AI-guided missiles that are cheaper, easier to manufacture, and resistant to electronic jamming.

The company is called Polemos Industries, and its business model is quite straightforward: take something that is currently very expensive, figure out how to make it considerably cheaper, and sell it to customers who have a very good reason to care about the price.

Pretty much the same logic behind thousands of startups.

Except we’re talking about missiles here.

And while most companies would probably describe their mission using something like “democratizing access to advanced technology”, Polemos went with a considerably more direct approach.

According to its investor one-pager, the goal is to reduce the “cost per kill”.

Well… at least nobody can accuse them of using vague marketing language.

From Raspberry Pis to actual missiles

Dandzbergs was born in Latvia and spent much of his childhood moving between Latvia, Indonesia, Australia, and Spain, with most of his early education taking place online.

He was interested in science and mathematics from an early age, and his room apparently looked more like an electronics workshop than a typical teenager’s bedroom, filled with robotics equipment, schematics, Raspberry Pis, and 3D-printed parts.

When his family returned to Latvia during Covid, he attended Astra Nova, an experimental school focused on lateral thinking and problem-solving.

A class taught by a NASA employee got him interested in aerospace engineering, and he eventually started working on projects involving a bionic arm, a thermal control system for a model rocket, and sensors for a self-flying drone.

So, clearly, engineering wasn’t a particularly new interest.

But there’s quite a difference between building a self-flying drone and deciding to manufacture guided missiles.

And that part of the story has a lot to do with where he grew up.

Latvia shares a border with Russia, and when Russia invaded Ukraine, the possibility of a military conflict became considerably more real for people living in the region.

Dandzbergs recalls seeing military jets flying overhead and hearing about stray missiles being intercepted at the Russian border.

His stated motivation is to help defend his country and its allies, while also trying to understand where autonomous warfare is heading.

Which brings us to the interesting part.

Getting into Y Combinator is difficult. Apparently, doing it twice is an option.

In April 2026, Dandzbergs traveled to London for a hackathon organized by Unicorn Mafia.

Together with Aleksandrs Tihenko and Kai Yang, he built a system that could use AI to generate hardware circuit board designs from natural-language instructions.

Basically, instead of manually designing a circuit board from scratch, you could describe what you wanted to build and have the system generate a design for it.

The team managed to create a working prototype in less than 48 hours.

They also won the hackathon.

Not bad for a weekend.

A week later, the three founders pitched their project to Y Combinator and were accepted into its summer program.

Now, getting into YC is already quite an achievement, especially for someone who isn’t old enough to vote.

But the three founders had different ideas about where the company should go.

Dandzbergs wanted to pursue defense technology. His co-founders didn’t agree with that direction.

So he left.

And instead of continuing with the project that had already secured a place in one of the world’s best-known startup accelerators, he decided to apply again.

This time, alone.

With a completely different company.

One that builds missiles.

And he got accepted again, into Y Combinator’s fall 2026 cohort.

I imagine that makes for a rather interesting school career counseling session.

Although, to be fair, the accelerator itself had some serious concerns.

Dandzbergs was considering leaving high school to focus on his startup, and YC had to decide whether supporting such a significant commitment was appropriate for someone his age.

There were also questions about whether the proposed technology could actually work.

Harj Taggar, the YC partner assigned to mentor him, acknowledged the responsibility that comes with influencing such an important decision in a young founder’s life.

Eventually, they decided to back the project.

And that’s how a 16-year-old ended up building a missile startup with Y Combinator’s support.

The idea is simple. Making it work is another story.

Modern precision-guided weapons can be extremely expensive.

Meanwhile, the war in Ukraine has demonstrated how relatively inexpensive drones can have a significant impact on the battlefield.

The obvious question is whether some of the capabilities traditionally associated with more expensive missiles could be delivered at a much lower cost.

That’s essentially what Polemos is trying to do.

The company wants to develop precision-guided missiles that are cheaper and easier to manufacture than conventional large rocket systems, can be launched from portable equipment, and can resist electronic jamming.

The missiles are also intended to offer greater precision and carry larger warheads than the drones commonly used in Ukraine.

Which sounds like a pretty impressive combination.

There’s just one small problem.

Every single one of those requirements introduces engineering challenges, and making the final product cheaper doesn’t exactly make any of them easier.

Tom Karako, director of the Missile Defense Project and a senior fellow at the Center for Strategic and International Studies, points out that reducing missile costs generally involves trade-offs.

Range, payload capacity, and reliability can all be affected.

And that’s before accounting for the additional complexity of a sophisticated guidance system.

You can reduce manufacturing costs, simplify components, or change the design, but at some point you still need the thing to perform reliably.

Especially when the intended application doesn’t leave much room for technical failures.

Polemos also has another problem: competition.

Anduril, CoAspire, and Leidos are already working on more affordable missiles that can be manufactured at scale.

These are established defense companies with significant resources and experience.

So it’s not as if Dandzbergs has discovered a problem nobody else noticed.

The problem is very real, and several companies are already trying to solve it.

His challenge is figuring out whether Polemos can do something meaningfully better, cheaper, or different.

Preferably all three.

What has Polemos actually built?

According to Dandzbergs, the company has developed the software system intended to guide its missiles and has started working on the hardware design.

The next objective is to demonstrate a prototype missile by the end of Y Combinator’s program in December 2026.

And it’s worth emphasizing that this is the objective, not a completed demonstration.

At this stage, the technical ambitions of Polemos remain unproven.

Having guidance software is an important step, but a missile is considerably more than a piece of software.

There’s the hardware, the guidance system, reliability under real-world conditions, production costs, and the rather important detail of making all those things work together.

At the price you’re promising.

It’s a complicated engineering problem even for companies that have been working on this technology for years.

And while those are interesting challenges on their own, they’re not necessarily the most important questions surrounding the project.

Making weapons cheaper is not quite the same as making the world safer

Let’s assume Polemos succeeds.

Not just with a prototype, but with the actual objective of producing sophisticated, AI-guided missiles at a fraction of the cost of conventional systems.

That could have some very real advantages for countries trying to defend themselves.

Precision weapons are expensive, military budgets are limited, and having access to capable defensive technology without spending enormous amounts of money could make a significant difference.

Especially for smaller countries.

So there’s a perfectly reasonable argument in favor of making effective defensive systems more affordable.

But there’s another side to it.

If a technology becomes cheaper and easier to manufacture, the barriers to producing it become lower.

And that’s not necessarily something you want to happen without considering who could eventually gain access to it.

Particularly when the technology in question is designed to deliver lethal force.

Then there’s autonomy.

AI-guided doesn’t automatically mean a weapon independently decides who to attack. Those are different capabilities, and the available information doesn’t establish that Polemos has developed a fully autonomous targeting system.

But the broader move toward increasingly capable AI systems in warfare raises questions that are difficult to ignore.

Who is responsible when something goes wrong?

What happens if a system makes an incorrect decision?

Could a technical failure result in friendly fire?

And how much control should we be comfortable delegating to software when the consequences can be irreversible?

Dandzbergs says he has considered these issues.

He describes his motivation as protecting the United States and its European allies.

And there is no reason to assume that isn’t his genuine intention.

But good intentions don’t solve engineering problems, and they certainly don’t eliminate the risks associated with increasingly accessible military technology.

There’s also an interesting contradiction here.

The company presents its work as a way to improve defense capabilities, while its own investor material describes success in terms of reducing the cost per kill.

Both can be true, of course.

But the contrast is worth thinking about.

So, what does this actually mean for the future?

The most interesting part of this story isn’t necessarily that a 16-year-old wants to build missiles.

Although that’s certainly not something you come across every day.

It’s the fact that the tools, knowledge, and development resources required to attempt projects like this are becoming increasingly accessible.

Not that building an actual missile is suddenly easy. Polemos still has an enormous amount to prove, and established defense companies are working on many of the same problems.

But consider how the story started.

A teenager interested in engineering, building things with Raspberry Pis and robotics equipment, eventually develops an AI system that designs circuit boards, gets accepted into Y Combinator, and decides to start a company focused on precision-guided weapons.

There’s an interesting progression there.

AI is making increasingly sophisticated engineering tasks more accessible, and startups are exploring areas that once required considerably larger organizations and resources.

That can be an incredible development.

The same capabilities that make it easier to create useful technology can also lower the barriers to developing things we’d probably prefer to keep under very strict control.

And unlike a chatbot generating questionable answers or an AI image with six fingers, the consequences of getting something wrong in this industry can be very real.

Polemos is still an early-stage startup. Its prototype hasn’t been demonstrated in the reporting available, its proposed cost advantages remain unproven, and there are significant technical and competitive obstacles ahead.

It might succeed.

It might discover that building affordable precision-guided missiles is considerably more complicated than the original plan suggested.

Either way, the underlying trend is worth watching.

Because even if this particular startup never manages to produce a commercially viable missile, the push toward cheaper, more capable, AI-assisted weapons isn’t going away.

And it’s probably something we should pay attention to before those systems become commonplace.

WTF Index: Terminator 5 / Idiocracy 0

This development receives a Terminator score of 5 because it represents an attempt to make sophisticated AI-guided weapons cheaper, easier to produce, and potentially more accessible.

The concern isn’t that the founder is 16, nor is it a prediction that Polemos will successfully deliver the technology.

It’s the broader direction of making increasingly capable lethal systems more affordable, and the questions that raises about control, proliferation, reliability, and the future of warfare.

There’s no meaningful Idiocracy component to the development, which is why it receives a score of 0 on that side of the index.