For the past couple of years, the cheapest way to knock out a slice of Europe’s internet has been a rusty tanker and a dropped anchor. Drag it across the wrong stretch of seabed, snap a cable or two, and by the time anyone notices, the ship is over the horizon and the official explanation is usually that the anchor slipped.
The response so far has mostly happened on the surface. Patrol boats, maritime aircraft, the occasional robot submarine sent down for a look. What almost nobody has done is actually listen to the seabed itself.
A three-person startup out of Copenhagen wants to fix that. On July 14, Triton Depth said it had raised a €1 million pre-seed round to build a network of passive underwater microphones that sit on the bottom and wait, designed to hear a submarine, a diver or a stealthy drone moving through the dark long before it reaches whatever it came to cut.
The cash came from Denmark’s state-owned Export and Investment Fund, known as EIFO, and London’s The Creator Fund, according to Defence Industry Europe. It’s a small round for a very big idea, and the idea holds up better than the check size suggests.
The seabed is the one battlefield nobody is really watching
We put a lot of effort into watching the sky. Radar, counter-drone systems, layers of air defense. We patrol the surface too, with ships and aircraft and satellites. The bottom of the ocean is the part almost nobody covers, and it happens to be where the important stuff runs.
Roughly 99% of the world’s intercontinental data rides on subsea cables, and there are well over 500 of them lying on the seafloor, alongside the power links and gas pipelines that keep the lights on across northern Europe. That’s a lot of critical infrastructure sitting in water that, until recently, nobody was watching at all.
The threat stopped being theoretical a while ago. NATO stood up a permanent Baltic patrol called Baltic Sentry in January 2025 after a run of cable and pipeline damage, and the seabed has become crowded with hardware since. The US Navy ran its own uncrewed drones across the Baltic floor off Latvia during this summer’s exercises, and Italy has since put an integrated seabed-monitoring network in the water aimed at spotting intruders from a long way off.
It isn’t just a Baltic problem, either. Triton points to the shadow fleet of tankers running with their transponders off, and to the Strait of Hormuz, which saw Iranian naval-mine activity during this year’s Gulf conflict. Jamie Macfarlane, CEO of The Creator Fund, put the blind spot bluntly: “Humanity has better maps of Mars than of the ocean floor.”
What all of it has in common is a response built around chasing the problem, sending expensive platforms out to have a look, rather than listening for it around the clock. That gap is exactly what Triton is aiming at.
How you catch a saboteur by listening instead of chasing
The system is built around small autonomous sensors the company calls Triton Nodes. Strip away the branding and each one is a hydrophone, an underwater microphone, wrapped in enough electronics to think for itself.
The nodes are designed to sit on the seabed or hang in the water column and stay there for long stretches with barely any maintenance. They’re passive, which is the whole trick. Active sonar works by pinging the water and reading the echo, but a ping also announces exactly where your sensor is. A passive node makes no sound at all. It just listens.
TechRadar framed the whole idea as a kind of CCTV for the seabed, which captures the gist even if there’s no camera anywhere in it. The nodes see with their ears.
When something moves nearby, an onboard AI model reads the acoustic signature in real time, right there on the device rather than in some distant data center, and tries to classify it. Is that a cargo ship, a fishing boat, a diver, or a drone that has no business being there? If it’s the last one, the node flags it and pushes an alert to whoever is watching.
Spread a bunch of them out and the numbers start to work together. Multiple nodes hearing the same contact can triangulate roughly where it is, and the company describes the whole arrangement as a kind of tripwire with no single point of failure. Knock out one node and the net still holds.
That’s a different philosophy from the machines already down there. Germany’s submarine-sized robot that hunts other subs goes out actively looking for trouble. Triton’s bet is a scatter of cheap, silent ears that never move and never stop. One approach chases. The other waits.
Why cheap and passive is the entire pitch
The hunters are impressive machines, and that’s also their limitation. You cannot park a robot sub that costs as much as a small warship over every cable, pipeline and harbor entrance. There aren’t enough of them, and there never will be.
The way the seabed gets monitored today, in the company’s telling and its backers’, is with systems that are too expensive, too sparse, or too slow to scale. A distributed net of cheap listening nodes flips that math. Instead of a handful of prized assets covering a fraction of the water, you blanket the areas that matter and let the cheap stuff do the watching.
It also cuts both ways commercially. The same acoustic data that tells a navy something is wrong can tell an offshore wind operator that a boat is loitering over its cables. Triton is pitching it as dual-use from the start, defense and civilian on the same hardware.
And there’s a longer game underneath the sensors. Macfarlane argues the company that wins this space will be the one with the strongest acoustic data advantage, because a giant, well-labeled library of underwater sound is what actually trains the classifier to tell a threat from a fishing trawler. Borg frames the ambition as building “the intelligence layer for the ocean.” The €1 million is really funding that: gather the data first, expand the coverage later.
A three-man company, a million euros, and nothing in the water yet
Triton Depth was founded in 2025 by Carl Borg, Markus Rytter and Frederik Himmer, who met as engineering students at the Technical University of Denmark and now run the company as CEO, CTO and COO respectively.
They’ve picked up some early validation along the way. DTU named them Student Startup of the Year in April, which came with a 25,000-kroner prize, and they landed a Heyfunding grant before this round closed. Having Denmark’s own state investment fund lead the pre-seed is a real signal too, not a random angel writing a check. Sara Sande of EIFO framed underwater awareness as a strategic necessity for Denmark and Europe, which is the kind of language that tends to unlock follow-on money.
The check size is the tell. This is a pre-seed, and the money buys exactly what Borg says it buys: accelerating development, expanding deployments and building the dataset. That’s founder-speak for “we’re at the beginning.”
There’s no deployed network at scale to point to, and no published figures on how far a node can hear, how accurately it classifies, or how often it cries wolf. Those numbers are the difference between a clever idea and a working system, and they don’t exist in public yet. What Triton has right now is a real problem, a sensible approach to it, and a million euros to find out whether the approach survives contact with a noisy, contested ocean.
A good bet that still has to survive the ocean
The stakes here aren’t abstract. The power in your wall and the data on your phone both cross that dark seabed, and right now the bottom is the least-watched part of the whole chain.
The bet Triton and its backers are making is a good one on paper. Listening is cheaper than chasing, and a thousand cheap ears really can beat one expensive hunter, if the ears work. Whether a scatter of student-built microphones can reliably pick a stealth drone out of the racket of a working ocean, all that engine noise and weather and marine life, is the thing that has to be proven. A €1 million pre-seed is the money you raise to start proving it, not the money you raise because you already have.
The seabed is still mostly deaf. Triton’s whole pitch is that it can hand it a set of ears before the next anchor slips.
