Finding a submarine that doesn’t want to be found is one of the most expensive problems in modern defense. Norway spent most of 2025 shopping for an answer and picked at least five British-designed Type 26 frigates, a deal Breaking Defense put at roughly $13.5 billion, which works out to about $2.7 billion per hull. A Munich company called Helsing thinks the future of the hunt looks less like a 6,900-ton warship and more like hundreds of 132-pound (60 kg) gliders drifting along at walking pace, each one running an AI that was trained the way you’d train a chatbot, except on decades of recorded ocean sound instead of internet text.
The software is called Lura. The glider that carries it is the SG-1 Fathom. The pitch is a swarm of mobile listening posts that Helsing compares to a satellite constellation, except underwater and pointed at submarines. It would have been easy to file all of this under defense-show vaporware when it was unveiled at Portsmouth in May 2025. Thirteen months later, the Royal Navy has run a technology demonstration with it, the UK has built an entire North Atlantic surveillance program around the concept, and investors have reportedly pushed Helsing’s valuation to $18 billion. The render phase is over.
The hardware is a glider two sailors can carry
Start with the machine, because the numbers are almost comically small for the job description. The SG-1 Fathom is 6.4 feet (1.95 m) long, 11 inches (28 cm) across, and weighs 132 pounds (60 kg), light enough to launch off a quay or the back of a workboat. There’s no propeller. The glider shifts its buoyancy and lets its wings convert that vertical motion into forward travel at one to two knots, which sounds like a flaw until you remember the entire mission is to be quiet. It runs on a lithium-ion battery, patrols for up to three months at a stretch, and can stop moving altogether and hold position on the seabed as a fixed listening node.
Depth is the one spec where the company has been unusually candid about the gap between brochure and hardware. The gliders are designed to work down to 3,280 feet (1,000 meters), but program manager Katie Raine told Navy Leaders at a London defense show last September that the team was still pushing toward that mark: “We’re not there yet, but we are getting closer all the time.” You don’t hear sentences like that from defense contractors very often, so it’s worth flagging when one slips out.
Sitting on the bottom for weeks isn’t science fiction anymore. A Canadian hydrogen drone already parked on the seabed for 16 days straight to watch a single cable junction. What Helsing adds is volume: the Fathom is containerized, rail-launched from shore or from a ship, and designed from day one to be built in the hundreds rather than the dozens.
Lura is the actual product
The glider, in Helsing’s framing, is mostly a delivery vehicle. The thing being delivered is Lura, what the company calls a large acoustic model: the same architectural idea as the large language models behind chatbots, trained on decades of archived ocean sound instead of text. Sonar operators spend entire careers learning to pick a submarine’s signature out of whale song, shipping traffic, and rain hitting the surface. Lura’s job is to do that classification on the glider itself, at the edge, where underwater bandwidth is close to nonexistent.
The performance claims are the company’s own, so treat them accordingly. Helsing says Lura hears targets ten times quieter than what existing AI models can pick up, tells individual vessels apart even within the same class, and works up to 40 times faster than a human operator. None of that has been independently benchmarked in public. What has happened in public is a string of in-water tests that kept not failing, which is its own kind of evidence.
When a glider hears something worth reporting, it surfaces and sends a compressed contact report to command instead of streaming raw audio, a detail Helsing walked trade press through at the launch briefing. Then it slips back under and keeps listening. “Lura detects so our navies can deter,” is how co-founder and co-CEO Gundbert Scherf compressed the sales pitch on launch day, and for once the slogan matches the architecture.
If the idea of a self-organizing underwater sensor network sounds familiar, it’s the same problem Germany’s hydrogen-powered Greyshark attacks with maximum endurance per vehicle. Helsing’s answer runs the other way: stop building exquisite individual machines and start stamping out cheap ones in bulk.
Hundreds of gliders, one person at a desk
Scale is the whole argument. Helsing’s launch material describes hundreds of Fathoms deployed per mission, all running Lura locally, with a single operator at a maritime headquarters tasking and monitoring the lot. The company claims the result delivers intelligence at 10% of the cost of crewed anti-submarine patrols, the kind of figure that should always arrive with a company-claim asterisk, but the direction of the math is hard to argue with. A frigate is a couple hundred people, a helicopter hangar, and a couple billion dollars. A glider is 132 pounds and no crew at all.
The business model is the part that would have sounded absurd a decade ago: navies can buy the gliders and run them, or hire the surveillance as a contracted service. Submarine detection as a subscription, basically. Helsing built the system with Blue Ocean Marine Tech Systems, Ocean Infinity and QinetiQ, the last of which also happens to operate the UK’s underwater test range, which turned out to be convenient.
The timing isn’t subtle either. Severed cables and damaged pipelines in the Baltic have become a recurring news genre, NATO now runs standing patrols over its own seabed infrastructure, and companies are piling into seabed-watching drones from every direction. Persistent, cheap underwater listening went from niche requirement to procurement priority in about three years.
The British went from demo to factory in six months
The development pace is the strongest part of Helsing’s story, and it’s verifiable. The system was unveiled in Portsmouth on May 13, 2025. Helsing, Blue Ocean and QinetiQ then ran a three-month sprint: acoustic simulation first, then in-water testing at a Scottish loch, then multi-glider trials at BUTEC, the Ministry of Defence-owned test range off Scotland’s west coast, wrapped up in late July. Naval News reported the at-sea campaign complete that September, with the hardware deliberately stressed along the way. “We pushed SG-1 to its design limits quite deliberately,” Helsing’s maritime general manager Amelia Gould told the outlet.
Gould’s team also collected the unglamorous lessons you only get from salt water. The trial gliders flew 3-D printed wings, iterated fast off hydrodynamic simulations; production wings won’t be printed, because the real ocean chews on them in ways the simulation didn’t predict. That detail matters more than any spec sheet, because it means production engineering is actually happening.
And it is. Helsing opened an 18,000-square-foot Resilience Factory in Plymouth in November, its first in the UK, with Defence Secretary John Healey cutting the ribbon and production already underway, part of a £350 million ($458 million) commitment Helsing made under the 2024 UK-Germany Trinity House Agreement, according to Janes. Army Recognition, citing the UK Ministry of Defence, reported Royal Navy test deployments in the Hebridean Sea in both single-glider and swarm configurations, with full operational capability targeted for the end of 2026.
Then December got busy. On December 8 the UK formally launched Atlantic Bastion, a North Atlantic undersea surveillance network that pairs autonomous systems with frigates, attack submarines and P-8A patrol aircraft, with the SG-1 Fathom as its most visible new hardware. Defence Minister Luke Pollard told Parliament that Helsing had been selected for a Royal Navy technology demonstration that same month, while a competitive process for the program’s first phase runs in parallel. Translation: trials, yes; signed production orders, not yet.
Helsing’s stated aim, per Naval News, was gliders available for deployment in 2026. We’re halfway through 2026, the Hebridean tests have happened, and the Royal Navy spent late April hosting demo days for more underwater surveillance suppliers as Atlantic Bastion advances. The race now is between Helsing’s production line and the MoD’s procurement paperwork, and only one of those is famous for speed.
The money treats this like infrastructure
The funding rounds say investors see a buildout, not a bet. In June 2025 Helsing closed a €600 million Series D ($693 million at the time) led by Prima Materia, the investment vehicle of Spotify founder Daniel Ek, who also chairs the company, at a €12 billion valuation, roughly $14 billion. Eleven months later, TechCrunch reported Helsing was closing in on a fresh $1.2 billion round at roughly an $18 billion valuation, led by Dragoneer with Lightspeed co-leading, oversubscribed multiple times according to the Financial Times’ sources, with the company staying about 80% European-owned. The man who built your playlist app now chairs Europe’s most valuable defense startup, and the submarine-hunting glider is one of the reasons the bids keep climbing.
The honest caveats, before anyone gets carried away: nobody outside Helsing and a few navies has publicly verified the ten-times detection claim, the 1,000-meter depth was still a work in progress as of last fall, and the UK is evaluating the Fathom against competing concepts rather than handing out contracts. Anduril is pitching a fixed-node cousin of the same idea called Seabed Sentry, so Helsing won’t have the ocean floor to itself. But the core bet doesn’t need every claim to land. If a swarm of 60-kg gliders can take over even the listening half of a frigate’s job, the budget math behind a $2.7 billion hull becomes a question somebody in a parliament will eventually have to answer. Norway’s new frigates will be superb ships. Helsing is betting they’ll spend their careers responding to contacts a glider heard first.
