We are making decisions about a seafloor we cannot see

The seabed is the least observed habitat on the planet, and one of the busiest. Offshore wind is being built out at record pace across Europe & Asia. Undersea cables now carry almost all of the world’s intercontinental data. Ports are dredging, coastlines are being reinforced, and a century of dumped munitions still sits in shallow coastal sediment on nearly every continental shelf.

All of that activity happens on top of living habitats. Seagrass meadows, kelp forests, cold-water corals, maerl beds and oyster reefs are not scenery. They are the nurseries where a large share of commercial fish species spend their early lives, the filters that keep coastal water clear, and the structures that hold sediment in place when storms hit. Cold-water corals grow a few millimetres a year, which means a colony damaged during a survey or an installation may have taken centuries to form and will not return in our lifetimes. And the seabed itself is one of the largest stores of organic carbon anywhere on Earth, most of it locked in sediment that stays put only if it is left undisturbed.

Benthic biodiversity, in other words, is not a nice-to-have layer on top of ocean health. It is most of what ocean health actually means below the surface.

And almost all of the infrastructure layered on top of this, is planned using sonar.

Sonar is efficient and it scales, but it returns shape, not identity. It can tell you that something is there. It cannot tell you whether it is a rock, a piece of debris, a gorgonian, or a protected coral. The alternatives are worse at scale: ROVs and divers deliver high resolution but cover tiny areas at high costs and operational complexities.

So the biodiversity layer, the part that matters most for ocean health, gets modelled, extrapolated from a handful of transects, or quietly skipped. Environmental impact assessments are legally required across most of these projects, but the evidence underneath them is often thin and based on missing or outdated science. Restoration projects face the same gap in reverse: it is hard to prove kelp or seagrass has recovered if you were never able to measure the baseline properly.

At Sustainable Ocean Alliance (SOA), we look for companies where the commercial product and the ocean outcome are the same artifact. Cosma is one of the clearest examples we have seen.

That is why SOA is proud to announce our investment in Cosma, a French ocean-data and subsea robotics company using swarms of autonomous underwater drones to produce centimetric, photoreal maps of the seafloor. Cosma has closed a €2.05M round, bringing total funding raised to more than €7M over the past 18 months. Every one of Cosma’s existing investors — Ifremer, Ternel, Wind Capital, Caisse d’Epargne Côte d’Azur and 50 Partners — reinvested above their pro-rata, and they are joined by three new investors: Region Sud Investissement, Ring Capital and Sustainable Ocean Alliance.

Related: Read Cosma's full press release here

Swarms that photograph the seabed, not just ping it

Cosma was founded in Nice in 2022 out of Ifremer, France’s national marine science institute. Rather than send one large, expensive vehicle over a site, the company flies a swarm of compact autonomous underwater vehicles roughly 1.5 metres above the seabed, carrying cameras alongside side-scan sonar and magnetometers. A cloud-native photogrammetry and AI pipeline then stitches that imagery into a photoreal, centimetric 3D model of the seafloor, delivered in 24 to 48 hours. Today the drones are designed for 300 metres and already operated from 2 to 200 metres and it will be possible to adapt the system to reach deeper waters.

The result is closer to a high-resolution aerial map of a forest than to a traditional survey chart. You can see individual species, count them, measure the area they cover, and come back a year later to see what changed.

Two consequences follow, and both matter for the ocean.

It is 10 to 100 times cheaper than conventional vessel-based survey, which turns high-resolution habitat mapping from an occasional luxury into something a municipality, a port authority, a marine protected area manager or an NGO can actually afford.

Every commercial mission returns a full biodiversity map as a by-product. A client paying for a cable inspection or a munitions survey also receives a species-level picture of the habitat they are operating in, with no additional mobilization, no additional vessel days, and no additional cost.

As Cosma’s founders put it: what has been seen cannot be unseen. Once a developer has photographic evidence of a cold-water coral colony inside a planned wind farm footprint, the conversation about where to site the turbines changes. This leads to smarter marine spatial planning and project permitting.
 

Deployment-ready today, and available to clients anywhere

This is the point we want to be unambiguous about, because it is unusual for a company at this stage: Cosma is not a prototype waiting on sea trials. The system is operational, it has been delivering paid missions for two years, and the team is available to mobilize for clients globally.

A large part of that is a design decision. Cosma owns no survey vessels. Their proprietary launch and recovery system is built to deploy from small local boats of opportunity, which means a 2 pax crew and a pallet of drones can mobilize almost anywhere there is a harbour, without waiting on the scarce and expensive specialised vessels that gate the traditional survey industry. That keeps costs down, keeps mobilization emissions down, and makes the technology reachable for coastal communities and clients far from the markets where this equipment is usually concentrated.

The work already spans three applications, each with real, paid deployments behind it:

• Biodiversity, restoration and environmental impact assessment. Kelp and posidonia monitoring with the Principality of Monaco and Seaforester. Habitat and macro-waste mapping with WWF, which demonstrated photogrammetry outperforming side-scan sonar and generated requests from neighbouring municipalities. Large-scale habitat mapping with the French Office for Biodiversity and SETEC. An offshore wind assessment with Pennavel covering more than 140 km, which mapped two protected coral species and put Cosma on the shortlist for roughly 25 years of monitoring.
• Inspection of subsea and coastal infrastructure. Seabed and infrastructure monitoring at the Port of Esbjerg in Denmark, including work in visibility conditions where conventional survey struggles, and shoreline engineering inspection with Egis, where the environmental assessment is required by law and the work is therefore recurring by default.
• Munitions clearance and seabed security. A feasibility study with RWE combining magnetometry and photography to find buried munitions where sonar fails, alongside work with the French Navy and DGA. Legacy munitions leach contaminants into the sediment around them, so finding and clearing them is an ocean health outcome as much as a security one, and here too the biodiversity map comes free with the survey.

Ten-plus deployments, ten signed contracts, and 100% client satisfaction across a customer list that already includes the French Navy and DGA, the Port of Esbjerg, RWE, WWF, the French Office for Biodiversity and the Principality of Monaco. Revenue today splits roughly evenly across those three applications, with a strong proposal book and pipeline behind it. Alongside the commercial work, the team runs pro-bono science campaigns, including in Greenland and Monaco, tracked as a distinct impact KPI each year.

Scaling that reaches further is one of the stated purposes of this round. Cosma has set out three priorities for the new capital: continuing to deploy in the most demanding environments, where visibility, temperature, depth, currents, obstacles and remoteness all work against you; building out dedicated capability for monitoring sensitive subsea infrastructure, including deployment from shore, from intervention vessels or carried by aerial drones, plus positioning in GNSS-denied conditions; and adding more advanced autonomous inspection, from sensor fusion and species recognition to time-lapse monitoring that tracks how a habitat changes over time.

Cosma’s CEO, Frédéric Mittaine, frames the problem the company exists to solve this way: benthic ecosystems and critical subsea infrastructure remain poorly observed, poorly understood and poorly monitored, even as we come to recognise how strategically important both will be over the coming decades.

Four Reasons SOA Invested

1. Impact that rides on every job, measured rather than modelled

Most ocean companies bolt an impact report onto a commercial product. Cosma cannot separate the two, because the biodiversity map is generated by the same flight that pays the bill.

In practice that means something specific. Every mission produces a georeferenced, photographic record of what actually lives on that patch of seabed: which species, how many, over how large an area, at a known date. That record is verified rather than modelled, which makes it usable as evidence. It can establish a baseline before a project breaks ground, show a regulator whether a protected habitat was avoided, prove whether a restoration site is recovering, or document a species assemblage nobody had catalogued. Because it costs the client nothing extra, it happens on jobs that would never have commissioned a biodiversity study at all: a cable inspection, a harbour survey, a munitions sweep. The company tracks this formally, weighting each project by the area surveyed, so impact scales with commercial throughput rather than competing with it.

That structural coupling is exactly what our diligence framework is built to find, and it is why species-level seabed evidence ends up in front of the people deciding what gets built, dredged or cleared.

2. A deployment-ready system with paying customers in every target market

Ocean buyers are deliberate, but when they adopt, they adopt at scale and stay for decades. Cosma has already cleared the hardest part of that curve: real assets, real conditions, real invoices, across biodiversity, offshore infrastructure and subsea security simultaneously. Several of those first contracts point toward recurring, multi-year monitoring rather than one-off surveys, which is what turns a survey business into a long-term record of how a habitat is changing.

3. A team we believe is the right one to build this

The caliber of this team was one of the strongest signals in our diligence. Solving this problem requires marine science, subsea robotics, data engineering and hard-won at-sea operational experience to sit in the same room, and that combination is genuinely rare. Cosma has it. Frédéric Mittaine leads the company after nearly two decades of seabed data focus; Yannick Pennecot runs robotics and operations and has experience in aerospace and aeronautics; Quentin Chenevier came from large-scale industrial data work to lead data and product; and Laura Huguenin, a marine biologist and professional diver, leads biology and delivery. They are all in the founding team, and it shows in how they work: careful with capital, focused on getting systems into the water rather than into a pitch deck, and grounded in the biology as much as in the engineering.

The moat here is not the robots. Anyone with capital can buy hardware. Piloting a swarm, launching and recovering it from a small boat in poor conditions, and turning what comes back into something a scientist or a regulator can actually use is the part few teams can replicate. This team has been doing exactly that at depth for two years, while holding costs flat. We came away from diligence convinced they are the team to build this.

4. A wave of demand pulling the market forward

In September 2022, the Nord Stream gas pipelines running under the Baltic Sea were ruptured by a deliberate act of sabotage. A cut to the Balticconnector pipeline followed a year later, and damage to subsea cables has been reported with growing frequency since. The effect has been to reclassify the seabed as critical infrastructure. Governments now need to know what is down there, in detail, and budgets have moved accordingly.

That security demand is currently the fastest door into the market, and it funds the biodiversity mission rather than competing with it. At the same time, the offshore build-out required to meet climate targets is accelerating worldwide, and every one of those projects needs an environmental impact assessment. Both curves point the same direction for Cosma.

Our vision: an ocean we can actually look at

SOA has long believed that what gets seen gets protected. Good evidence is what unlocks good decisions, and the seabed is finally becoming something we can see properly. A wind farm can be sited around a coral colony once someone has mapped it. A restoration project can attract serious funding once it can show, in photographs, that the seagrass came back. A coastal community can manage a protected area with a real picture of what lives inside it. Cosma is making that kind of evidence affordable and routine, at the level where decisions are actually made, one project, one permit, one seabed at a time.

The opportunity here is global. The seabed habitats that matter most for fisheries, carbon and coastal protection sit off every coastline, and the biggest gains are in places where high-resolution survey has simply never been within reach. Cosma's model, which mobilizes from local boats rather than dedicated survey ships, travels well to exactly those places. We are excited to support the team's global expansion, including as they look toward the United States.

We are proud to back Cosma and to welcome the team into the SOA Ecopreneur Network as they scale internationally, as part of our commitment to a better-understood, better-defended ocean.

If you are building or funding solutions at the intersection of robotics, ocean data and the blue economy, let’s connect. Send a message to Alicia Hernández Arriaga at alicia@soalliance.org.

By: Alicia Hernández, Investment Manager, and Matt Mulrennan, Head of Investments