Protecting undersea infrastructure in the Baltic: the market for seabed surveillance, uncrewed systems and sensors, 2024-2026

The Baltic seabed-surveillance market did not exist as a labelled category before 2023. It was built, announcement by announcement, in response to a specific run of incidents: Balticconnector's gas pipeline and a telecom cable cut in October 2023, then C-Lion1, Estlink 2 and other cables damaged around November-December 2024. NATO launched Baltic Sentry on 14 January 2025 at a Helsinki summit, adding frigates, patrol aircraft and 'a small fleet of naval drones' on top of two existing Standing NATO Maritime Groups - not a newly created uncrewed force, but an uncrewed layer bolted onto ships already there and run through the same command structure.

Alongside Baltic Sentry, NATO's Allied Command Transformation (ACT) launched Task Force X-Baltic (TFX-Baltic) the same month to test whether uncrewed maritime systems could actually do useful work in this environment. Its most-cited number is that a three-week live demonstration in June 2025 fielded 'more than 70' air, surface and subsurface systems - drawn from over 40 contracted assets spanning at least 13 named USV/UUV/UAV types - running at roughly 75% availability for eight hours a day and tracking hundreds of vessels daily, including suspected 'shadow fleet' shipping. That is a genuine field-test statistic, not a permanent standing fleet.

NATO also claims its uncrewed fleet ran at 'one-third the cost' of equivalent crewed-frigate coverage. That figure comes from NATO itself, with no published cost model behind it - worth citing, not worth treating as an audited number.

Sensors are useless without a place to fuse their data. NATO built Mainsail, a cloud-based analytics tool developed by the Centre for Maritime Research and Experimentation (CMRE) and delivered to Allied Maritime Command around January 2025, to pull satellite imagery, sonar and underwater-sensor feeds into one real-time picture and flag 'patterns of life' that look like sabotage prep. During TFX-Baltic, vehicle data reached Mainsail via NATO's CWIX interoperability exercise using STANAG 4817, a maritime-unmanned-systems data standard that as of October 2025 was still awaiting formal ratification.

• MANGROVE consortium (Saab-led, 10 organisations, 12 sponsoring nations) - selected 16 July 2025, started 1 September 2025, building the Allied Underwater Battlespace Mission Network
• NATO DIANA - ran a 2026 challenge programme with a dedicated 'Maritime Operations' track, though this research could not verify winner-level detail for that track through a directly opened source
• CMRE's research vessel Alliance - led TFX-Baltic's underwater workstream directly, rather than only coordinating from shore

On 12 February 2026, eight Allies - Denmark, Estonia, Finland, Germany, Latvia, Lithuania, Poland and Sweden - agreed to fast-track acquisition of uncrewed naval capabilities, building on TFX-Baltic's 2025 testing (this report calls this step 'Phase II' for clarity; NATO's own release does not use that label). NATO's text describes acquisition 'through the development, purchase or leasing of capabilities' rather than explicitly promising national ownership - but it is still the point at which this market stops being solely a NATO-funded trial and starts becoming, at minimum, eight separate national procurement conversations. Poland's own rotation into this structure runs on a longer clock still: one analyst notes Poland is due to take command of Task Force Baltic only in 2028.

On 11 February 2026 the European Commission adopted a Cable Security Toolbox and set aside close to €350 million for submarine-cable projects: €60m (2026) for repair equipment, €20m for damage-detection and monitoring tools, €267m across 2026-2027 for 13 named 'Cable Projects of European Interest,' and a further €20m to pre-position repair kit at ports and shipyards - whose first pilot targets the Baltic Sea specifically.

Separately, the Commission's OceanEye initiative (presented 3 June 2026) commits more than €92 million - split roughly €50m to global ocean monitoring, €12m to data-sharing systems and €30m to startups building underwater drones, AI sensors and other ocean tech - aiming for the EU to supply over a third of the world's ocean data and 35% of the global ocean-tech market by 2035.

The 2025 European Defence Fund round put roughly €1.07 billion behind 57 defence projects EU-wide; of those, 'SHIELD,' 'SWORD,' 'DEEP-TECH,' 'ABYSSA' and 'SOUND2' are the ones specifically about underwater warfare or subsea sensing - a small slice of a much larger fund, not a dedicated seabed-security budget line.

Saab is the most concrete single vendor in this market for one buyer: it holds two separate FMV contracts, one for the mine-countermeasure AUV62-MR (signed December 2024; contract value undisclosed) and one worth SEK 60 million to design, build and test a new 'Large Uncrewed Undersea Vehicle' (LUUV) as an unarmed seabed-monitoring sensor platform (signed August 2025, sea trials due summer 2026). Both use Saab's shared 'Autonomous Ocean Core' control software rather than one-off code.

HII's REMUS family is the volume play: more than 700 units sold globally as of March 2025, marketed across ISR, mine countermeasures, anti-submarine warfare and electronic warfare. Exail's approach is narrower and more concrete in the Baltic specifically: its DriX H-8 uncrewed surface vessel went to Poland's Orlen Petrobaltic in October 2025 - the first commercial DriX deployment in the country - fitted with a specific echosounder for a specific hydrographic job, not a general-purpose sensor platform.

Anduril's Seabed Sentry and ThayerMahan's SeaPicket both sell the same basic idea - a recoverable, rechargeable, buoy- or seabed-mounted acoustic sensor - to slightly different buyers. Seabed Sentry leans on Anduril's Lattice software and its own Dive family of AUVs; SeaPicket is a simpler moored hydrophone array that shipped its first unit to an unnamed international customer in April 2026 under what ThayerMahan calls a 'rapid procurement' framework.

Kongsberg's August 2026 Aegir SSA launch illustrates a pattern worth watching: rather than building a new platform, Kongsberg paired its sonar with Fugro's existing SEAWATCH ocean buoy, demonstrated live from the Oslofjord Testbed, and referenced two separate 2026 contracts (Belgium's Princess Elisabeth Island, and one unnamed customer) as proof of demand. Sensor vendors are increasingly renting existing ocean-monitoring infrastructure rather than deploying from scratch.

The same logic shows up in partnerships rather than products: Helsing, Blue Ocean Marine Tech Systems, Ocean Infinity and QinetiQ announced a joint push against underwater threats in April 2025, and Helsing then bought Blue Ocean outright in October 2025 to control AUV production directly. Saab's MANGROVE consortium for NATO's AUWB-MN project runs twelve companies under one prime. No vendor in this research offers a single-company, end-to-end sense-fuse-respond stack; buyers are integrating pieces from several suppliers.

Poland's Armaments Agency ran a market consultation for an uncrewed-surface-vessel programme between 19 March and 10 April 2026, drawing nine responses spanning Kongsberg, Fincantieri, Poland's own WB Group and several smaller regional integrators. The tender referenced a 'Magura V7-class' configuration - a term borrowed from a Ukrainian missile-armed strike USV design - which trade press flagged as potentially misleading, since Poland's own brief describes protecting other USVs rather than confirmed strike armament.

Finland's Defence Forces opened a market-participation call in spring 2026 for 'Raivaamiskyky 2030,' replacing ageing Kuha- and Kiiski-class minesweepers with vessels usable manned or unmanned, targeting operational capability in the early 2030s - a decade-long horizon, not a near-term buy. Sweden, by contrast, already has money moving: FMV's Gavia AUV follow-on order (March 2026) sits alongside its two live Saab contracts, making Sweden the most concretely committed Baltic buyer in this research as of September 2026.

Finland's Elisa completed testing of distributed acoustic sensing (DAS) on its own Baltic cables, working with the Finnish Navy and Border Guard, reported around June 2026; the pilot's full partner list - Border Guard, Navy, Naval Academy, Fingrid, Gasgrid, the Geological Survey of Finland and University of Helsinki seismologists - shows how far outside telecoms a cable-monitoring pilot now reaches. The system is being turned into an automated alert service; the tests used Border Guard and Navy equipment, though the source does not name a specific vessel.

The motivation is concrete: Elisa says a December 2025 anchor-drag incident, suspected to involve a vessel sailing from Russia, damaged one of its cables (no material disruption, thanks to redundancy), and separate reporting puts the repair cost of at least one 2024 Baltic cable incident at €60 million or more - though the source of that estimate is unclear.

The most ambitious DAS bet is structural rather than retrofitted: Amber Cable, a Latvian special-purpose company incorporated February 2025, is designing a roughly 1,500 km cable linking Finland, Estonia, Latvia, Lithuania, Poland and Denmark through nine landing stations, with DAS built in from day one across the whole route. As of July 2026 the project was still in Phase 1 of 4 - feasibility and stakeholder engagement - with no cable laid and vendors not yet chosen. Elisa's retrofit pilot and Amber Cable's from-day-one design use the same dark-fibre backscatter physics, and between them show DAS moving from single-operator pilot to default design requirement within about a year and a half.

Every documented case in this research where a sensor flagged suspicious seabed activity - Mainsail's analytics, Elisa's DAS pilot, TFX-Baltic's vessel-tracking - was a monitoring or forensic success, not a prevention. No source found here shows a system stopping a cable-damage incident before contact. Endurance is part of why: small battery AUVs typically need daily recharging, Exail's larger DriX H-8 is rated up to roughly 10 days, and Saildrone markets multi-month persistence - which is consistent with why 2024-2026 deployments favour fixed buoys or surface USVs for continuous coverage, keeping AUVs for shorter mine-countermeasure and survey jobs.

• Treat every vendor cost-per-mile or cost-per-unit figure as a marketing claim until you can see the contract scope behind it.
• A 'contract' with a shipbuilder or defence prime is usually a design-and-build programme, not a fleet order - ask how many units it actually buys.
• Standards (STANAG 4817) and consortia (MANGROVE) are still forming; buying now means buying ahead of the standard, not to it.
• Analysts (CEPA, Brookings) argue procurement fragmentation and slow attribution are bigger blockers than sensor technology itself.
• NATO, Kongsberg and the EU all lean on the same pitch - AI automation lets a handful of humans supervise many sensors at once - so ask what happens when that AI is wrong, not just how many vehicles it can watch.

This topic sits inside a wider research program covering the whole Baltic; the legal and insurance detail behind EU cable-security rules (CER, NIS2, national transposition) is deliberately out of scope here and covered separately - this report's EU material is limited to the money and the hardware.

• What each of the eight Baltic Allies actually committed to spend under the February 2026 national-capability agreement - not disclosed at the time of this research.
• Whether NATO DIANA's 2026 maritime-operations track has named winners working specifically on seabed/undersea problems - the primary DIANA cohort page could not be opened directly in this run.
• What Amber Cable's DAS system will actually cost per kilometre once it leaves Phase 1 - no vendor or price has been chosen yet.
• Whether any Baltic state will sign a fielded (not pilot) seabed-sensor-network contract before the end of 2026.