Stingray’s Laser Robots: How Norway Shoots Sea Lice Off Salmon With AI

Off the Norwegian coast, in the same open-net pens that produce most of the world’s farmed Atlantic salmon, thousands of underwater machines now spend every hour hunting a parasite the size of a fingernail. Each unit is a submersible "node": stereo cameras, lights, software and a laser, hung in a sea cage so that as salmon swim past, the system can spot a sea louse and fire a pulse at it, without touching the fish.

The Oslo company behind the machines, Stingray Marine Solutions, said that as of late 2025 it had more than 2,500 active nodes along the Norwegian coast, live-streaming from about 15,000 cameras, and that the system was monitoring and protecting 20 to 25 percent of farmed salmon in Norway. On December 31, 2025, it reported that it was protecting 112,218,373 salmon and trout across 38 customers (Stingray Fish Health Report 2025). Those figures are higher than a widely shared claim that Norway had "more than 2,000" laser robots already guarding "more than 50 million" fish. The 2,000-node mark was real, but earlier: Stingray delivered its 2,000th node in February 2025 and its 3,000th in November 2025. A mid-2025 company article already put the number of fish above 100 million (Stingray).

The machines do not make sea lice disappear. They are one attempt to deal with a parasite that has become the central constraint on Norwegian salmon farming: costly for companies, harsh on farmed fish when they have to be treated by other means, and a documented threat to wild salmon and sea trout. Here is how the system works, why the lice are such a problem, and what the evidence actually shows.

Why sea lice are such a problem

The salmon louse, Lepeophtheirus salmonis, is a naturally occurring marine copepod. It latches onto salmon, trout and Arctic char and feeds on mucus, skin and blood. A heavy infestation can open wounds, throw off the fish’s salt balance, slow growth and, in the worst cases, kill the host. Young wild salmon heading to sea are especially vulnerable: Norway’s Food Safety Authority notes that more than 10 lice on a smolt can be fatal (Mattilsynet).

Lice have always been part of life in the North Atlantic. What changed is the number of hosts. Farmed salmon and rainbow trout live in open cages, so lice and their larvae move freely between the pens and the surrounding water. The Institute of Marine Research (Havforskningsinstituttet) estimates that aquaculture is the source of more than 97 percent of salmon lice in Norwegian waters. That extra load spills onto wild fish. In its 2026 risk report, the institute again listed poor welfare on the farms and the effect of lice on wild salmonids as the main obstacles to a sustainable industry. It judged the risk of lice-induced mortality in migrating wild salmon smolts as high in two of Norway’s 13 production areas, moderate in eight and low in three. For sea trout and Arctic char, which stay longer in coastal water, the risk was high in nine areas (Institute of Marine Research).

Norwegian farms are legally required to keep lice down. For most of the year, the average number of adult female lice must stay below 0.5 per fish; a tighter cap applies in spring, when wild smolts migrate to the sea. If a farm cannot get under the limit, it may have to slaughter the fish. Counts, treatments and sea temperature are reported weekly when the water is 4°C or warmer (Mattilsynet). Those rules also feed the country’s "traffic light" system, which can allow, freeze or cut a region’s production capacity according to modeled lice-driven mortality in wild salmon.

The economic stakes are large. Norway exported 1.41 million tonnes of farmed salmon in 2025, worth 124 billion kroner, according to the Norwegian Seafood Council figures cited in the same marine-research report. A Nofima analysis cited by Benchmark Animal Health has put the annual cost of sea lice to the Norwegian industry at about 6.6 billion kroner. A 2023 bio-economic study in Preventive Veterinary Medicine found that a farmer could spend, on average, about 5.4 million kroner (then roughly €535,000) per cage in a production cycle to prevent four thermal delousing treatments and still break even (Walde et al., 2023).

The old toolkit, and why it hurts the fish

For years the industry’s main answer was medicine: drugs in the feed, or pesticide baths in a wrapped cage or a wellboat. Repeated use selected for lice that could survive those drugs. By around 2015, resistance to most available medicines had pushed Norwegian farmers toward non-medicinal methods (Jensen et al., 2020).

The methods that filled the gap knock lice off fish, and are hard on the fish themselves. In a thermal treatment, salmon are typically bathed in water heated to 28°C or 34°C for about 30 seconds. In a mechanical treatment, they are pumped and flushed or brushed with seawater. Freshwater baths are also used; they still require handling, but a laboratory comparison in Aquaculture International found freshwater performed best, while a 34°C bath reduced lice but left more injuries and poorer growth (Thompson et al., 2023).

Handling is now central to the welfare problem. In 2025, figures from the Directorate of Fisheries showed that 67 million farmed salmon were lost: more than 54 million died in the cages, about 2.5 million were discarded at slaughterhouses, 26,000 escaped, and nearly 10 million were recorded as lost for "other causes." That was three million fewer deaths than in 2024, but the Institute of Marine Research still judged the risk of poor welfare as high in most production areas. Non-medicinal delousing was the most common type of welfare incident reported to Mattilsynet in 2025, at 47 percent of notices, and it had the highest associated mortality (Institute of Marine Research). Mattilsynet is blunt: treatments against salmon lice cause the biggest animal-welfare and health problems in the industry.

Cleaner fish — lumpfish and wrasse put into the pens to eat lice — were supposed to be a gentler form of biological control. Their own welfare record is poor, and their use is falling. That approach is a long way from viruses that eat bacteria, but it shares the same hope: let a living organism do the dirty work. Stingray, citing sector figures, says national stocking of cleaner fish fell to 19.1 million in 2025, close to 70 percent below the level of six years earlier, and that cumulative mortality among cleaner fish in Norway is estimated at 82.1 percent. The company wants the practice phased out no later than 2029, in line with a recommendation from the Norwegian Veterinary Institute (Stingray Sustainability Report 2025).

How the camera, AI and laser work

The idea is older than the current fleet. Esben Beck patented optical delousing in 2010. Stingray was founded in Oslo in 2012 and launched commercially in 2014, when it produced 12 nodes (Stingray; Sustainability Report 2024). Danish investment firm Novo Holdings bought a majority stake in October 2024 (Fish Farming Expert).

A Stingray node hangs in the pen and can be steered remotely. The company says it developed its own thruster so the submerged unit can rotate a full 360 degrees and hold a heading. Several cameras photograph fish as they swim by, with LED lamps, including royal-blue light at 450 nanometers, meant to keep the pictures usable day and night and to heighten the contrast between a louse and the fish. The lights and cameras are specified to work from the surface down to 25 meters; a 2025 sustainability report says the laser unit itself can descend to about 20 meters and is moved around the pen to stay with the fish (Stingray Technology; Sustainability Report 2025).

Machine-learning detectors, trained on years of labeled images of fish and lice, try to pick out individual lice in real time. When the software is confident it has found one, a laser pulse is fired at that spot on the skin. Stingray says the pulse is meant to kill or knock off the parasite while the salmon swims on unharmed, and that the lasers can shoot up to five pulses per second, around the clock. The system is aimed at mobile and adult lice, the stages that reproduce, rather than the smallest attached larvae. The software has to decide in a fraction of a second, the same class of problem as using AI to steer fusion plasma. Farmers can watch live video and historical counts in a portal called Stingray Online. Dedicated "laser operators," a job the company says it helped create, adjust the nodes so enough fish pass through the camera’s field of view; by 2025, Stingray said more than 100 full-time-equivalent operators were employed outside the company.

They are a narrower kind of industrial robot than the machines trained to work around molten steel, but the job is similar: see a target and act without a person in the loop. The hardware is built in Oslo. Customers have included large producers such as Mowi and Cermaq. In May 2026, Stingray said it had finished a second factory next to its existing plant in Stålfjæra, almost twice the size of the first 2,200-square-meter building, with production scheduled to start in August 2026. CEO John Arne Breivik then put the workforce at 275 (Fish Farming Expert). The company has also expanded in Iceland, where it opened a service center in March 2025 and reported 56 nodes deployed that year.

What the evidence actually shows

Does it work? The honest answer is mixed, and it depends on which study you read and how much the technology has changed.

In 2020, researchers at the Institute of Marine Research, the University of Bergen and the University of Melbourne ran a 54-day commercial-scale trial at a Bremnes Seashore site. Laser nodes were installed in three of six sea cages that already used snorkel cages, a barrier meant to keep salmon below the lice-rich surface layer. At the end there was no difference in mobile lice between cages with and without lasers, and adult female counts in all cages were close to the legal trigger of 0.5 per fish. The lasers fired a large number of pulses relative to the lice present, which the authors took as a sign that lethality, not a failure to aim, was the limiting factor. If every pulse had worked, they estimated the devices should have removed 4 to 38 percent of mobile lice each day. They found no effect on welfare indicators such as skin condition or eye status. The paper’s title was blunt: "Salmon lice survive the straight shooter" (Bui et al., 2020).

That trial is now several years and several product generations old, and it tested lasers alongside snorkel cages at a single site. A much larger, but different, kind of evidence appeared in 2025. In a peer-reviewed paper in Aquaculture, Stingray researchers working with statistician Jan Bulla of the University of Bergen analyzed official 2023 lice and treatment data for the entire Norwegian coast, then flagged which site-weeks used optical delousing. After accounting for sea temperature, sites with the lasers had about a 50 percent lower weekly probability of needing some other lice treatment: 5.06 percent versus 10.1 percent at a reference temperature of 10°C. Laser sites accounted for 10.9 percent of production weeks but only 5.9 percent of treatment weeks, with no sign that lice counts were higher there, even though those sites also used fewer cleaner fish (Worm et al., 2025).

Study	What was tested	What it found
Bui et al., 2020	54-day commercial trial, lasers in 3 of 6 cages	No difference in mobile lice; no welfare harm detected
Worm et al., 2025	All Norwegian sites, 2023 public data	About 50% lower weekly chance of other treatments at laser sites
Stingray 2025 operations	Company analysis of 2025 site-weeks	Laser sites: 23.2% of production weeks, 13.1% of treatment weeks

Two caveats belong next to that national result. First, several of the authors work for Stingray; the analysis used public regulatory data and a University of Bergen statistician, but it is not an independent experiment like the 2020 trial. Second, association is not the same as a controlled demonstration that each laser pulse kills a louse. What the 2023 numbers show is that sites using the system needed fewer reactive treatments.

Stingray’s own 2025 operating data, published in 2026, point in the same direction in a warmer year: modeled weekly treatment probability was 8 percent at laser sites versus 12 percent at others, a 30 percent reduction despite higher sea temperatures (Sustainability Report 2025). CEO Breivik has also said the company has more than 3,000 "node-years" of continuous operation without documented injury to fish, a claim that has not been independently audited in a public study. The lasers are sold as a continuous, preventive control that can cut how often farmers have to pump fish into a heated or mechanical treatment, not as a guarantee that those treatments disappear.

What comes next

The lice problem is not receding. Marine heatwaves in 2024 in northern Norway and in 2025 in the west sped up lice reproduction and forced more delousing. Production is still rising: farmed Atlantic salmon output jumped by about 182,000 tonnes in 2025, the largest annual increase in decades, and only about 40 of roughly 900 aquaculture sites were using submerged cages, another way to keep fish out of the lice-rich surface layer (Institute of Marine Research). The government has been looking at lice-emission quotas and at letting farms use otherwise cut capacity if they switch to closed units.

Stingray is betting that continuous optical control, plus the welfare and lice-count data its cameras already collect, will be part of that shift. Automated lice counting on its system has been approved by regulators, according to the company, which would reduce the need to hoist fish out for manual counts. A second Oslo factory is meant to raise output in the second half of 2026. Iceland is a second market, in part because wellboat capacity there is tight and handling in cold water is riskier. The company also wants to help cut unwanted mortality in the sea phase to 5 percent by 2030, the target set in Norway’s animal-welfare white paper.

None of that makes a laser a miracle. The 2020 trial is a reminder that a clever sensor can still leave lice on the fish. The 2023 national data are a reminder that, at scale, the same idea may still earn its keep by keeping farmers out of the treatment wellboat. For an industry that lost tens of millions of salmon last year, and that is legally on the hook for the lice it sheds onto wild fish, that narrower claim is the one worth watching.

Further reading