Aquaculture Goes Autonomous

Aquaculture is becoming more important for a simple reason: demand for seafood keeps growing, while wild fish stocks cannot just be harvested more aggressively forever. Overfishing has already put pressure on many fisheries, so a larger share of future production will have to come from farming.

At the same time, aquaculture itself is getting harder. Warmer water and lower oxygen levels can hurt fish health, while more extreme weather makes farm operations less predictable. This increases the need for systems that can monitor what is happening and react faster.

This landscape looks at the companies building those systems. Some improve existing farms with smarter equipment, while others rethink the infrastructure itself. It is still a relatively small startup market, but the underlying need is clearly growing.

Growth & Feeding Optimization

What is this category about?

• This category covers technologies that help aquaculture farms make better decisions about how animals grow, with feeding as the most important use case. Fish and shrimp are hard to monitor because farmers cannot easily see what is happening underwater. Farmers often have to infer what is happening from periodic samples or visual observation. The companies in this category try to replace that uncertainty with continuous measurement and software that can influence daily production decisions.

• Feeding is the obvious problem because small mistakes become expensive very quickly at farm scale. Farmers need to give animals enough feed to maximize growth without continuing once appetite falls. Too little feed slows production, while too much feed means paying for pellets that are not converted into biomass. The difficulty is that appetite changes continuously, so a feeding plan based on fixed assumptions can become wrong within the same day.

• Farmers also need a much better view of the biomass they actually have in the water. Traditional weighing requires catching a sample and extrapolating from it, which gives farms an imperfect picture between measurements. Better biomass estimation makes feeding decisions more accurate and gives operators an earlier indication when growth diverges from expectations.

• The broader objective is to move aquaculture away from production decisions based heavily on human observation. A farmer can watch fish behaviour and adjust a feeder manually, but this becomes harder as farms become larger or more distributed. Continuous sensing allows the farm to react to what is happening in the water rather than follow a predetermined production plan.

What do products in this category do?

• Some products use smart feeders to decide automatically when and how much to feed the fish. Cameras or other underwater sensors track whether fish are still eating, and the system adjusts feeding in real time. ReelData does this with cameras and AI, while Tidal combines underwater sensing with automated feeding. CageEye uses hydroacoustics to measure how fish react during feeding and help farms adjust it.

• Another product type focuses on continuously measuring the animals rather than controlling the feeder itself. Better production decisions depend on knowing whether fish are growing as expected, but manual sampling only provides occasional snapshots. Aquabyte and OptoScale place camera systems inside farms and use computer vision to estimate biomass without handling the fish. Their software then turns those measurements into information that farm teams can use when adjusting production.

• Sonar becomes useful when cameras cannot reliably see what is happening underwater. Shrimp ponds are an obvious example because visibility can be poor and the animals are difficult to observe directly. Minnowtech uses sonar to estimate shrimp biomass so farmers can make better feeding and harvest decisions without physically sampling the pond. CageEye applies hydroacoustics in salmon cages, where the position and movement of the fish can reveal how they are responding to feed.

• A final product model is becoming the software layer that sits above the farm rather than solving one sensing problem. Once farms generate more production data, there is value in combining it into a system that helps operators plan what to do next. UMITRON has expanded from smart feeding into FARM, its production management software, while GoSmart connects its monitoring system to a cloud application that supports ongoing farm decisions. The product ambition here is closer to an operating system for production than a standalone monitoring device.

What types of customers are using these products?

• Large salmon producers are the clearest customers because even small improvements become valuable across many cages or sites. Mowi has worked with Tidal since its early days and is now extending the technology into its genetics operation. SalMar announced a multi-site collaboration with Tidal in 2026. ReelData followed a similar path with Samherji, where a trial eventually became a three year agreement covering all of the customer's Icelandic sites.

• Customer adoption often starts with a limited deployment before expanding across the operation. Kaldvík decided to install Aquabyte across eight farming sites, while Cermaq Canada tested ReelData at two facilities. This makes the GTM model relatively clear: prove that the system works in the customer's production environment first, then expand the number of sites once the economics are visible.

• Shrimp creates another important customer base, especially outside Northern Europe. Minnowtech describes Ecuador as its main market and has built local installation capacity there. Algaeba's first SeaThru CURRENT customer moved from validation into a subscription covering six shrimp ponds. The underlying need is similar to salmon farming, but the sensing technology and farm economics can be quite different.

What are the major trends shaping this category?

• Aquaculture itself is becoming a much larger part of global seafood production, which increases the value of making farms more productive. FAO estimates that aquaculture produced more than 100 million tonnes of aquatic animals in 2024, while capture fisheries have remained broadly flat for decades. Growth in seafood supply is therefore increasingly dependent on farming rather than catching more wild fish. That creates a structural market for technology that can improve output from existing farms.

• Feed efficiency is becoming a software problem rather than only a feed formulation problem. Mowi's 2025 results show how changes in feed costs and feed conversion directly affect production economics. At the same time, a 2026 review of precision feeding describes the industry moving away from fixed ration systems toward technology that senses demand and adjusts feeding automatically. The important shift is that farms can increasingly optimize how feed is used in real time rather than simply buy a better feed product.

• The technology is moving from monitoring toward closed loop control. Computer vision and other sensing technologies have existed in aquaculture for years, but newer systems connect observation directly to the feeder or another farm operation. A recent review of intelligent feeding describes real-time monitoring and system stability as central priorities for the next generation of feeding technology. This is important because the value of knowing that fish have stopped eating is much higher when the farm can immediately react to it.

• Fish welfare is becoming harder for large producers to treat as a secondary metric. Norway recorded sea phase salmon mortality of 15.4 percent in 2024, while the government has set a long term objective of moving mortality toward 5 percent. The European Commission is also pushing for clearer welfare standards and better use of monitoring data. This should increase demand for systems that can detect changes in fish condition early, even when their original product was built mainly around growth optimization.

How does the funding environment look?

• The category has already produced several companies capable of raising meaningful venture rounds. Aquabyte raised a $25 million Series B, while ReelData raised an $8 million Series A. UMITRON reached a S$15 million Pre-Series B and Ace Aquatec raised £10 million in 2025. Investors are therefore financing the category beyond the experimental stage when a company can demonstrate that its technology is being used in commercial production.

• The more interesting signal is that capital is starting to come from later stage investors rather than only early aquaculture specialists. Aquabyte became majority owned by Vitruvian Partners in 2025, while Insight Partners acquired a majority stake in OptoScale the same year. Both companies had spent years building monitoring platforms before those transactions. This suggests that mature aquaculture technology can become attractive as a real software and hardware business rather than remaining a niche climate or food-tech investment.

• The long tail remains much more dependent on smaller rounds or public support. Minnowtech has used NSF funding alongside private capital, while CageEye received Norwegian grant funding and Algaeba has relied partly on government-backed projects. That makes sense for technologies that need extensive field validation before farms will rely on them for production decisions.

• Overall, I would describe this category as clearly VC compatible, but still slow rather than trending. The ROI can be straightforward because better feeding directly affects farm economics, and successful products can expand across large customers after an initial deployment. The constraint is that these are not frictionless software products. They have to work reliably inside real farms, which makes validation and commercial expansion slower than in conventional SaaS.

Health & Water Management

What is this category about?

• This category covers technologies that help farms detect health or water problems earlier. Fish live directly in the environment the farm is trying to control. If water conditions deteriorate, the animals can become stressed or sick very quickly.

• Many problems are difficult to spot before they become serious. A farm may only discover that something is wrong once fish start behaving differently or mortality increases. Traditional testing can also rely on manual samples sent to a lab. The companies in this category try to give farms a more continuous view of what is happening.

• Water conditions can also change quickly. Knowing the current oxygen level is useful, but knowing that it is likely to fall tomorrow is much more useful. This is why some products are moving from simple monitoring toward forecasting and early warning.

• The same logic applies to animal health. Instead of treating a disease after it spreads, farms can monitor biological changes earlier or use preventive treatments. The goal is to intervene before the problem has a large impact on production.

What do products in this category do?

• Several products continuously monitor water conditions and warn farms when something is changing. BiOceanOr takes data from environmental sensors and uses AI to forecast problems such as low oxygen. AQUAWISE is trying to do something similar without installing sensors on the farm. It uses satellite data and AI to estimate water conditions remotely.

• Another group looks directly at the biology inside the water. The idea is to detect changes that normal water sensors cannot see. FP AQUA uses an underwater microscope to continuously observe microscopic organisms. Hadl uses AI to analyse microscope images from shrimp farms. KYTOS focuses on the microbiome and helps farms understand whether the microbial balance of their system is changing.

• Sensor Globe focuses on what fish experience during farm operations. Its sensors move through handling systems with the fish and measure the conditions they are exposed to. This helps farms identify parts of a transfer or treatment process that may be causing stress or injury.

• AquaNab takes a very different approach because it is developing a treatment rather than a monitoring product. Its first product uses nanoantibodies mixed into salmon feed to protect fish against sea lice. The idea is to make treatment easier by giving it through normal feeding instead of handling the fish separately.

What types of customers are using these products?

• Large fish farmers are the most obvious customers because health problems become very expensive at scale. BiOceanOr says Chilean salmon producer INVERMAR expanded its technology from a few sites to all of its active sites. In France, Aquafrais Cannes is using BiOceanOr to better anticipate oxygen conditions. Sensor Globe names salmon farmer Måsøval among its customer references.

• Land based farms also have a strong need for these products because the farm controls much more of the water environment itself. Oceanloop has used KYTOS at two shrimp farming sites to monitor the microbiome inside its recirculating systems. KYTOS also has a commercial collaboration with RASLab.

• Some companies reach farms through larger aquaculture technology providers rather than selling everything directly. Innovasea integrated BiOceanOr's oxygen forecasting into its own monitoring software, while ScaleAQ has a cooperation agreement with BiOceanOr. This could become an important GTM model because farms already use these larger platforms for other parts of their operations.

What are the major trends shaping this category?

• Disease remains a very large economic problem for aquaculture. FAO's General Fisheries Commission for the Mediterranean estimates that aquatic animal diseases cost the industry more than $6 billion every year. Even in Norway, where fish health is closely monitored, 14.2% of farmed salmon died during the sea phase in 2025. That was an improvement from 2024, but it still means more than 54 million salmon died.

• Climate change makes health management harder because farms depend on environmental conditions they cannot fully control. FAO identifies temperature changes and poor water quality as causes of mortality, while harmful algal blooms can also create sudden problems. Norway's 2025 Fish Health Report added a dedicated chapter on climate change because its impact on fish health is becoming more important.

• The industry is being pushed toward prevention rather than simply treating problems afterwards. FAO links the overuse of antimicrobials in aquaculture to antimicrobial resistance and argues that better disease prevention can reduce reliance on these treatments. At the same time, the European Commission is calling for stronger fish welfare monitoring and better use of farm data. Both trends increase the value of products that identify problems before animals become seriously sick.

How does the funding environment look?

• BiOceanOr is the clearest example of a company reaching conventional venture scale in this category. It raised €2.5 million in 2023 and another €2 million Series A in 2025. Its development also shows what investors may find attractive here: the product is software based and can be deployed across many farms without building a completely new system for every customer.

• Most of the other companies are still much earlier. Sensor Globe raised a $600,000 pre-seed round in 2025, while AquaNab recently raised an undisclosed seed investment from Nordic Foodtech VC. Hadl also has Hatch Blue backing, but there is little evidence of larger institutional rounds across the rest of the category.

• Overall, I would describe Health & Water Management as selectively VC compatible and currently slow. The problem is real and customers have a clear reason to pay when a product can reduce mortality. But most companies are still small, and there is very little competition between heavily funded startups. This is an active technology category, not a VC trend yet.

Farm Task Automation

What is this category about?

• This category covers technologies that automate specific physical tasks on fish farms. Many farm operations still require people to inspect fish or maintain equipment manually. These companies build machines that can do part of that work automatically.

• Some of these tasks are difficult because they happen underwater. Cage nets need to be cleaned and checked regularly, but doing this manually takes time and can expose workers to difficult conditions. Robots can stay in the water and perform the same task much more often.

• Other tasks require handling large numbers of fish individually. Hatcheries need to inspect or sort fish, but doing this manually becomes expensive at scale and can stress the animals. Computer vision can identify relevant characteristics quickly and let machines sort the fish automatically.

What do products in this category do?

• Some products are underwater robots that take over cage maintenance. Remora Robotics makes robots that stay on fish cages and clean the nets automatically. The same robot can inspect the cage and detect damage. Remora has also added software that lets farms see the condition of their nets remotely.

• Another product type automates fish inspection and sorting in hatcheries. A camera looks at each fish, AI decides how it should be classified and the machine sends it to the right group. Aquaticode uses this model to sort salmon based on traits such as sex or maturation. OctaPulse uses computer vision to inspect fish for physical characteristics and is adding robotic sorting to the same workflow.

• BioSort is trying to automate intervention at the level of individual fish inside a cage. Its technology identifies fish as they pass through the system and can use that information to decide which fish need attention. The company developed this approach through the iFarm project and is now focusing on a simpler product for earlier sea lice intervention.

What types of customers are using these products?

• Large salmon producers are already using these systems in commercial operations. Remora signed its first commercial contract with MOWI and later deployed robots with Bjørøya. Its first international deployment was at a salmon farm in Chile.

• Fish sorting products are also moving from trials into larger deployments. Aquaticode says AquaChile and Australis expanded their use of its system, with agreements covering more than 60 million fish. In 2026, it signed another agreement with İlknak Aquaculture in Turkey covering sea bass and sea bream hatcheries.

• Smaller startups can win meaningful contracts early because the task they automate is very specific. OctaPulse has a six figure annual contract with Riverence and has been deploying its system at additional farms. This suggests farms are willing to buy automation before the supplier itself has reached much scale.

What are the major trends shaping this category?

• More aquaculture work is moving toward robots instead of people doing underwater tasks. Net inspection and cleaning are obvious examples because they are repetitive and difficult to perform manually. Recent research describes autonomous net operations as a clear direction for sea based fish farming, although underwater navigation and biofouling still make full autonomy difficult.

• The move toward larger and more exposed farms should make automation even more useful. Offshore and exposed sites are harder for people and service vessels to access regularly. Research on exposed aquaculture points to remote operations and autonomous systems as important parts of making these farms safe enough to operate.

How does the funding environment look?

• The largest rounds are going to companies that already have hardware working on farms. Remora raised NOK 164 million in 2025 after reaching commercial deployments and serial production. BioSort followed with more than NOK 100 million in 2026 as it moved from the long iFarm development programme toward commercialization.

• Aquaticode shows that investors are also willing to fund automation earlier in the product cycle. It raised a $6 million Series A in 2022 before its current wave of larger customer deployments. The company has since moved from development with Salmones Camanchaca into contracts covering tens of millions of fish.

• There is also new startup activity entering the category. OctaPulse is only a 2025 company but joined Y Combinator's W26 batch and received YC's $500,000 standard investment. It already has paid deployments, which makes it a useful example of how quickly a narrow automation product can reach customers.

• I would describe Farm Task Automation as VC compatible and starting to trend, but it is far from hype. The strongest products replace a task that already costs farms money, so the value is easy to understand. The harder part is building reliable hardware and proving it inside real farms. That slows down scaling, but the recent funding and commercial activity suggest this is becoming a real venture category.

Aquaculture Infrastructure

What is this category about?

• This category covers the physical systems that make aquaculture farms work. Some farms need much more than cages and feeders. They need systems that control the water or supply oxygen continuously.

• This is especially important for land based aquaculture. When fish or shrimp are raised in tanks, the farm has to recreate the conditions that animals would normally get from the surrounding environment. Water has to be treated and reused, while oxygen has to remain at safe levels.

• The main problem is making these farms reliable enough to operate at commercial scale. A failure in oxygen supply or water treatment can affect the whole farm very quickly. Infrastructure therefore has to work continuously and often needs backup capacity.

What do products in this category do?

• Some companies build complete land based farms. Oceanloop develops modular recirculating aquaculture systems for shrimp and marine fish. Vertical Oceans has designed a vertical prawn farm where tanks are stacked and controlled automatically. Atarraya sells Shrimpbox, a modular system that can be installed in locations closer to the final customer.

• Other companies focus on oxygen, which is one of the basic requirements of intensive aquaculture. Poseidon Ocean Systems sells oxygen generation and diffusion systems for salmon farms. Nernst Electric is developing equipment that produces oxygen directly on site, so farms do not have to rely as much on delivered oxygen.

• There are also technologies that improve how oxygen is delivered into the water. nanobOx creates very small bubbles that remain in the water longer and can improve oxygen availability. The company is testing this in aquaculture environments, including a shrimp trial with Three Sixty Aquaculture.

What types of customers are using these products?

• Large salmon producers buy infrastructure that can be added to existing farms. Poseidon has a five year agreement with Cermaq Chile covering four sites and has also deployed technology with Mowi Canada West. These customers already operate large farms and buy equipment to improve how those farms work.

• Land based aquaculture companies are another customer group, but the business model can look different. nanobOx is currently testing its technology with Three Sixty Aquaculture. Nernst Electric is also targeting fish farms as the first market for its oxygen generation system, although it is still preparing commercial trials.

• Some infrastructure companies operate the farms themselves instead of selling the system to another farmer. Vertical Oceans sells the prawns produced by its own technology. Atarraya has followed a similar model in the US, where it built Shrimpbox farms and secured buyers for the shrimp. Oceanloop also uses its own farming operations to bring seafood to market. This makes these companies partly technology providers and partly aquaculture operators.

What are the major trends shaping this category?

• More aquaculture production is moving into controlled systems. RAS lets farms reuse most of their water and gives operators much more control over growing conditions. Recent research describes adoption as increasing, especially for valuable species and in places where water availability is limited.

• Energy remains one of the biggest problems for land based aquaculture. Water has to keep moving through the system and oxygen must be supplied continuously. A 2026 review describes RAS as water efficient but energy intensive, while another commercial scale study found that water circulation and oxygenation were among the largest sources of energy use. This creates demand for infrastructure that can reduce the energy needed to keep the farm running.

• Climate change is making water control more valuable for traditional farms as well. Norway's Institute of Marine Research says warmer water and lower oxygen levels are becoming bigger risks for aquaculture. That makes oxygen systems and better water management useful beyond fully land based farms.

• The scale of new land based projects is becoming very large. The EIB is currently supporting projects in Iceland where individual facilities require hundreds of millions of euros of investment. This shows both the opportunity and the problem with the category: infrastructure can support very large farms, but building those farms requires a lot of capital.

How does the funding environment look?

• This category attracts larger rounds than most of the aquaculture landscape because the products are expensive to build. Poseidon raised a $20.75 million Series B in 2024. Oceanloop secured €35 million from the European Investment Bank the same year and added up to €6.5 million of new equity in 2026.

• There is also funding for younger infrastructure technologies. Nernst Electric raised $2 million in 2026 to move its oxygen technology into pilots. nanobOx raised around €900,000 for its first institutional round. These companies can start with individual pieces of infrastructure before trying to scale into larger deployments.

• I would describe this category as selectively VC compatible and active, but not trending. There is clear investor interest and several companies have raised meaningful rounds. But the capital requirements are high and scaling depends on physical deployments. The strongest companies may therefore need a mix of venture capital and infrastructure financing rather than following a normal VC path.