The aquaculture industry has invested heavily in feeding technology over several decades. Sensors. Cameras. Feeding systems. Digital platforms. More and more data. So why is feeding still the industry’s largest unrealised margin opportunity?

The answer is uncomfortable. Not because the technology hasn't worked, but because we're asking it to do the wrong job. We've been collecting more information when what we actually need is to make better decisions with what we already have — across thousands of pens, every hour, every day.

Good feeding is an art. Scalable feeding is a discipline. The difference between the two is the difference between marginal improvement and a real leap in performance.

AKVA group
Modern feeding operations involve hundreds of parallel decisions every single hour. Photo. AKVA group

When feeding reaches the capacity ceiling

Picture a feeder along the coast right now. She has been doing this for fifteen years and is among the best in the company. She probably drinks a little too much coffee – and understandably so. She is monitoring twelve pens in parallel, each with its own cameras, fish, appetite patterns, and biological history.

Every hour she makes hundreds of decisions. She interprets behaviour in pen four, checks feed response in pen six, adjusts the feeding rate, and monitors oxygen levels in pen nine. Then a camera feed freezes in pen three. A winch jams. Behaviour changes abruptly. One pen suddenly demands all her attention.

The question is not whether she is good at her job. The question is what's happening in the eleven other pens while she handles pen three — and how many decisions slipped while the undivided attention was on pen three.

Even the best feeders can't be in twelve places at once.

This is the reality every feeding operation eventually faces: thousands of micro-decisions, a large number of variables, parallel pens, and divided attention. We have some of the best feeders in the world – people who can read fish in ways few outside the industry truly understand. This is not a competence problem. It is a capacity ceiling. And for large parts of the industry, we have already reached it. To push the limits of the capacity ceiling further we must upgrade the support system around the feeders.

The difference that is not biology

The cost of inaction is measurable. In before-and-after analyses with farmers across multiple regions, we see feeding precision vary by as much as 40 percent between pens at the same site. Same fish. Same team. Same feed.

The difference is not biology. It is precision. The largest margins lie in making perfect feeding the standard – every day, in every pen, everywhere. Achieving consistency across an entire operation is one of the most difficult improvements an industry can make.

To achieve consistent precision at the company level, we need to go deeper than the site level. We need to go down to each individual pen – minute by minute, hour by hour, day by day. Because that is where the variation actually occurs. But this isn't an unprecedented challenge. Other industries have done it before — by upgrading the support systems around their best people, and pushing the boundaries of what's possible.

What Formula 1 solved – that aquaculture has yet to solve at scale

In Formula 1, the early decades were about raw horsepower and courage. Only about half the cars finished the races. Then engineering took over: better materials, aerodynamics, and sensors. Variability decreased, and three out of four cars finished.

The big step change came around 2005, when telemetry systems enabled control rooms to process hundreds of variables in real time and provide continuous decision support to drivers and teams – before, during, and after every race. Today, around 95 percent of cars finish. Victories are decided by milliseconds. The sport has not become less competitive. It has become more precise.

The aquaculture industry stands at a similar crossroads. Infrastructure, sensors, cameras, feeding systems, and digital platforms are already in place. What is still missing at scale is the layer that turns all of this into precise decisions executed in real time – across the entire operation, not just in individual pens. This is where artificial intelligence becomes the tipping point.

AKVA group
Technologies such as AKVA submerged provide continuous underwater insight and support more precise and scalable feeding operations. Photo: AKVA group

What AI actually optimises for

AI does not replace the feeder. It handles the routine work – monitoring, observation, and continuous follow-up – allowing the feeder to focus on the assessments only humans can make. It watches where human attention cannot reach, filters out noise, and highlights what actually matters. Multiple parallel data streams are transformed into real-time decision support.

The result is that the two-hundredth decision in a shift can be made with the same quality as the first – in parallel, in real time, across pens, shifts, and sites. The standard set by the most skilled feeder on her best day becomes accessible to the entire organisation. Feeding moves from art to discipline – from individual skill to organisational performance.

But feeding precision is more nuanced than many people think. Precision is not simply about reducing waste. Anyone can do that by feeding less. The problem is that it is rarely good feeding. Good feeding is about hitting the biological optimum. The point where every pellet delivers maximum growth, allowing the fish to reach harvest weight as quickly as possible

Avoiding underfeeding is therefore at least as important as avoiding overfeeding. In practice, underfeeding is often what costs the most: longer production time, higher accumulated biological risk, and weaker price realisation at harvest. For an operations manager, precision in feeding means hitting the right level every day, in every pen – where growth, feed efficiency, and time at sea work together. That is where the greatest value is created.

Predictable precision is no longer just a competitive advantage. It is a requirement. And predictability is the foundation for regulatory trust, social license, and both access to and cost of capital.

The aquaculture industry has historically had a structural disadvantage: biological volatility makes it harder to regulate, harder to defend publicly, and more expensive to finance.

Regulations are tightening. Welfare and sustainability expectations are rising. Both margins and predictability now have to come from many small decisions made correctly, every single day.

The operators who build precision into their daily operations do not just achieve better margins. They earn the right to operate differently — with greater regulatory, operational, and financial freedom. This is also where the industry’s two often conflicting goals converge. Less waste. Lower environmental impact. More efficient resource use. Healthier fish reaching harvest weight faster. The same precision that drives profitability also drives sustainability.

Sustainability without profitability is not sustainable. In feeding, we do not have to choose.

What this means in practice – and how little it actually takes

This is not theory or pilot projects. AKVA group has worked with feeding precision for more than 45 years, and with AI-supported feeding for the past nine. The solutions have been developed in close collaboration with farmers – feeders, operations managers, and experienced site personnel in real production environments.

Today AI-assisted feeding is running in daily operation across more than 2,000 pens, with customers achieving sustained FCR improvements above 0.1 points.

The economics are more forgiving than many think. Less than 0.01 FCR points of improvement covers the cost of an AI solution – in practice, roughly a quarter pellet per fish per day. With actual improvements above 0.1 FCR points, the value is around ten times higher, corresponding to approximately five percent margin improvement across the entire operation, purely from better feeding precision. The strongest performers go even further.

The path to value is to introduce AI step by step, under controlled conditions. By building on investments already made in cameras, sensors, and infrastructure, it also becomes cost-efficient – with gains from day one.

In an industry where feed is the single largest cost, this is the difference between treating feed as an expense to control – and treating it as the daily strategic lever it actually is.

In aquaculture, feed is not just a cost. Every pellet is a profitability decision.