AI and the energy transition in greenhouse horticulture: from cost item to control variable

Why energy is now at the top of the list

The cost increase has two clear causes. The tax advantage on energy for greenhouse horticulture is being phased out in steps, and since 2025 a CO2 levy has applied to the sector. Together, that means a structurally higher energy bill, not a temporary peak. On top of that, there is grid congestion: in many places, a larger grid connection is simply not available in the short term, so businesses have to work more intelligently with their existing capacity instead of adding to it.

The consequence is that energy shifts from a cost you pay afterwards to a variable you have to control actively: when to light the crop, when to buffer heat, when to buy or sell electricity, and how that relates to what the crop needs at that moment.

Where AI can really add value

The gain is rarely in a spectacular new growing recipe, but in making better use of data that is often already there: climate computer, energy prices, weather forecast and crop measurements. AI can help combine those streams into better short-term decisions, just as with yield forecasting and climate control, so that energy use and crop needs are better aligned.

• Forecasting heat and electricity demand on the basis of weather, crop stage and historical data, so that buffers and lighting are used in a more targeted way.
• Responding to dynamic energy prices: working out when buying, buffering or feeding back into the grid is advantageous, within the limits the crop allows.
• Flagging deviations in the climate or energy consumption before they lead to damage or waste.
• Setting scenarios side by side, so that the grower can compare them instead of blindly following an outcome.

Where the hype begins

Some vendors suggest that an algorithm controls the greenhouse autonomously and optimally. For most businesses, that is not a realistic picture. A model works with assumptions and averages, while cultivation is full of exceptions: rising disease pressure, a batch that reacts slightly differently, a fault in the equipment. Also be critical of unsubstantiated savings claims. The effect varies greatly by crop, by greenhouse and by energy contract, and a generic percentage says little about your own situation. Ask vendors for a trial with your own data instead of a marketing figure.

Subsidies and investment windows

Making your business more sustainable does not have to be paid for entirely out of your own pocket. For 2026 there are again schemes such as Energie-efficiëntie en hernieuwbare energie glastuinbouw (EG, the subsidy for energy efficiency and renewable energy in greenhouse horticulture) and Marktintroductie energie-innovaties (MEI, the subsidy for the market introduction of energy innovations) through RVO (Netherlands Enterprise Agency), with specific application windows. That also makes energy an investment decision with a timetable: an application can be combined with a smarter set-up of control and measurement, so that the investment delivers more than the hardware alone.

What matters is that data and control go hand in hand with the physical measure. A heat buffer or LED lighting delivers the most when its control follows price, weather and crop, and that is precisely where AI-assisted software can make the difference.

The grower decides and signs

At AIBA, our starting point is that the human decides and signs. For energy control, that means in practical terms that the grower remains responsible for the settings, even if a system proposes a setpoint. That is a matter of professional skill and of risk, because a wrong energy decision directly affects the crop.

In addition, Article 4 of the EU AI Act (Regulation (EU) 2024/1689) has applied since 2 February 2025. Since the amendment by Regulation (EU) 2026/1744, in force from 27 July 2026, it has asked organisations to take measures that support the AI literacy of employees who work with AI systems. They do not have to guarantee a specific level. For growers and cultivation managers, what matters in practice is that they understand what a control system can and cannot do, where the assumptions and limitations lie, and when a proposal must be adjusted manually.

Where to start

Start small and make it measurable. Choose one specific process where energy and cultivation come together, for example the control of lighting or the use of the heat buffer, and first record how it is done now: what it costs, which decisions the grower takes and when, and on what basis. Only after that can you judge whether AI support measurably saves time or money there.

Our free Practice Scan helps you choose that first process and sharpen the baseline measurement, so that any next step rests on real figures and not on a promise. How such a project then proceeds, including the question of how you gain access to the data from your own climate computer, is set out on our page on AI implementation in greenhouse horticulture (in Dutch).