Artificial intelligence (AI) is a megatrend that is affecting the entire industrialized world. AI has the potential to fundamentally transform our society and economy, and the potential is enormous.
And just as enormous as the potential of AI is the energy requirement of data centers, the backbone of this technology. Data centers are already among the largest energy consumers worldwide. According to calculations by the consulting firm International Data Corporation (IDC) global data center power consumption will double between 2023 and 2028 with a five-year annual growth rate of 19.5 percent. The Electric Power Research Institute predicts that data centers are expected to require up to nine percent of all electricity generated in the US by the end of the 2020s.
The problem is similar in many markets: the expansion of power grids is not keeping pace with the data center boom. Grid connections are delayed, high-performance connections are scarce and approval procedures take a long time. In spring 2026, for example, the city of Frankfurt am Main announced that large, high-performance new connections cannot be expected in the home of one of the world's largest internet hubs until the mid-2030s.
The insatiable power demand of AI is overtaking the decades-long development cycles of the power grid, leading to a critical bottleneck.
Goldman Sachs
VP Analyst at Gartner [4]
More and more data center operators are therefore turning to on-site power generation. Generating power directly on site ensures greater planning security, faster project implementation and greater resilience to grid outages or delays in grid conversion. On-site power can therefore serve as a primary supply, as a bridge to the grid connection or as a supplement to the grid.
Three technology paths have emerged as energy suppliers: Gas turbines for large, flexible output, gas engines for modular and often more efficient multi-engine concepts and fuel cell systems for quiet, scalable and lower-emission continuous power solutions.
Gas turbines
Gas turbines are particularly suitable for larger power leaps and for projects in which high power must be available quickly. They are particularly interesting when a large data center campus or grid bottlenecks require strong on-site generation. The benefits of gas turbines for data centers are their high power density and suitability for large plant concepts. Their disadvantage is usually their lower modularity compared to multi-engine setups.
Gas engines
Gas engines are considered a particularly strong on-site power model when data centers want to grow modularly, ramp up loads gradually or ensure high availability with many smaller units. They can be used very well in multi-engine architectures, making maintenance and redundancy easier and reducing the impact of individual unit failures.
They are also attractive for operators because they can be ramped up quickly and can also be used as an island grid or for grid support. Gas engines are therefore often the pragmatic answer to a lack of grid capacity and long waiting times for grid connections, particularly in regions where electricity is scarce.
Fuel cell systems
Fuel cell systems are an interesting option for a quiet, scalable and locally low-emission power supply. They can operate with very low emissions on site, especially when running on hydrogen. This technology can therefore significantly reduce the ecological footprint of data centers and thus contribute to air quality and climate protection.
Fuel cell systems can be used as a primary supply or as a supplement to the grid and are therefore of interest to operators who want to combine security of supply with a reduction in emissions.
Filtration is a decisive factor for all three types of power generation
Filtration is a performance, availability and efficiency lever for AI infrastructure in all on-site power technologies: it protects sensitive components from particles, moisture and other contaminants, reduces pressure losses, extends service life and thus reduces failure risks and operating costs. Filtration therefore has a direct impact on the performance, service life and cost-effectiveness of on-site power.
Gas turbines need clean intake air because particles and moisture can contaminate the turbine blades, compressor stages and the entire system. The aim is not only to protect the gas turbine, but also to ensure stable operation: reliable filtration helps to reduce power losses, wear, maintenance costs and fuel consumption. Hengst therefore offers Air intake filter systems including Filter mats, Pocket filters, Filter cartridges and Compact filters for gas turbines.
Filtration is important for gas engines because the intake system and the engine compartment need to be protected from dust, dirt and other airborne contaminants. This protection is particularly important for data centers that rely on high availability, as unplanned downtimes can quickly become expensive. The right filters reduce wear on cylinders, pistons and other components and have a direct impact on maintenance costs and service life. For gas engines, Hengst primarily offers Air intake filter systems with the corresponding Pocket filters or Filter cartridges.
For fuel cells , the Air treatment for cathode air filtration is crucial because the cell needs clean, low-emission and sufficiently available air to generate electricity efficiently. Hengst's customized filtration solutions, for example, are designed to keep the air supply to fuel cells clean and free of particles and harmful gases, which increases their service life and optimizes efficiency.
How air filters influence energy requirements in data centers
While gas turbines, gas engines and fuel cell systems are solutions for the energy supply of data centers, the energy requirements of a data center's ventilation technology can also be positively influenced with filtration.
Data centers are equipped with air conditioning units that cool down the air heated by the IT technology. This is usually done in recirculation mode with high air exchange rates. The greatest energy requirement here is caused by the fans that draw in the air. With differential pressure-optimized and therefore energy-efficient air filters, the systems can be designed in such a way that energy consumption can be reduced. This also reduces CO₂ emissions and saves costs for the data center operator at the same time.
Conclusion
AI, on-site power and filtration are more closely connected than it seems at first glance. With the growing energy requirements of data centers, local power generation is becoming increasingly important. In order for gas turbines, gas engines and fuel cells to reliably fulfill this task, cleanly designed filtration is required.
For technical decision-makers, this means that anyone thinking about on-site power in the data center, security of supply and energy efficiency should include filtration in the system design at an early stage. It influences availability, efficiency, maintenance and service life and is therefore a relevant building block for high-performance and future-proof AI infrastructure.
FAQ
What is on-site power in the data center?
On-site power in the data center is the generation of electricity directly on site. It is used to bridge grid bottlenecks, increase security of supply and implement projects more quickly. On-site power can be used as a primary supply, as an interim solution or as a supplement to the power grid.
Why is the energy demand of data centers increasing due to AI?
The energy requirements of data centers are increasing due to AI, because the training and operation of modern AI models require a great deal of computing power. The power requirements for cooling, airflow and electrical infrastructure are also growing. The higher the power density, the more important resilient energy concepts become.
Why do data centers rely on their own power generation?
Data centers rely on their own power generation because grid connections and grid expansion are not keeping pace with growth fast enough in many places. On-site power improves predictability and reduces dependencies on external infrastructure bottlenecks. For operators, this is particularly relevant for rapidly scaling projects.
Which technologies are suitable for on-site power in the data center?
Gas turbines, gas engines and fuel cell systems are particularly suitable for on-site power in data centers. Gas turbines are suitable for high outputs, gas engines for modular and redundant concepts, and fuel cells for quiet and locally low-emission supply solutions. The choice depends on the application scenario.
Why is filtration important for gas turbines in data centers?
Filtration is important for gas turbines in data centers because compressors and turbine blades depend on clean intake air. Particles and moisture can cause wear, loss of performance and higher maintenance costs. Good filtration supports stable and efficient operation.
Why is filtration important for gas engines in data centers?
Filtration is important for gas engines in data centers because the intake system and engine components need to be protected from dust and air contamination. This reduces wear and tear and improves the predictability of maintenance and availability. This is a key operating factor in high-availability data centers.
Why is Filtration Fuel Cells important in data centers?
Filtration Fuel Cells in data centers are important because fuel cells require clean cathode air for stable performance. Particles and harmful gases can have a negative impact on efficiency and service life. Coordinated air treatment therefore contributes directly to operational safety.
How does filtration affect energy consumption in the data center?
Filtration influences the energy requirements in the data center because air filters also determine the pressure loss in air conditioning and ventilation systems. The lower the pressure required, the less energy the fans need. Efficient air filters can therefore reduce operating costs and power consumption.
What are the consequences of unsuitable filtration in the data center?
Unsuitable filtration can lead to increased wear, reduced efficiency, more maintenance and unplanned downtime. This applies to on-site power systems as well as air conditioning and ventilation technology. In critical infrastructures such as data centers, such effects can cause considerable costs.
Is filtration only relevant for power generation?
No. Filtration is also relevant for cooling, air quality and energy efficiency in data centers. It influences how efficiently air can be moved and cleaned, especially in recirculation and air conditioning systems. Filtration therefore affects several infrastructure levels simultaneously.
How does filtration contribute to more sustainable data centers?
Filtration contributes to more sustainable data centers because it reduces wear and tear, stabilizes efficiency and can reduce the energy requirements of air systems. This reduces resource consumption, maintenance costs and indirect emissions. Sustainability is achieved primarily through more reliable and efficient operation.
