Magnetic technology in AI data centres: Small components for critical infrastructure

Artificial intelligence is changing more than software, business models and digital services. It is also transforming the physical infrastructure that supports these applications. Modern AI data centres require enormous computing capacity, high power densities and, above all, a power supply that is always available.

Development is moving rapidly. According to recent analysis by the International Energy Agency (IEA), global electricity consumption by data centres rose by around 17 per cent in 2025. Consumption by AI-focused data centres increased by approximately 50 per cent. Total electricity consumption by data centres is expected to almost double, from around 485 TWh in 2025 to approximately 950 TWh by 2030.

This creates new demands across the entire energy and infrastructure chain. Grid connections need to become available sooner, on-site generation is being added in some locations, and backup systems are becoming increasingly important. This is precisely where components that initially seem to have little to do with artificial intelligence come into play: solenoids. 

They control valves, move armatures, switch fuel and gas systems, actuate safety devices and perform precise switching functions. Injector solenoids in generator sets are a particularly interesting example, bringing the demands of AI infrastructure into direct contact with high-precision electromagnetic actuation.

An injector solenoid may need to operate reliably for a very long time while withstanding an extremely high number of switching cycles. Several stresses act on it simultaneously.

The first challenge is thermal loading. Engines and generators operate in environments where elevated temperatures can occur. The solenoid coil must maintain its electrical and magnetic characteristics as consistently as possible, even at high temperatures.

Dynamic response is another challenge. An injector must open and close within a very short time. Even small deviations in magnetic force, armature travel or switching time can affect metering.

Service life presents a further challenge. Stationary power generation systems can accumulate very high switching cycle counts. The actuator must therefore do more than perform a single switching operation: it must maintain repeatable characteristics over a substantial part of its service life.

Finally, materials, coatings, seals and the magnetic circuit must be matched to the application. Gaseous fuels and hydrogen, in particular, create additional requirements for media compatibility and leak-tightness. Modern hydrogen injectors, for example, are developed using specialised sealing and material concepts. For hydrogen applications, We explain how these requirements influence the design of electromagnetic components in our article on solenoid valves for fuel cell systems and hydrogen combustion engines.

For amanufacturer of magnetic technology , the real challenge is not simply making a solenoid switch. It is ensuring that a precisely defined switching operation can be reproduced over millions of operating cycles under real thermal, mechanical and media-related conditions.

Data centre power supply concepts are also evolving. Alongside conventional diesel generators, natural gas generators, hybrid systems and, in the longer term, hydrogen solutions are being considered.

The IEA expects natural gas to make an important contribution to meeting the additional electricity demand from data centres in the coming years. At the same time, renewable energy and energy storage solutions are expanding.

Industry is already responding to this development. In 2026, for example, Cummins announced natural gas generators for a large-scale AI data centre campus in Texas. The planned systems are intended to provide primary power behind the grid connection point, helping to address constraints on grid capacity.

This opens up additional applications for electromagnets. Regardless of the energy source used, modern combustion systems require precise actuators to regulate fuel and gas flows.

Hydrogen introduces further requirements. Its physical properties differ from those of diesel or natural gas, placing demanding requirements on seals, materials and the design of components in contact with the medium. Electromagnets for hydrogen technology must therefore be engineered to work with these components and the relevant pressure and temperature conditions. Hydrogen engines can use either port or direct injection systems. Manufacturers of hydrogen injection systems describe hydrogen injectors as components designed to ensure precise metering and low leakage, even under demanding engine conditions.

This creates an interesting development area for custom-engineered electromagnets. The actuator must do more than function electromagnetically: it must become part of an overall system that operates with a demanding medium, high temperatures and stringent requirements for safety and availability.

The development of AI infrastructure is also changing the technical requirements for its supporting systems. Microsoft, for example, describes its new AI data centres as facilities specifically designed for large AI clusters. The company is building extensive power, network and cooling infrastructure to support them.

Google is also developing its own AI accelerators and data centre infrastructure, while working on solutions that allow its data centres' power demand to respond more flexibly to the needs of the electricity grid.

AWS is likewise investing heavily in AI infrastructure. The company explicitly identifies growing demands for computing power, networks and resilient infrastructure as a challenge for the next generation of AI applications.

This creates a market in which computing capacity is not the only thing increasing. Demands on power supply, cooling, safety and automation are also rising.

For electromagnetic actuators, this means more demanding requirements in the following areas:

• Force density: Greater performance must be available within a smaller installation space.
• Dynamic response: Switching and control operations must be fast and repeatable.
• Thermal performance: Solenoid coils must operate reliably even at elevated ambient temperatures.
• Service life: High availability places correspondingly demanding requirements on fatigue resistance.
• Safety: Gas-handling systems may introduce additional requirements for explosion protection and functional safety.
• Diagnostic capability: As systems become increasingly digitalised, early detection of operating conditions and faults becomes more important.

These requirements cannot be met by a single standardised solenoid solution for every application. Custom-engineered electromagnetic actuators are needed.

The pace of the AI market also affects the supply chain. When major technology companies build new data centre capacity within short timeframes, buildings, servers and power supplies are only part of what needs to be available. Numerous components of the technical infrastructure must also be developed, validated and prepared for series production on time.

The IEA now highlights significant bottlenecks throughout the supply chain for data centre and energy infrastructure. These include transformers, gas turbines and other electrical and energy-related components. At the same time, data centres can be built much faster than much of the energy infrastructure that supplies them.

Time to market is therefore becoming increasingly important for manufacturers of electromagnetic actuators.

For custom electromagnets, technical quality alone does not determine project success. What also matters is how quickly a technical requirement becomes a robust concept, a concept becomes a functional prototype, and a prototype becomes a product ready for series production.

This is where short decision-making processes, rapid prototype production, simulation, close customer communication and early industrialisation play a decisive role.

At Magnetbau Schramme, this means treating development projects as more than isolated design tasks. Development, project management, manufacturing and quality assurance work closely together from the outset. This allows technical decisions to be made faster, prototypes to be supplied at short notice and findings to be incorporated directly into the next stage of development.

We explore the factors that make rapid development and successful industrialisation possible in our article, “The 8 success factors for rapid development and successful industrialisation”.

The computing power of an AI data centre is generated by GPUs, CPUs and specialised accelerators. For these components to operate reliably, however, they need a highly complex infrastructure comprising power supply, cooling, switchgear, safety systems and backup power.

Electromagnets are small but functionally critical components within this infrastructure. They control valves, move armatures, meter fuels and other media, and perform switching and holding functions.

The injector solenoid in a generator set demonstrates particularly clearly how these requirements come together: fast response, compact dimensions, thermal resilience, exceptionally long service life and reliable operation even when the data centre's power supply is under critical stress.

The rapid expansion of AI infrastructure is therefore also increasing demands on suppliers. Technical performance alone is not enough. Development speed, prototype availability, industrialisation expertise and the ability to reliably bring custom solutions into series production under demanding conditions are equally important.

For an electromagnet manufacturer, this creates a new and technologically demanding field of activity, at the intersection of electromagnetic precision and one of the most important infrastructure developments of our time.

1. IEA – Key Questions on Energy and AI (2026): https://www.iea.org/reports/key-questions-on-energy-and-ai
2. Microsoft – Inside the world’s most powerful AI data centre (2025): https://blogs.microsoft.com/blog/2025/09/18/inside-the-worlds-most-powerful-ai-datacenter/
3. Google – How we’re making data centres more flexible to benefit power grids (2025): https://blog.google/innovation-and-ai/infrastructure-and-cloud/global-network/how-were-making-data-centers-more-flexible-to-benefit-power-grids/
4. AWS – Powering innovation at scale: How AWS is tackling AI infrastructure challenges (2025): https://aws.amazon.com/blogs/machine-learning/powering-innovation-at-scale-how-aws-is-tackling-ai-infrastructure-challenges/
5. Cummins – Natural Gas Generators to Power Large-Scale Data Centres in West Texas (2026): https://investor.cummins.com/news/detail/697/cummins-natural-gas-generators-to-power-large-scale-data
6. European Commission – ATEX: Equipment for explosive atmospheres: https://single-market-economy.ec.europa.eu/single-market/goods/european-standards/harmonised-standards/equipment-explosive-atmospheres-atex_en
7. IEC – IEC 60079-0:2026 / Explosive atmospheres: https://webstore.iec.ch/en/publication/71519
8. VDI 2221 – Methodology for the development and design of technical systems and products:
   https://www.vdi.de/richtlinien/details/vdi-2221
9. ANSYS – Maxwell Electromagnetics Simulation:
   https://www.ansys.com/products/electronics/ansys-maxwell