• • • DATA CENTRE WORLD • • •
CENTERS OF INTELLIGENCE: HOW AI IS RESHAPING
DATA CENTRE ELECTRIFICATION BY JORGE LIS,
GLOBAL SEGMENT LEADER FOR DATA CENTRES, ABB ELECTRIFICATION
A
I workloads are putting data centres’ conventional electrical systems under strain.
Increasing power density, scale and volatile load profiles are pushing designers to rethink long- established system architectures. ABB’s Jorge Lis highlights the tech trends, including Direct Current (DC) distribution, future-proofed infrastructures and standardisation, that are helping data centres grow at scale without compromising availability. AI is driving warp-speed pressure on the capacity and resilience of today’s data centre infrastructures. Gigawatt factories are pushing data centre engineering development to unprecedented extremes. Bigger rack counts, higher power densities and volatile AI-driven workloads are stressing electrical and cooling systems. And as deployment timelines keep shrinking, data centre owners must flex to meet market demand without jeopardising performance. Incremental gains in today’s conventional architectures are no longer sufficient to meet the insatiable demands of AI. As the industry shifts from a megawatt to a gigawatt era, the need for fundamental adjustment of data centres’ electrical systems is clear.
DC: enabling dynamic AI workloads
DC power distribution offers a compelling alternative to traditional AC systems. By reducing conversion stages and components, DC systems deliver better energy efficiency, reliability and
smaller footprints while lowering maintenance costs. For data centres, especially in remote locations, DC native systems integrate seamlessly with renewables, green hydrogen, Battery Energy Storage Systems (BESS) and microgrids, reducing dependence on aging AC grids not built for AI- scale loads.
DC systems also provide faster fault protection critical for power-intensive GPUs and dynamic AI workloads. Higher power density and fault currents demand ultra-fast reaction; solid-state breakers and UPS respond in microseconds versus milliseconds. While DC excels in AI-focused applications, it will complement AC’s proven stability and operational familiarity depending on each application.
Flexibility: building for
AI demand volatility Many data centre operators, particularly colocation providers, face a variety of challenges including unpredictable tenant requirements, evolving power standards, infrastructure investment pressure and risk of obsolescence. This uncertainty has created some paralysis in the market, with operators hesitant to commit to infrastructure designs that may become obsolete within 3-5 years.
A strategic response to this market paralysis needs to be addressed by enabling operators to deploy infrastructure today while maintaining flexibility for tomorrow’s unknown requirements. Scale will require cross-ecosystem partnerships and building a future-proof infrastructure resilient to volatility.
Standardisation: enabling
scalable AI infrastructure To meet the accelerating pace of AI deployment, the industry should converge on shared standards to enable faster, more flexible infrastructure
38 ELECTRICAL ENGINEERING • FEBRUARY 2026
scaling. Industry collaboration through initiatives like the Open Compute Project (OCP) is establishing critical standards for high-density AI data centres. OCP’s Power Distribution sub-project addresses voltage standardisation, with systems evolving from 48V DC to 800V DC and potentially 1,500V DC to support megawatt-scale racks. This standardisation enables centralised architectures to transition to distributed designs built from prefabricated modular elements. Systems constructed from these standardised blocks deploy above 30-50 per cent faster than traditional builds, while component reuse reduces re-engineering costs and supply chain risks. Modular approaches also enable flexible scaling across multiple sites with reduced dependence on specialised workforce training. As voltage and interface standards mature, the industry gains a unified framework for rapid, efficient AI infrastructure deployment.
Adapting to the needs of an AI era
Integrated system engineering end-to-end thinking is essential in an AI-driven data centre landscape, where flexibility and standardisation are key to creating electrical systems that can meet simultaneous demands for reliability, efficiency and scalability. Data Centres are no longer a single category split by application but instead bring their own class of infrastructure, requiring a balance between proven AC systems and emerging DC architectures. As the industry adapts, lessons from early AI- scale deployments and cross-industry collaboration will play a critical role in shaping the next generation of data centre design.
https://www.abb.com/global/en
electricalengineeringmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48
