04 Mar 2026
10 minutes to read
The traditional power structures that supported Data Centres even five years ago are no longer sufficient.
AI is accelerating demand. Grid capacity is constrained. Planning, permitting and emissions requirements are tightening. Carbon neutral roadmaps are now embedded into investment decisions. Energy strategy has become central to Data Centre viability.
At RED Engineering Design, we have seen this shift at scale. In 2017, we master-planned 1.2GW of capacity. By 2024, that had increased to 8.3GW. In 2025 alone, we master-planned more than 15GW of new campuses to support the AI revolution.
This is not incremental growth. It represents structural change in digital infrastructure - and it demands a structural response.
Data Centre operators are navigating an environment shaped by:
The industry is moving rapidly, and our designs must anticipate what will happen over the next 10 to 15 years - not simply respond to today’s requirements.
Futureproofing energy systems is highly dependent on the IT equipment installed within facilities. Machine learning workloads and inference workloads behave very differently. Power profiles are increasingly erratic. Demand spikes are sharper. Cooling loads are less predictable.
Meanwhile, water usage is becoming as critical as energy consumption. Cooling strategy can no longer be an afterthought - it is central to infrastructure design.
Power availability is one of the defining challenges for new development. Network constraints, renewable intermittency and DNO connection conditions require detailed power studies - including assessment of fault currents, harmonics and AI demand spikes. Grid interaction has become a complex relationship. Resilience strategies increasingly include:
Data Centres are no longer passive loads. They are active participants within wider energy ecosystems. Resilience must extend beyond the building envelope - across grid, generation and storage infrastructure.
AI has introduced a step change in power density and cooling requirements. RED is designing liquid-cooled hyperscale environments capable of supporting up to 150kW per cabinet. Higher density reduces white space and concentrates load. Demand volatility becomes a design condition. Protection of supply must extend upstream - to grid, generators and BESS - to manage ramp-up characteristics and erratic demand profiles.
Water usage is also becoming critical alongside the energy used for cooling. Cooling strategy now influences site selection, permitting and long-term operational performance.
Engineering resilience in this context requires coordinated thinking across electrical, mechanical and infrastructure systems.
RESILIENCE AND SUSTAINABILITY
Resilience and sustainability increasingly intersect. Redundancy can increase carbon impact. Cooling strategies influence water consumption. Backup assumptions may conflict with decarbonisation objectives.
The response is not to prioritise one over the other. It is to define a clear roadmap to carbon neutrality and minimum water use over a 10–15 year horizon - and engineer resilience within that framework.
THE ROADMAP TO ZERO CARBON: THE 4DS FRAMEWORK
At RED we frame the transition to sustainable resilience through four strategic pillars:
1. Demand Reduction
We accept that overall energy demand will rise. But optimisation remains critical.
This includes software streamlining, server load mapping, cooling system optimisation, reduction of distribution losses, adoption of DC power systems, and CFD modelling to refine airflow regimes.
2. Decarbonisation
Renewable integration - wind, solar, geothermal - must be paired with flexible strategies such as PPAs, seawater cooling, e-fuels, biofuels, SMRs and carbon capture. The key is adaptability. Technology maturity will evolve over the next decade. Infrastructure must be capable of integrating new supply chains as they develop.
3. Decentralisation
Behind-the-meter generation and storage are becoming essential. Microgrid architectures enable bridging power, permanent generation, and participation in grid stabilisation services.
4. Digitalisation
Digital twin modelling, power ramp-up simulations, RAMS analysis, CFD dispersion studies, and machine learning-driven optimisation are now embedded throughout the lifecycle.
EVOLVING DELIVERY MODELS
longside technical change, we are also seeing commercial evolution. Independent Power Producers are increasingly partnering with Data Centre operators under long-term concession models - delivering scalable primary and backup power, integrated cooling, and continuous investment through back-to-back SLAs. This approach can reduce upfront capital burden while enabling long-term carbon optimisation. It reflects a broader truth: energy infrastructure is no longer static. It must evolve continuously over the asset lifecycle.
LOOKING FORWARD
The pace of innovation will not slow. Quantum computing, advanced nuclear technologies, AI-optimised energy management and sovereignty-driven infrastructure strategies are already influencing long-term planning. The precise mix of technologies may change. The need for adaptable infrastructure will not.
A FINAL THOUGHT
As you have seen, the conversation has moved well beyond simply securing grid capacity. Designing modern Data Centres now means balancing density with flexibility, resilience with decarbonisation, and immediate delivery with long-term adaptability. We must stop designing for today’s equipment cycle and start engineering for the next decade of uncertainty. If we embrace integration - technical, digital and strategic - complexity becomes manageable. And futureproofing becomes achievable.
04 Mar 2026
