We obsess about model parameters and throughput-and with good reason-but we rarely treat data-centre energy as an architectural constraint that shapes everything upstream: hardware, software, operations, and even national policy. That assumption is breaking down. The emergence of institutions focused explicitly on the intersection of compute and grid planning is a timely reminder: energy is now a first-class citizen in system design.
Context
Oak Ridge National Laboratory has launched a Next Generation Data Centers Institute to study how power, thermal systems, grid integration, security and system-level modelling must evolve to support rapidly growing AI infrastructure. The initiative treats datacentres not merely as compute barns but as coupled assets in energy ecosystems.
What this means for architecture and strategy
Treating datacentres as isolated compute silos is a strategic mistake. As compute density increases, cooling and power distribution become the dominant drivers of cost, reliability and carbon footprint. From an enterprise-architecture perspective, this forces several mindset shifts:
– Systems thinking over siloed optimisation. Compute, power electronics, cooling design, workload schedulers and energy markets must be architected together. The knobs you expose in your control plane should include power-capacity, thermal headroom and grid signals-otherwise orchestration cannot make informed trade-offs.
– Software-defined energy control. Just as we manage networking or storage via software, the next wave is software-defined power and cooling: dynamic voltage/frequency tuning across server fleets, thermally-aware scheduling, and fine-grained load-shedding tied to market signals. These controls change the calculus between expensive over-provisioning and resilient, adaptive operations.
– Build vs. buy decisions now include energy supply. Proprietary cooling or power platforms can yield efficiency gains but increase integration and lock-in costs. Open, standards-based interfaces for power and thermal telemetry make it easier to negotiate with vendors and evolve incrementally.
– Security and resilience baked into the physical layer. “Cyber-informed engineering” is not a buzzword: if the control loops that manage power or cooling are compromised, the damage is physical as well as operational. Zero-trust principles, redundancy in control paths, and quantum-safe links where appropriate should be considered for mission-critical campuses.
– Long-term modelling becomes a strategic capability. Integrated systems modelling-digital twins that couple grid behaviour, workload forecasts and supply-chain constraints-lets organisations anticipate capacity bottlenecks and policy impacts years ahead, turning uncertainty into actionable scenarios.
Actionable priorities for CTOs and founders
– Start exposing energy and thermal telemetry at the application and orchestration layers. If schedulers can see energy prices, carbon intensity and thermal headroom, they can optimise placement and timing automatically.
– Pilot liquid cooling and direct-to-chip solutions on a segment of workloads where density justifies the integration cost.
– Treat demand-response and on-site storage as feature flags: integrate them into your DR and capacity planning exercises today.
– Implement strong segmentation and immutable infrastructure for control-plane components that manage power and cooling; apply Zero Trust.
– Build partnerships with local utilities and research institutions to align capacity planning with grid upgrades and renewables procurement.
– When evaluating vendors, prioritise open telemetry and API access over marginal efficiency gains that come with closed ecosystems.
Relevance for India (and the Northeast)
India’s data-centre growth is similarly stress-testing distribution networks and land-use constraints. There is an opportunity for strategic alignment: colocating capacity where renewable or hydropower is abundant, deploying heat-reuse for local industry, and using region-specific advantages-cooler hill stations in the Northeast, for instance-to reduce cooling loads. For government-led DPI projects, energy-aware architecture reduces long-term operational expenditure and climate impact.
Closing thought
We are entering an era where compute strategy and energy strategy are the same conversation. Architects who design for this coupling-exposing energy as an explicit resource in their systems-will unlock not just efficiency, but new levers of resilience and competitive advantage.
About the Author
Sanjeev Sarma is the Founder Director of Webx Technologies Private Limited, a leading Technology Consulting firm with over two decades of experience. A seasoned technology strategist and Chief Software Architect, he specializes in Enterprise Software Architecture, Cloud-Native Applications, AI-Driven Platforms, and Mobile-First Solutions. Recognized as a “Technology Hero” by Microsoft for his pioneering work in e-Governance, Sanjeev actively advises state and central technology committees, including the Advisory Board for Software Technology Parks of India (STPI) across multiple Northeast Indian states. He is also the Managing Editor for Mahabahu.com, an international journal. Passionate about fostering innovation, he actively mentors aspiring entrepreneurs and leads transformative digital solutions for enterprises and government sectors from his base in Northeast India.
