Why AI Places Data Centres at a Sustainable Power Crossroads

NTT’s latest sustainability report sets out a stark reality for data centre operators and their telecommunications partners: the rapid expansion of generative AI is driving a power requirement that existing infrastructure struggles to support. 

In the NTT Global Data Centers 2025 Sustainability Report, the company notes that AI-driven workloads could increase global data centre power consumption by a factor of 13 by 2030, with implications for grid capacity, cooling design and long-term commercial viability.

Doug Adams, CEO at NTT Global Data Centers, describes the challenge: while “data is central to our digital age”, its environmental impact “could increase dramatically with the rapid rise of generative AI.” 

Doug positions sustainability as essential to operational continuity, framing it as a “commitment to ensuring the AI revolution is a sustainable one”.

Power, sustainability and the definition of a premier provider

The report argues that the traditional definition of a leading data centre partner has shifted. The priority is no longer limited to latency or connectivity but the ability to secure power, guarantee stability for high-density AI deployments and manage associated emissions.

NTT’s own ambition is framed in these terms, with Doug setting out the goal of “being the world’s premier data centre provider in a digitally connected world that is both harmonious and sustainable”. The report supports this vision with targets aligned to science based targets, including net zero emissions across the full value chain by 2040.

This redirection reflects rising pressure on data centre operators to ensure that growth in digital workloads does not exceed the limits of available infrastructure or regulatory tolerance.

A three-part strategy for future data centre design

NTT Global Data Centers builds its approach on a three part operational framework: consume less, do better and do different. Each pillar connects directly to the technical requirements of running large-scale data centre estates under increasing power constraints.

Applying sustainability principles to high-density workloads

The report outlines how these principles translate into operational practice. A key example is the deployment of direct liquid cooling at NTT’s Navi Mumbai facility, designed to support demanding AI environments. The report describes how the site manages heat loads “higher than 20kW per rack” and has improved overall energy efficiency by “almost 30%”.

The Mumbai project highlights how AI workloads are accelerating the adoption of cooling technologies that reduce energy consumption while supporting higher rack densities. As liquid cooling becomes more common, data centres are beginning to reverse the traditional link between compute intensity and inefficiency.

Another case study comes from Berlin, where NTT is supplying captured waste heat to a nearby district network. By acting as a heat source for local users, the facility demonstrates how data centres can optimise thermal output and reduce strain on local energy systems.

Risks from water usage and Scope 3 emissions

NTT’s reporting also details the constraints that may influence future facility design. Water usage rose by 29%, driven by evaporative cooling at high load sites. The report highlights the “Water Power Nexus” as a growing concern, especially where water scarcity limits cooling strategies.

Scope 3 emissions remain another challenge. Client IT load is still the largest obstacle to increasing NTT’s renewable energy share, underlining the difficulty of decarbonising workloads hosted within data centres.

NTT’s analysis sets out a framework for how operators can manage rising AI power requirements while balancing efficiency, renewables procurement and the technical realities of advanced cooling.