Digital transformation and rapid AI adoption are fueling an unprecedented demand for AI servers and high-performance computing. This surge is driving a steep rise in electricity and water consumption across data centers. By 2030, global electricity demand from data centers is set to more than double by 2030 to around 945 terawatt-hours (TWh), surpassing the current annual consumption of Japan. AI-optimized data centers will contribute significantly to this increase, with its energy use expected to quadruple by 2030.
This puts pressure on operators to reduce their environmental footprint while maintaining uptime and performance. This shift is prompting a complete rethink of how power is sourced, how systems are cooled, and how hardware is designed. The focus is on long-term scalability, operational resilience, and environmental responsibility.
Looking at what Commercial Agreements are taking place within the industry provides a real-time view into the latest data center developments. This article highlights the key deals, partnerships, and innovation milestones within the budding sustainable data center ecosystem to help leaders spot emerging opportunities.
Objective: Foxconn, the world’s largest AI server supplier, is working to lower its partial Power Usage Effectiveness* (PUE) and reduce carbon emissions in data centers.
Overview: SoftBank, ZutaCore, and Foxconn are co-developing a rack-integrated, high-efficiency cooling solution based on ZutaCore’s two-phase Direct Liquid Cooling (DLC) technology. Foxconn is deploying the system in AI servers powered by NVIDIA H200 GPUs, hosted in SoftBank-designed racks with integrated cooling.
Outcome: In a February 2025 evaluation, the system achieved a PUE of 1.03 at a SoftBank data center—far outperforming traditional setups. It also passed NVIDIA’s NVQual temperature test, confirming its reliability and performance. The partners plan to scale this solution across other data centers and factories, supporting the growth of sustainable data center infrastructure.
-
Strategic Partnership: Castrol, Supermicro, and Submer
Objective: To demonstrate how immersion cooling can serve as a scalable, sustainable thermal management solution for high-performance computing.
Overview: Castrol, Submer, and Supermicro collaborated to integrate Submer’s immersion cooling technology and Castrol’s dielectric coolant into Supermicro’s AI server systems. These systems support power-hungry AI workloads with thermal loads of up to 1000W per node.
Outcome: This solution significantly lowers energy consumption while improving performance. It also reduces reliance on water-intensive cooling methods. It is an important step toward building sustainable data centers with lower carbon and resource footprints.
-
Service Agreement: Digital Realty and Coolgradient
Objective: Digital Realty is targeting carbon neutrality for Scope 1 and 2 emissions by 2030, with a strong focus on optimizing energy efficiency across its data centers.
Overview: Digital Realty, a global operator with more than 300 data centers, is partnering with AI platform Coolgradient. The collaboration seeks to improve energy and operational efficiency. Coolgradient’s technology integrates with Digital Realty’s Apollo system to deliver real-time insights that help:
- Enhance energy efficiency
- Reduce water consumption
- Predict and prevent system failures
- Support smarter, data-driven site operations
Outcome: The partnership has already saved 18 GWh of energy over 2.5 years. It has helped contribute to the company’s broader sustainability targets and accelerate the transition toward sustainable data centers.
Objective: DTU aims to reduce energy usage, cut costs, and improve grid interaction at its data center facilities.
Overview: Cartesian is deploying its Thermal Box™ cold energy storage system at DTU’s Copenhagen campus. The system produces and stores cold energy during off-peak hours when temperatures are lower and uses it to cool data centers during peak hours.
Outcome: This approach enhances cooling efficiency and supports peak shaving (reducing electricity usage during peak demand periods) and load shifting (moving energy use to off-peak times). It reduces energy costs and stabilizes demand besides aligning with DTU’s commitment to sustainable data center practices.
Objective: NeevCloud seeks to reduce water use by 90% and energy consumption by 40% across its AI data centers in India.
Overview: ZutaCore’s HyperCool® technology is a waterless direct-to-chip liquid cooling system that is being deployed in NeevCloud’s AI infrastructure. The partnership is backed by a revenue-sharing model with local data center providers.
Outcome: The solution reduces water usage by up to 10 million litres per megawatt of IT load per year. It also cuts electricity usage by more than 100 million kWh annually. These improvements lower operational costs, reduce emissions, and contribute to a more scalable model for data centers in high-growth markets.
Explore how Commercial Agreements are boosting the adoption of Sustainable Data Centers
These commercial agreements reveal how Climate Tech is becoming a core aspect of data centers’ operations. Furthermore, each deal represents a strategic collaboration that accelerates decarbonization and boosts resource efficiency that can meet rising digital demand without compromising environmental goals.
For enterprise operators, investors, and technology leaders, tracking these developments is key to understanding the innovation landscape. Whether it’s new cooling methods, AI-driven energy optimization, or water-saving systems, these partnerships offer a window into what’s working—and what’s next—for sustainable data centers.
Have a look at our dataset today and discover the most impactful commercial activity in the data centers vertical. Request a free trial to uncover key trends, deals, and partnerships shaping the future of green data centers.
* PUE stands for Power Usage Effectiveness. It’s a metric used to measure the energy efficiency of a data center by comparing the total energy consumed by the facility to the energy used by its IT equipment. The closer the number is to 1.0, the more energy efficient the data center is.
