Data Centre Trends for 2025: Vertiv Predicts Industry Innovations to Support AI

Innovation in AI rack cooling and powering, as well as energy and emissions control, will be priorities in the coming year.

Singapore, December 3, 2024―As artificial intelligence (AI) continues to revolutionise the data centre industry, 2025 is set to be a pivotal year marked by significant advancements. Vertiv (NYSE: VRT), a global leader in critical digital infrastructure solutions, has unveiled its predictions for the year ahead. These trends focus on enabling high-density computing, addressing regulatory challenges, and driving sustainability and cybersecurity innovations.

AI and Industry Evolution

Giordano (Gio) Albertazzi, CEO at Vertiv, says the adoption of AI is accelerating rapidly, creating unparalleled demand for advanced technologies. “Our experts rightly identified the proliferation of AI and the shift to sophisticated liquid- and air-cooling strategies as a key trend for 2024. This momentum is set to intensify in 2025,” Albertazzi stated. He highlighted the need for scalable solutions to manage AI-driven rack densities, which now range from hundreds to thousands of kilowatts. “The industry must respond with innovative systems to power and cool these AI factories efficiently while minimising environmental impact,” he added.

Here’s a closer look at Vertiv’s key predictions for 2025:

1. Power and Cooling Infrastructure to Match Computing Densification: AI’s compute-heavy workloads will continue to challenge traditional power and cooling systems. As data centres shift from CPUs to GPUs for their superior parallel processing capabilities, the higher thermal demands of modern chips will necessitate new solutions.

• Hybrid Cooling Systems: A mix of liquid-to-liquid, liquid-to-air, and liquid-to-refrigerant cooling methods will become standard, deployed across brownfield and greenfield sites alike.
• Liquid Cooling Integration: Systems such as cold plate and immersion cooling will increasingly be paired with high-density uninterruptible power supply (UPS) systems to maintain continuous operation.
• Factory-Integrated Servers: Servers pre-integrated with liquid cooling systems will improve manufacturing efficiency, accelerate deployment, and reduce energy consumption.

This shift will also affect enterprise data centres as AI expands beyond early adopters in cloud and colocation sectors.

2. Tackling Energy Availability Challenges: The global rise in energy consumption driven by AI presents growing challenges. While data centres currently account for 1-2% of global energy use, this figure could rise to 3-4% by 2030. Such demands may strain electrical grids, driving up costs and emissions while attracting increased regulatory attention.

• Energy Alternatives: In 2025, we will see accelerated adoption of microgrids, fuel cells, and alternative battery chemistries to meet these challenges.
• Long-Term Solutions: Multiple companies are developing Small Modular Reactors (SMRs) are being developed to provide efficient power for large-scale facilities, with implementation anticipated later this decade.

3. Industry Collaboration to Enable AI Factories

AI Factory racks, predicted to require power densities of 500 to 1,000 kW or more, represent an unprecedented disruption to the data centre industry. Addressing these demands will require collaboration across stakeholders, including chip manufacturers, power and cooling solution providers, and utility companies.

• Integrated Solutions: Development tools powered by AI will streamline the design and manufacturing of IT infrastructure.
• Standardisation and Partnerships: Industry players will work together to create roadmaps that integrate IT systems with supporting infrastructure, paving the way for seamless AI adoption.

4. AI’s Dual Role in Cybersecurity

AI complicates and simplifies cybersecurity: Ransomware attacks are becoming more frequent and severe, prompting a broader look at cybersecurity protocols and the data centre community’s involvement in avoiding them. One-third of all attacks last year featured ransomware or extortion, and bad actors are using AI to increase their attacks, broaden their reach, and use more advanced methods. Attacks increasingly begin with an AI-supported breach of control systems, embedded devices, or connected hardware and infrastructure systems that may not fulfil network security standards. Even the most advanced data centre might fail without adequate care.

Cybersecurity professionals, network administrators, and data centre operators must create advanced AI security systems to keep up with hackers’ use of AI to boost attack frequency. Defence in depth and extreme diligence remain unchanged, but changing attack types, sources, and frequencies complicate cybersecurity operations.

5. Regulatory Focus on AI and Sustainability

Government and business regulations address energy use and AI applications: In 2025, we anticipate that regulations may progressively address the use of AI itself, whereas in 2023, they concentrated on government rules for energy usage. Globally, governments and regulatory agencies are rushing to evaluate the ramifications of AI and create policies for its application. The European Union’s Artificial Intelligence Act and China’s Cybersecurity Law (CSL) and AI Safety Governance Framework both centre on the trend towards sovereign AI, which refers to a country’s control or influence over the creation, application, and regulation of AI as well as regulatory frameworks intended to govern AI. As evidence of the trend’s trajectory, Denmark recently unveiled its own sovereign AI supercomputer, and numerous other nations have started their own sovereign AI initiatives and legislative procedures to advance regulatory frameworks. Restrictions are feasible, if not likely, and guidance of some kind is unavoidable.

Driving Innovation in Asia Pacific

Paul Churchill, Vice President and General Manager at Vertiv Asia, highlighted the region’s leadership in AI adoption: “Enterprises in Singapore, Malaysia, and Australia are at the forefront of using AI to unlock efficiency, enhance customer experiences, and solve complex challenges. Our 2025 predictions underscore the urgency of investing in energy-efficient, innovative infrastructure to maximise AI’s potential.”

He assured that Vertiv is well-equipped to support customers with a robust portfolio of end-to-end solutions designed to meet the demands of high-performance computing environments.

With 2025 poised to be a transformative year for data centres, the industry must rise to the challenges of AI, sustainability, and cybersecurity. Vertiv’s forward-looking approach promises to equip businesses with the tools they need to thrive in an increasingly complex landscape.