Cadence issues 2025 data centre predictions

Author: Simon Rowley

Experts at Cadence, an electronic systems design specialist, have given their data centre predictions for the year ahead – and have addressed whether or not data centres can keep up with the AI demand in 2025. Here’s a round-up of their commentaries.

Balancing AI Growth and Sustainability – Mark Fenton, Product Engineering Director at Cadence

In 2025, data centres will face mounting pressure to reconcile AI’s surging energy requirements with strict sustainability goals, sparking an industry-wide rethink on AI applications. The infrastructure required to deliver on AI is poised to drive a 160% increase in data center power demand. This challenge is creating a pivotal moment for data centres to support high-density compute loads while advancing their environmental commitments.

Companies will face a new crossroads. Many that initially rushed into AI, driven by competitive urgency, will now re-evaluate its financial and energy impact, with some in-house setups costing up to $300,000 in hardware alone. This shift is likely to push organisations toward selective, high-value AI applications that provide stronger operational returns, including within data centres themselves.

However, demand will still remain at a high level, stretching capacity to its limits. As such, tools like digital twins will be essential for data centres to meet AI goals sustainably, allowing operators to proactively manage power, integrate renewable sources, and optimise cooling measures to meet AI’s GPU usage demands. With these advancements, data centres can help organisations make AI investments both impactful and environmentally responsible.

Data Centres AI Era Revamp – David King, Senior Principal Product Engineer, Cadence

Data centres will face a pivotal moment next year as energy usage, especially to power AI, continues to rise. Facilities will start to realise that there really isn’t enough energy. While newly built, AI-optimised facilities can be better suited to handle these requirements, retrofitting older data centres to support increased power, as well as cooling, is required to meet demands. However, it is costly and complex. This pressure is prompting operators to plan both infrastructure upgrades and invest in purpose-built facilities designed to power AI.

Amid these changes, digital twins will be crucial for enhancing data centre efficiency and sustainability in both new and existing data centers. By simulating physical facility environments, digital twins allow operators to optimise power distribution, improve cooling techniques, and test energy changes, helping to maximise resource use and reduce stranded capacity. This technology not only makes the most out of existing space, it also supports sustainable growth, setting a new standard for energy-efficient, AI-capable data centres.

EU’s 2025 Energy Efficiency Directive Will Prompt Data Center Accountability – Mark Fenton, Product Engineering Director at Cadence

The EU Energy Efficiency Directive’s new reporting requirements, starting in May 2025, will mark a significant step in measuring energy and water usage within the data centre industry on a wide scale. By establishing initial data points, the directive will enable an ongoing comparison of industry performance, potentially paving the way for new regulations or targets that promote greater energy efficiency. The results will likely reveal a diverse landscape, where some companies, particularly within tech, show measurable progress, while others lag behind, exposing varying levels of commitment to environmental goals.

What’s more, with public interest in data centres’ resource use on the rise, these findings could invite heightened scrutiny, especially if data points to excessive energy consumption or local grid strain. This may lead to ‘naming and shaming’ by the media, heightening societal pushback even further.

However, the EU’s transparency-driven approach and heightened scrutiny should encourage data centres to adopt greener practices and utilise tools like digital twins, both to meet compliance standards and mitigate public backlash. Ideally, this will set a new benchmark for sustainability and accountability across the sector.

Gen Z’s Purposeful Mindset Will Help Close the Skills Gap – David King, Senior Principal Product Engineer, Cadence

From 2025 onwards, the data centre industry will see a substantial generational shift as seasoned professionals retire and younger, tech-savvy talent bring in specialised skills in AI, automation and sustainability.

Traditionally focused on physical infrastructure, data centre roles are evolving to require advanced technical skills like proficiency with simulation software such as digital twins. These tools are crucial in modern data centres for optimising energy use, airflow, and resource allocation, marking a proactive shift toward efficiency and sustainability.

Gen Z’s entry into the workforce also brings a strong focus on purpose-driven careers, where sustainability is a core value. This generation’s commitment will likely drive more talent to the data centre industry as they hope to help promote greener practices, such as implementing advanced cooling methods, reducing inefficient resource utilisation and bringing down stranded capacity.

As these professionals assume leadership roles, data centres are set to become more agile, eco-friendly and resilient. This evolution will not only close the current skills gap but redefine the industry’s approach to sustainable innovation, positioning it for a future led by Gen Z’s commitment to impactful, environmentally conscious operations.

Space Crunch and Cost Push Data Centres Back In-House – Aitor Zabalegui, Senior Principal Application Engineer at Cadence

Next year, data centres will continue to be challenged with intensified space constraints and rising operational costs, driven largely by surging AI workloads. For example, in Dallas-Ft. Worth, there is a record 372.2 MW of data centre space under construction, with 91.8% pre-leased. This forces all organisations, including cloud giants, to compete for limited real estate, pushing up costs. As a result, many organisations will reassess the benefits of moving certain operations back in-house and making use of pre-existing infrastructure. By doing so, these companies aim to mitigate the financial strain associated with using cloud providers or competing with large companies for colocation space.

Despite the in-house appeal, AI-led business demands may quickly outpace the limitations of organisations’ existing infrastructure, putting the need for new data centre builds back in focus. However, the significant time and complexity required for new builds – from permitting to power utility negotiations – will sustain a continued need for cloud resources in the short term.

The squeeze is on. In any case, both cloud providers and in-house data centres may increasingly turn to digital twin technology as a vital tool to maximise the current resources they have. This approach will be critical if the industry is to adapt to the rising demand created by technological innovation, including AI.

The Rise of AI Agents in Data Centres: Balancing Power, Efficiency and ROI – Mark Fenton, Product Engineering Director at Cadence

AI agents are poised to secure their position in workplaces industry-wide in 2025, and data centres are no exception. Software providers in this sector will increasingly explore the integration of powerful large language models (LLMs), such as those based on ChatGPT, into data centre tools. While this offers transformative potential, it also introduces significant challenges due to the resource-intensive nature of LLMs.

Through Cadence’s own testing of advanced LLMs, it’s become clear that even relatively modest setups require substantial infrastructure for training and operations, with each model functioning as a ‘personal assistant’ for operators. In order for an AI assistant to be well informed, they need significant training to understand the questions asked and provide the correct answers. As such, for assistants to be quick in providing those answers, businesses will need low latency and high-speed compute running the show.

Organisations will need to weigh the ROI of AI agents carefully. For example, at Cadence, integrating LLMs could enable digital twins to become personal assistants that provide on-the-spot insights for automating cooling and optimising energy efficiency. This synergy between AI and digital twins has the potential to redefine data centre operations, creating cost and energy efficiencies at a time of heightened environmental scrutiny. Ultimately, implementing LLMs across any platform will require strategic planning to ensure that their financial and environmental benefits outweigh the energy costs required to support them.

For more from Cadence, click here.