Evaluation Consultant’s Reflection on the Datacloud Global Congress 2025

So, another year at Data cloud in Cannes and time to reflecton this. For Evaluation Consultants our attendance is not just about the conference, it is also about maintaining existing connections and generating new connections as well a chance to discuss over food and drink what challenges and opportunities we are all encountering now and in the future within the data centre sector.

What came out of these conversations was no surprise, but it is was how the issues were being approached and overcome that was interesting.

From our perspective, the following topics were key discussion points but are not listed in any particular order of importance or number of occasions they were discussed:

·     Demand for DCs

·     The impact of AI and HPC on DC design

·     Shortage of adequate resources across consultants and GCs

·     New build vs Refit of existing DCs

·     Over ambitious project delivery schedules

·     Chinese AI advancements and their impact on the ‘West’
‍

The key points from these discussions for each of these topics above are summarised below:

‍

Demand for DCs

Demand for data centres was very common point of discussion, which remains high across the globe with Artificial Intelligence (AI) and High-Performance Computing (HPC) headlining the surge in demand, to the point where some large Hypers calers (arguably nowMegaScalers) had cancelled some data centre projects and laid off staff to freeup capital and focus efforts on AI research and deployment, reflecting a strategic pivot to prioritize cutting-edge technology over traditional infrastructure expansion.

But the usual challenges of optimal location, power availability, planning and permitting restrictions are proving to be increasingly problematic to overcome in certain Tier I & II markets coupled with ever increasing environmental and sustainability targets that must also be achieved.

Geopolitical tensions, data sovereignty, trade disputes and regulations requiring local data storage are also dictating where data centres are located. Countries are investing in domestic infrastructure to comply with sovereignty laws, while international expansion faces regulatory challenges.

So, enter the Tier III markets, which are the other regions previously not focused upon but being viewed as credible alternatives for developing data centres where power, planning and permitting may be less restrictive to development, although questionable whether more than 20MW of IT power at best per site could be made available.

There were discussions around DC sites being developed with limited grid power capacity in the short term bolstered by gas powered generators or other power sources such as solar, wind and even nuclear as the primary power source alongside what grid power is available pending further grid power capacity becoming available in future years.

Demand for data centres was very common point of discussion, which remains high across the globe with Artificial Intelligence (AI) and High-Performance Computing (HPC) headlining the surge in demand, to the point where some large Hypers calers (arguably nowMegaScalers) had cancelled some data centre projects and laid off staff to freeup capital and focus efforts on AI research and deployment, reflecting a strategic pivot to prioritize cutting-edge technology over traditional infrastructure expansion.

But the usual challenges of optimal location, power availability, planning and permitting restrictions are proving to be increasingly problematic to overcome in certain Tier I & II markets coupled with ever increasing environmental and sustainability targets that must also be achieved.

Geopolitical tensions, data sovereignty, trade disputes and regulations requiring local data storage are also dictating where data centres are located. Countries are investing in domestic infrastructure to comply with sovereignty laws, while international expansion faces regulatory challenges.

So, enter the Tier III markets, which are the other regions previously not focused upon but being viewed as credible alternatives for developing data centres where power, planning and permitting may be less restrictive to development, although questionable whether more than 20MW of IT power at best per site could be made available.

There were discussions around DC sites being developed with limited grid power capacity in the short term bolstered by gas powered generators or other power sources such as solar, wind and even nuclear as the primary power source alongside what grid power is available pending further grid power capacity becoming available in future years.

‍

The impact of AI and HPC on DC design

AI, in particular, was a key theme for discussion with the continual rise of AI, particularly generative models and machine learning, requiring immense computational power beyond traditional CPUs and so GPUs andTPUs are used instead to handle these workloads. This has necessitated architectural, structural, mechanical, and electrical redesign to support high-performance, power-hungry hardware. It has been forecast that by 2030,AI-driven data centres are expected to represent a significant share of global demand, pushing operators to upgrade their existing data centres as well as building to specifically facilitate AI and HPC computing.

Higher cooling demands driven by AIand HPC deployments are driving the need for liquid cooling. This in turn is requiring some tweaking of ‘standard’ DC designs to accommodate heavier equipment and distribution loads in the data halls as well as alternative cooling distribution arrangements direct to the racks. The higher power densities are also leading to smaller building footprints, mainly relating to data hall area requirements.

There were discussions centred around liquid cooling deployments having no standard model design as yet, due to this being are latively new technology and also the variety of approaches to liquid cooling currently on the market based around Direct liquid cooling (or direct to chip) and Immersion cooling.

‍

Shortage of adequate resources across consultants and GCs

‍

A very common discussion topic was the volume of projects currently and in the pipeline, are continuing to put a strain on the resources of both consultants and general contractors across the sector and as a consequence those consultants and general contractors not typically associated with delivering data centres are entering the market with varying degrees of success. These limitations of success are influenced by a lack of track record and expertise in the data centre market which has left clients less likely to engage with these GCs.

There is also the growing issue of clients hiring project, design and cost management consultants to work directly for them and this is draining the sector of very experienced data centre expertise who would have otherwise directly delivered data centre projects for these clients.

‍

New build vs Refit of existing DCs

The need to review older legacy data centres to consider whether they can be upgraded to meet current and future expected standards or are at the point of obsolescence was an interesting topic discussed as there is a growing quantity of data centres that have been in operation for 10 or more years which need to be evaluated to assess their potential for ongoing usage in their current form. Many of these buildings were designed around delivering a lower power density and so would not necessarily lend themselves to compliance with current and future expected standards of design without extensive re-design, especially where AI and HPC deployments are being considered.

Some of these legacy data centres could be ripe for re-classification as ‘Edge’ Data centres where they do fulfil the criteria necessary to service the technology that requires low latency between the source and the end-user.

Environmental sustainability would also have an influence on the upgrade potential to enable replacement plant and equipment that is more energy efficient and therefore achieving both environmental and operational cost benefits.

‍

Over ambitious project delivery schedules

This topic was popular and frankly has been a common point of discussion over many past years but the frustration in more recent times is the ever-increasing number of projects which are squeezing more time out of schedules that is not there to be squeezed out. With so many moving parts involved in delivering a data centre project from concept to handover, there is the risk and inevitability of one or more activities not going to plan. Not allowing sufficient and realistic durations with some float included to achieve RFS dates is setting up a project to fail from the outset and when delays either during the pre-construction or post construction stages are encountered there is often limited scope to claw back that time due to external or stakeholder pressures.

We all understand and appreciate the pressures that our data centre clients are exposed to in order to satisfy their stakeholders and customers, as well as generally satisfying the exponential and immediate demand for data centre space, however this approach cannot be sustainable as the outcome in terms of lower quality and higher cost is generally being and will continue to be casualties in the endeavours to achieve schedule. This imposes greater pressures across the supply chain to deliver to ever decreasing project timescales which are very much driven by main plant and equipment that sit on the critical path and their lead in periods are not getting any shorter.

Chinese AI advancements and their impact on the West

‍

Chinese investment and their rapid development in the AI space was mentioned by more well-informed individuals on such matters. The riseof Chinese AI models such as DeepSeek's R1, Alibaba's QwQ-32B, and Tencent's Yuanbao, has disrupted the Western data centre sector, driving an increased focus on innovation, competition, and market adaptation.

The ‘Wests’ initial market reaction to Chinese AI breakthroughs was sharp—tech stocks like Nvidia’s dropped as investors feared aloss of market edge and share. However, Western companies have since adapted through:

·     Investment Shifts: Some funds are moving from large-scale data centre builds to optimizing existing infrastructure and developing efficient, smaller AI models that require less hardware.

·     Edge Computing Focus: Emphasis on de centralized facilities has grown to improve cost and latency, aligning with AI’s real-time needs.

·     Market Pressure: Lower-cost Chinese models have squeezed profit margins, forcing price reductions and spurring R&D investment to regain a competitive edge.

‍

The geopolitical and economic fallout due to the rise ofChinese AI has intensified data security concerns and scrutiny of foreign techin Western markets. Economically, it has become a double-edged sword:competition drives innovation, but it has also challenged profitability asWestern firms adjust to a more cost-conscious landscape.

 These key topics and many other key issues raised that weare facing in the Data Centre sector enriched and enhanced our team’s knowledgeand experience developed through our own involvement in data centres andoffered some different perspectives and approaches that we could learn from andas well as appreciate that we all exposed to the same challenges andopportunities.

We are looking forward to further engagement on what the hottopics happen to be at the Datacloud Global event next year in 2026.

‍

‍

‍