Familiar issues, but higher stakes: why 2026 will be a year of challenge and opportunity for European data centers

Demand for data center capacity in Europe will shift up another gear in 2026 due to a combination of ongoing pressures and new trends. With the expected breakout of AI inference, the emergence of agentic AI and ongoing demand for general cloud computing, the data center industry is likely to grow at its fastest rate yet. 

Data center operators who have focused strategically on higher rack densities, power availability and advanced cooling are best placed to take advantage of this wave of demand. As we look forward to 2026, I wanted to share some of our thoughts on how and to where the market will shift, what that means and what it will take to stay ahead.  

Power becomes Europe’s competitive battleground in 2026 

Access to reliable and scalable power will determine which data center operators grow and which are left behind. 

Power will continue to reshape the competitive landscape, with two significant forces impacting the market. First, the scale of data center requirements is increasing at an unprecedented rate. Demand no longer comes in the form of a few megawatts, as individual deployments can take the form of tens or hundreds of megawatts. 

Second, Europe is beginning to experience the growth surge experienced in the US over the last couple of years. As the power supply in the US tightens, customers are looking to increase their global footprint. 

With escalating density requirements, grid constraints and growing competition for power, a significant beneficiary of demand will be secondary markets, including the Nordics, which are expected to grow at double the rate of core markets in the next few years. 

AI scale opens a game of infrastructure Tetris 

The scale of infrastructure supply and customer demand will define where and how AI workloads are deployed. 

The global data center industry operational footprint is currently estimated to be an already immense 50 GW, growing to over 100 GW or potentially even 200 GW by 2030. One AI company alone, OpenAI, has committed $1.4 trillion to deliver 30 GW of compute during the next eight years, with its stated goal to be deploying a gigawatt every week in the future!  

With this kind of growth, we will see data center operators having to take creative steps towards meeting demand, while end users scour the market for capacity. We have already seen operators initiating the development of 1-2 GW campuses. Expect more of this. 

On a macro scale, it will become a giant game of Tetris with enormous blocks of AI infrastructure dropping into place across the map wherever power and capacity is the right size and shape. 

Inference and agentic AI go mainstream 

Data center operators equipped to handle the density and advanced cooling requirements of scalable AI infrastructure will lead the charge. 

What we are experiencing today began four years ago: models like OpenAI were developed and trained around the first major generation of GPUs for large language models. As NVIDIA's GPUs have seen dramatic changes in performance, parallelism and efficiency to achieve unprecedented levels of scale-up and scale-out, AI has started to move from training to inference.  

The rapid growth in training will continue, but with the massive increases in computational capability, enormous infrastructure builds and the accelerating capability of AI models, the next wave of growth is going to be driven by inference and agentic computation, which will accelerate rapidly in 2026. 

Neoclouds and hyperscalers expand NVIDIA’s strategic footprint 

NVIDIA’s influence beyond chip sales will continue to shape the distribution of compute capacity across Europe. 

Hyperscalers such as Google, with its own TPU chip architecture, will begin to impact the market this year, but expect NVIDIA to continue to dominate the market for AI silicon.  

NVIDIA's influence comes not simply through selling chips. It is investing across the data center ecosystem and in some of the data center neoclouds. The neocloud companies backed by NVIDIA accounted for roughly half of the European data center capacity signed by third parties during 2025. 

Expect the market share of neoclouds to continue to increase during 2026 as their credibility solidifies. At the same time, expect some consolidation with a handful of neoclouds emerging as dominant players. And, while much of the activity during the last 12 months focused on a small number of AI end users, we will likely see more hyperscale activity from the likes of Microsoft, Meta, AWS and Google. 

Liquid cooling hits critical mass 

Operators investing in the engineering and operational skills for liquid cooling will be best positioned for the coming wave of AI. 

2026 will mark the beginning of widespread adoption of liquid cooling as a new generation of NVIDIA GPUs is adopted by the market. 

After years of designing and engineering for liquid-enabled infrastructure, Verne is now experiencing the arrival of liquid-cooled hardware, such as our NVIDIA GB300 installation with Nscale. GB300 GPUs peak at 120 kW per rack, well beyond the limits of air-cooled infrastructure. As successive generations of NVIDIA processors arrive, liquid cooling’s adoption will accelerate. 

The shift to liquid will mean a change of infrastructure design and skills to service and operate these mission critical environments. Design, engineering and operations teams will need to work with liquid dynamics rather than airflow, a capability and skillset that only a select group of operators possesses today.  

Mastering this new frontier will delineate those able to capitalise on liquid cooling from those struggling with its complexities. 

The Verne advantage 

Europe’s data center landscape will be transformed in the years ahead, with 2026 set to be a pivotal period in time. Those who have invested in forward-looking designs and have made strategic choices on location, power availability and delivery are best positioned to navigate these changes. Future-ready AI data center providers positioned to succeed will be those prepared not just to meet these requirements, but to shape them. To discover more about the Verne advantage, contact us here.