As we look back at 2025, it is tempting to describe this as a year of contrasts. Momentum slowed, then returned. Certainty gave way to questions, before clarity began to emerge again. But in reality, 2025 was not a break in the data centre growth story. It was a recalibration.
Demand continued to rise at a remarkable pace, driven by cloud and accelerated AI adoption. Yet the ability to deliver new power capacity struggled to keep up.
As a consequence, the market slowed down and nearly paused this year. Not because demand disappeared, but because the system around it needed time to catch up. Power availability, permitting, grid constraints, and the pace of AI-driven infrastructure change all converged at once. That pause gave the industry something it rarely has: space to reflect, reassess, and redesign how growth should happen next.
For AVK, this period reinforced a simple truth: the companies that continue to move forward are not those chasing volume alone, but those building resilience, integration, and long-term capacity into their operating models.
Despite the early-2025 slowdown, AVK continued to grow. Activity increased across regions, projects progressed, and new opportunities emerged even as others paused. This was not accidental. It reflected deliberate choices made years earlier: to invest in skills, to own the full power chain, and to design solutions that remain relevant even when the market becomes more complex.
Throughout the year, we strengthened our internal Academy, expanded our engineering and delivery teams, and advanced our integrated model covering generation, microgrids, storage, controls, modular PowerPods, and long-term service. We also introduced new capabilities, including emission reduction services and early carbon-capture deployments, helping operators think differently about emissions, reuse, and future regulatory pathways.
By the end of the year, AVK had delivered several projects, secured significant jobs across standby, almost commissioned microgrids, and continued to diversify its leadership team. Growth, in 2025, was not about speed alone. It was about depth.
The slowdown itself was both visible and measurable. After the aggressive AI-driven expansion of 2023 and 2024, early 2025 saw operators reassess pipelines, investors slow decision-making, and utilities pause new connections in several regions.
In Europe, permitting delays and grid bottlenecks became more pronounced. Yet this was never a contraction. Projects were delayed, not cancelled. Demand did not disappear; it queued.
Several forces converged at once. AI infrastructure itself began to change, requiring redesigns that slowed deployment timelines. National and regional regulations diverged, creating uncertainty for long-term commitments. Land availability tightened in key hubs. Above all, power and grid access emerged as the single largest barrier to growth.
This was not a failure of demand forecasting. It was a systems constraint. And it highlighted a fundamental shift: access to power is no longer a secondary consideration for data centres.
It is the primary one.
By the second half of the year, the picture began to change again. Deferred projects moved forward. Hyperscale demand resumed. AI workloads continued to scale, reigniting the need for large, reliable blocks of power across Europe.
At AVK, this shift was tangible. Projects moved into active negotiation, with decisions compressed into weeks rather than quarters. It turns out, the pause had concentrated demand rather than weakened it.
As growth returned, it came with new expectations. Traditional models built around grid dependency and standby-only systems no longer felt sufficient. Operators began asking different questions: how to reduce reliance on constrained grids, how to design hybrid systems that can flex with AI workloads, and how to secure independent dispatchable capacity without waiting years for connections.
Microgrids moved from theory to priority. On-site generation, hybrid architectures, and grid-interactive systems were no longer niche discussions. They became central to how new data centres are planned.
The growth cycle resumed, but the rules had changed.
Perhaps the most important shift of 2025 was how data centres began to be perceived within the wider energy system. No longer just consumers of power, they are increasingly seen as potential contributors.
By combining standby generation, flexible grid connections, microgrids, PPAs, and storage, data centres can become dispatchable assets. They can support the grid during peak demand, absorb renewable variability, and add resilience rather than strain.
This transition changes the narrative. Data centres move from being framed as villains of the energy transition to becoming part of its solution. Dispatchable power is not just a technical concept; it is a new role for digital infrastructure in a constrained, decarbonising energy system.
2025 reminded us that growth is not linear, and that resilience matters more than speed. The challenges the industry faced this year are not temporary. But neither are the solutions.
As we move into 2026, the focus will be on systems that can adapt, integrate, and evolve. Dispatchable power, microgrids, and long-term capacity planning will define the next phase of data centre growth.
At AVK, that is exactly where we are building.