I believe the discussion around DC and AC is bigger than one technology or one facility.
I recently joined the EUDCA Technical Committee Working Group on Emerging Technologies, which focuses on the future direction of the European data center sector. My first meeting with the group was this week and, sure enough, the topic was the DC future for data centers. The discussion was led by Mart Duitemeijer of VEIR, a company developing superconducting power delivery systems for high-density AI infrastructure.
That timing is telling.
Europe is at a critical point. It wants digital sovereignty. It wants AI capability. It wants industrial competitiveness. It also wants decarbonisation, efficient use of energy, better grid integration and public trust.
Those goals can either collide with each other or reinforce each other.
The difference will come down to policy, standards and execution.
This is where I believe EUDCA is particularly well positioned. The association gives operators, suppliers and national trade bodies a mechanism to engage with European policymakers before the rules are written in ways that unintentionally slow down useful innovation.
That matters because European policy is already moving in this direction. The European Commission’s roadmap for digitalisation and AI in energy points towards structured dialogue between data center operators, energy stakeholders and public authorities. That is exactly the kind of conversation the industry needs.
Regulation should not simply ask data centers to report more data or consume less energy. It should enable better infrastructure.
That means creating space for architectures that improve efficiency, support grid flexibility, safely integrate storage, and help Europe build AI capacity without undermining its energy-transition goals.
Europe also has an advantage that should not be overlooked: it already has deep expertise in renewable generation, power electronics, battery storage, grid integration and industrial energy systems.
That expertise is directly relevant to the DC discussion.
A good example is GreenYellow, an Ardian-controlled platform specialising in decentralised energy, solar PV, energy efficiency and battery storage. At Verne, we are actively working with GreenYellow on energy solutions for our recently announced Île-de-France campus, which was launched at Choose France.
That announcement received the sort of attention data centers do not usually receive. Dominic Ward, our CEO, posted afterwards that it is not every day President Emmanuel Macron tweets about your company.
That is not just a nice communications moment. It is a signal that digital infrastructure, energy infrastructure and industrial policy are now part of the same conversation.
GreenYellow’s experience with PV and BESS architectures, including the practical realities of high-voltage DC systems, is extremely valuable input for the data center industry.
The data center world has a tendency to assume that every new challenge is unique to the data center world. It usually is not.
The PV and storage industries have already wrestled with questions around 1,500 VDC systems, DC-coupled storage, protection equipment, maintenance safety and grid interaction. We should be learning from them, not rediscovering everything ourselves.
So where does that leave us?
Not with a winner.
Not with AC declared dead.
Not with DC declared inevitable.
It leaves us with work to do.
The data center industry needs to collaborate more seriously with the energy sector, the renewables sector, standards bodies, equipment manufacturers, policymakers and grid operators. We need to make sure that emerging regulation enables innovation rather than freezing yesterday’s architecture into tomorrow’s rules.
We need to develop safe, standardised and maintainable approaches to higher-voltage DC where it makes sense. We need to understand how AI loads behave during grid disturbances. We need to design facilities that are not only resilient for customers, but also responsible participants in the wider electricity system.
And we need to keep the purpose in mind.
The point of this debate is not to choose a fashionable voltage.
The point is to build infrastructure that can support the next generation of AI safely, efficiently and responsibly.
The winning architecture will not be AC or DC.
It will be the architecture that delivers the greatest amount of useful compute from every constrained watt, while remaining safe for operators, acceptable to communities, and constructive for the grid.
That is the real challenge.
And it is one Europe has a real opportunity to help solve.
To conclude this mini-series: the question is no longer whether AI will change data center infrastructure. It already has.
The challenge now is ensuring that power architecture, energy systems and policy frameworks evolve together. If they do, the industry has an opportunity to deliver more compute from every available watt while supporting wider economic and sustainability goals.
The answer may not be AC. It may not be DC. Most likely, it will be a combination of both. But whatever architecture emerges, success will depend on the industry's ability to make it safe, repeatable and practical at scale.