For most of the data centre industry’s history, the substation was infrastructure that existed before the project began. It was delivered by the utility, managed by the Distribution Network Operator (DNO), and treated as an external dependency rather than an internal asset. The project boundary started at the LV switchboard. Everything upstream was someone else’s responsibility.
That model is breaking down. AI infrastructure’s power demands, grid connection timelines, and the commercial imperative to control energy strategy are driving a fundamental shift in substation ownership. Operators are increasingly procuring, building, and in some cases operating their own high-voltage infrastructure. And with that ownership comes a set of assurance responsibilities that the industry is still learning to manage.
Grid Timelines Have Forced the Conversation
The shift toward operator-owned substations is not primarily driven by preference. It is driven by constraint. Grid connection timelines in many European markets now average seven to ten years, with some extending further. In parts of North America, large-scale connections can face delays of seven to ten years, with some delays of up to 13. Those wait times are fundamentally incompatible with the speed at which the industry needs to move. When an operator needs to secure access to hundreds of megawatts of power within eighteen months and the grid timeline extends years beyond that, the response has been to bring the power infrastructure within the project boundary.
That response takes various forms: private substations, behind-the-meter generation, battery energy storage systems, and in some cases hybrid configurations that combine grid, generation and storage assets on a single campus. Each of these approaches moves high-voltage infrastructure from external dependency to internal asset. And each requires assurance capability that the traditional commissioning model was not designed to address.
Ownership Without Assurance Is Liability
Choosing to build a substation is not the same as understanding how it works. Many data centre operators and their delivery partners have deep expertise in IT infrastructure, mechanical systems and low-voltage distribution. But high-voltage (HV) power systems sit outside that traditional competence. Protection coordination, grid interface compliance, transformer commissioning, switchgear verification and safety management systems all require specific technical knowledge and rigorous assurance processes.
When operators take control of HV assets without a corresponding assurance framework, they are also taking on significant risk. Incorrectly commissioned protection schemes, inadequately verified switchgear, or poorly coordinated grid interfaces can result in catastrophic failure. Unlike an internal systems failure, the consequences (already unthinkable in a world of guaranteed uptime) extend beyond the data centre to the grid itself. Regulators and network operators are acutely aware of this risk, which is why oversight requirements for private HV infrastructure are tightening across multiple jurisdictions.
Governments Are Now Actively Encouraging This Shift Towards Operator-Owned HV Infrastructure
What began as an operator-led response to grid constraints is now being reinforced by government policy.
In the United States, the Trump administration has explicitly told major technology companies to build their own power generation rather than rely on an ageing grid. In his February 2026 State of the Union address, the President framed this as a ratepayer protection measure, directing data centre developers to secure or develop a dedicated power supply for their own loads. The sector is already embracing this new paradigm, with research from Cleanview identifying 46 US data centre projects representing 56 GW of planned behind-the-meter generation, with 90%of those projects announced in 2025 alone. At the federal level, FERC’s December 2025 ruling approved colocation agreements allowing data centres to connect directly to power plants, bypassing the transmission grid entirely.
In the UK, the government’s AI Growth Zones programme is taking a different but equally significant approach. The November 2025 policy paper commits to working with Ofgem to enable data centre developers to build their own high-voltage grid infrastructure rather than waiting for network operators to deliver connections. Four zones have already been designated, including Blyth and Cobalt Park in the North East of England. The government estimates that these interventions could reduce time to power for data centre projects built within these zones by as much as five years.
These are not future proposals. They are current policy, and they carry a clear implication: if operators are being encouraged, and in some cases directed, to own their power infrastructure, the assurance frameworks supporting this new kind of infrastructure buildout must be in place before the first transformer is energised.
The Insurance Market Is Paying Attention
The insurance industry’s approach to data centre risk is evolving in parallel with infrastructure complexity. As operators take ownership of generation and HV distribution assets, insurers are asking more detailed questions about commissioning quality, protection philosophy, maintenance strategy and operational competence. A data centre with a private substation carries a different risk profile from one connected to a utility-managed grid supply, and that risk profile comes with a new, justifiable increase in scrutiny.
Underwriters are recognising that the evidence supporting substation commissioning — the protection study verification, the switchgear test records, the grid compliance documentation — must be as rigorous as anything else in the commissioning process for a new facility. Commissioning evidence that stops at the LV switchboard is no longer sufficient to satisfy the risk assessment requirements of major insurance programmes.
Bringing HV power infrastructure inside the project boundaries is a decision made to smooth over pain points. If the commissioning of those HV assets is ineffectual, however, operators risk creating new headaches worse than the old ones.
HV Assurance Requires a Different Discipline
High-voltage commissioning is a distinct discipline with its own standards, safety protocols, testing methodologies and competence requirements. It is not, as some operators may be tempted to believe, an extension of LV commissioning. The equipment is different, the failure consequences are greater, the regulatory framework is more complex, and the people who do the work require a separate subset of qualifications and experience.
That distinction is important because some commissioning providers attempt to cover HV scope with teams whose primary experience lies in LV and mechanical systems. That approach may satisfy a contractual line item, but it does not deliver the kind of verification and assurance that HV infrastructure demands. Protection testing, transformer commissioning, and grid interface verification require engineers who work in that environment regularly, understand the failure modes, and can interpret results with confidence.
Assurance Must Start Before the Substation Is Built
Effective HV assurance does not begin at the point of commissioning. It begins at the design stage, when protection philosophy is defined, resilience requirements are established, and the interface between the private HV infrastructure and the distribution network is agreed. If a commissioning perspective is absent from those early decisions, the assurance framework inherits constraints that are difficult and expensive to resolve later.
On programmes where Global Commissioning is engaged from early design through to energisation, the protection strategy, resilience modelling and grid interface risk are considered alongside commissioning delivery from the outset. That early involvement means the assurance framework is shaped by (and shapes) the infrastructure rather than applied to it after the fact.
A Structural Shift, Not a Trend
Operator ownership of HV infrastructure is not a temporary response to grid constraints. It reflects a permanent change in how AI-scale data centres are powered. If the data centre industry is to meet the power and capacity demands of the AI boom, then it must take ownership of, and responsibility for, its own HV infrastructure. Most importantly, it must do so responsibly with assurance and certification at heart of its approach, not treated as an afterthought.
As campuses incorporate generation, storage and transmission alongside computing, the operational boundary of the data centre has expanded to include assets that were previously external to the project.
Assurance models must expand alongside project scope and remit. The businesses that can provide integrated assurance across the full energy chain, from grid interface through high-voltage distribution to the final point of utilisation, will be better positioned to support the infrastructure programmes that are now being planned and delivered. The businesses that remain configured for downstream verification will find that the scope they can credibly address is narrowing. That boundary has already moved. The question is whether the assurance model has moved with it.
