Flexible Grid Links Can Speed Up AI Data Centre Power Access, Says Report

AI data centres can cut their power connection wait times from five to ten years down to two to four years by using flexible grid connections along with their own power generation and storage systems. A report by Boston Consulting Group states that this flexible access lowers the need for major grid upgrades.

Market Impact

The study looks at five Northwest European markets: the UK, Germany, France, Belgium and the Netherlands, covering the investment period from 2030 to 2035. Global data centre power demand is expected to reach about 210 gigawatts by 2030, which is up from 84 gigawatts in 2025. Power availability has turned into a major roadblock for data centre growth, leaving connection queues stretching for years in multiple markets.

Operations And Costs

Under this method, data centres take non-firm electricity connections where power supply can drop during grid congestion, avoiding long waits for guaranteed links. They can use on-site power generators, batteries and flexible workloads to keep running during those times. Connecting three years earlier could bring an estimated USD 1.28 billion in extra profits for a modeled 150-megawatt AI inference data centre in Northwest Europe, before counting on-site power asset costs. At the same time, an off-grid setup could raise yearly power supply costs by 70 to 230 per cent compared to a firm grid connection.

Recommendations

The report asks governments to build national data centre plans, push projects to spots with open grid capacity, and speed up approvals for on-site power assets. It also urges grid operators to make flexible connections a standard choice instead of a rare exception.

Context

Rapid growth in artificial intelligence workloads requires vast amounts of electricity, putting heavy strain on existing public power grids. Finding alternative energy setups helps technology companies bypass long infrastructure delays while keeping local power networks stable.