Huawei, ABB and VIRTUS Highlight Converging AI Data-Centre Power Trends

Huawei presented its grid-interactive AIDC solution at HUAWEI CONNECT 2026 in Shanghai, with the Huawei AIDC Facility Summit held on 20 September. The offering is designed to integrate AI data centres more closely with the electricity system through a multi-layer hybrid energy-storage system, electricity-compute synergy, liquid cooling and modular prefabrication.

The launch addresses the infrastructure requirements associated with increasingly dense AI workloads. S1 cites potential future chip designs requiring more than 1 MW per rack, although this is presented as a prospective requirement rather than a current deployment specification. In this context, grid interaction, storage and cooling architectures are becoming central design considerations for AI-oriented facilities.

ABB separately launched Infinitus, a DC solutions portfolio for powering AI data centres. ABB estimates that 25% to 40% of new data-centre capacity installed by 2030 could use DC distribution rather than conventional AC systems. This is an estimate by ABB, not a market-wide deployment forecast, but it indicates where power-system vendors see demand developing as AI loads reshape facility electrical design.

On the financing side, VIRTUS Data Centres closed a £2.45 billion (US$3.24 billion) financing package with a consortium of 13 banks. The funding is intended to support expansion in the UK and continental Europe while supporting existing operations. The source does not identify individual sites, capacity additions, project-level CAPEX allocations or new facilities entering service.

The broader demand backdrop remains material. According to S1's summary of IEA analysis, global data-centre electricity demand is expected to more than double by 2030. S1 also cites a World Economic Forum estimate that US$700 billion to US$1 trillion of investment is being lost to project delays and cancellations. The source frames early consideration of power, water and land availability in site selection and project design as a potential means of reducing such constraints.

Taken together, these developments suggest that AI data-centre growth is increasingly a power-infrastructure and delivery challenge, not solely a compute deployment question. Technology vendors are advancing architectures for high-density loads, while established operators continue to access large-scale financing; however, the ability to execute new projects may still depend on resource and grid constraints being resolved at an early stage.