NVIDIA has introduced a new modular system, the DSX POD, which aims to reduce the time required to set up AI data centres at prepared locations from several months to 14 days. According to The Hindu, the system is designed to help operators activate high-cost artificial intelligence infrastructure more rapidly. The core mechanism involves moving complex assembly and testing processes out of the data centre site and into factory environments. This approach allows for faster on-site installation, addressing the need for quicker scalability in the AI sector.
This development occurs against a backdrop of massive investment in data centre infrastructure. The International Energy Agency (IEA) reported that capital spending by the five largest technology firms surpassed USD 400 billion in 2025. The agency projects this figure will rise by another 75% in 2026. Such financial growth has driven a dramatic acceleration in project pipelines. The IEA noted that the capacity of dedicated AI factories has more than tripled over the previous 18 months. This rapid expansion is placing significant strain on planning and regulatory systems, creating a demand for more efficient deployment methods like those offered by NVIDIA.
Energy consumption remains a central challenge in this expansion. The IEA stated that power usage from data centres jumped by 17% in 2025, with AI-focused facilities growing even faster. Although efficiency per AI task is improving, the rise of energy-intensive applications such as AI agents is driving overall consumption upward. The agency projects that global electricity use for data centres will double by 2030, reaching approximately 945 terawatt hours. This growth is outpacing the broader increase in global electricity demand, which is projected to expand at a much slower rate.
To manage these energy demands, the tech sector is adopting new power sourcing strategies. The IEA reported that the technology sector signed around 40% of all corporate renewable energy agreements in 2025. Additionally, the pipeline of conditional offtake deals between data centre operators and small modular reactor nuclear projects has expanded from 25 gigawatts at the end of 2024 to 45 gigawatts. These developments suggest that the momentum behind AI is accelerating the commercialisation of new energy technologies, including nuclear and advanced geothermal sources, to support the physical infrastructure required for AI processing.