AI: Transforming the data centre Real Estate game
One of the key factors affecting data centre location is latency, with low latency being crucial for applications like high-frequency trading and gaming, or where data access is needed in high-risk, safety-critical contexts. While low latency is essential for certain sectors, many AI models are less sensitive to it, allowing data centres supporting AI to be located away from major urban centres at a potential lower cost base.
Currently, access to energy, whether renewable or not, is one of the main challenges to the data centre sector and this is no different for AI-enabled data centres. The demand for power is expected to increase dramatically and despite advancements in energy efficiency, the development of infrastructure/electrical grids, and an increased scope of renewable power networks, supply currently lags behind the rising demand.
The following article highlights the rapid growth in European data centre take-up, with a 65% increase in the first half of 2023 compared to the same period the previous year. The emergence of generative AI, with its high computer power requirements, is driving a fundamental change in data centre design. Graphics processing unit (GPU) chips, rather than the traditional central processing unit (CPU) chips, are needed to support AI's high density and performance requirements, leading to a surge in demand for AI chips, as seen in Nvidia's recent Q2 earnings with a 171% year-on-year jump in data centre revenue.
With many of the recommendations raised in the recent Independent Review of The Future of Compute pointing to the need for a step-change in the infrastructure that will be needed over the coming decade, and increasing broader concerns from industry and end users around security, scalability and reliability, AI is set to significantly transform the data centre landscape in the UK and beyond.
Link to article