Managing the energy demands of the AI revolution

A data centre is a physical facility designed to house computing infrastructure such as servers, storage systems, and networking equipment, enabling the processing, storage, and distribution of digital data (Refer to figure 1).

- 

Introduction

These facilities are critical for delivering services like cloud computing, artificial intelligence (AI), online transactions, streaming, and enterprise IT operations. Modern data centres operate 24/7 and require high reliability (99.99 percent uptime). They integrate power systems, cooling systems and redundancy mechanisms.

Types of data centres

Hyperscale data centres: These are large-scale facilities operated by tech giants such as Google, Amazon Web Services, and Microsoft.

Which comes with the features like: Massive scale (100,000+ servers), designed for cloud computing and AI workloads, and highly optimised for efficiency (low PUE). Hyperscale data centres account for 40-50 percent of global data centre capacity.

Example: Google's global data centre network supporting Search, YouTube and AI services.

Co-location data centres: These are third-party facilities where businesses rent space for their servers and IT infrastructure. Features of these include shared infrastructure (power, cooling, security), which are cost-effective for small/medium enterprises with flexible scalability.

Enterprise data centres: Owned and operated by individual companies for internal use, these feature customised infrastructure, limited scalability compared to hyperscale, and gradually decline due to cloud adoption.

Working of data centres

Core components: Data centres are built around three fundamental computational pillars: compute (servers), storage and networking. These components collectively enable data processing, storage and transmission. (Refer to Figure 2)