AI Driving a 6-Fold Spike in Data Centre Demand: Explained
It’s no secret that the surge in AI applications is reshaping infrastructure requirements across the data centre industry.
As enterprises accelerate deployment of Gen AI systems and chatbots, AI will trigger at least a six-fold increase in data centre demand by 2027, according to research from networking equipment provider Ciena.
Ciena's survey of data centre professionals has found that 53% identify AI workloads as the primary driver of data centre interconnect demand – which are high-speed network connections that link separate data centre facilities – over the next two to three years.
Meanwhile, AI use cases are now expected to surpass cloud computing workloads, which account for 51% of demand and big data analytics at 44 %.
This is a big change in how data centres allocate resources and plan capacity.
“The rapid expansion of AI is reshaping the data centre networking infrastructure landscape,” says Jürgen Hatheier, Ciena's International CTO.
“Hyperscalers and data centre providers must be able to scale their networks to meet these performance requirements, carving out a clear path to an AI-driven future.”
Ciena’s research reveals distributed AI training requirements
The complexity of AI model training is forcing operators to rethink their infrastructure strategies.
Ciena's research indicates that 81% of respondents expect large language model (LLM) training to occur across distributed data centre facilities rather than single locations.
This distributed approach requires robust interconnections between facilities.
On average, data centre professionals expect 43% of new facilities planned by their companies will be dedicated to AI workloads.
“The amount of training required to build trustworthy, high-quality AI is prompting network operators and hyperscalers to investigate how their infrastructure will evolve to meet demand,” Jürgen explains.
The survey also identified key factors shaping where AI inference will be deployed.
- AI resource utilisation over time is the top priority (63%)
- Reducing latency by placing inference compute closer to users at the edge (56%)
- Data sovereignty requirements (54%)
- Offering strategic locations for key customers (54%)
“This means we can expect the AI ecosystem of tomorrow to be a network of interconnected data centres, all with unique roles to play,” Jürgen says.
“Edge data centres, for instance, will handle inferencing as well as offer strategic locations to improve performance of latency-sensitive applications, such as security applications using facial recognition.”
Power demands raising sustainability questions for operators
The International Energy Agency predicts that data centre electricity consumption will more than double globally between 2022 and 2026, driven primarily by AI applications.
As a result, network performance requirements are escalating in response to AI demands.
Ciena's survey found that 87% of data centre professionals believe they will need fibre-optic capacity of 800 gigabits per second per wavelength or higher for data centre interconnects – which is a substantial increase from current typical capacities.
Operators are also focusing on solutions that balance performance with efficiency.
High-capacity pluggable optics, which are modular components that can be inserted into network equipment to provide connectivity, offer one approach to managing power consumption whilst maintaining network performance.
“These concerns underscore the importance of solutions that help minimise power consumption and physical footprint, such as high-capacity pluggable optics,” Jürgen says.
“The International Energy Agency predicted that data centre electricity demand will more than double globally from 2022 to 2026 because of traffic driven by AI applications – and it's a subject that's being taken seriously by the entire industry.”
Custom strategies emerging as operators adapt architectures
Data centre operators are developing tailored approaches to meet AI infrastructure demands rather than adopting standardised solutions – and the diversity of AI applications and customer requirements is driving operators to create custom network strategies.
- Interconnectivity, efficiency and sustainability are some of the most critical areas that will need attention
Now, cloud providers and data centre operators must balance multiple considerations including performance, cost, sustainability and regulatory compliance.
Different business models require different architectural approaches, with some operators focusing on centralised facilities whilst others pursue distributed edge strategies.
Network architecture decisions will vary based on specific customer needs and geographical requirements.
Furthermore, operators serving financial services clients may prioritise ultra-low latency connections, whilst those supporting AI training workloads may focus on high-bandwidth interconnects between facilities.
“How data centre operators tackle these challenges will vary. Ultimately, cloud providers and data centre operators will need to adopt custom network strategies tailored to their specific business needs and customers,” Jürgen explains.
“Over the coming months and years, we will see different network architectures and expansion strategies that fit specific business models and which best cater to their customers.
“Regardless of the approach taken, the key will be high performance DCI connectivity. Operators must ensure their overall network infrastructure is ready for an AI-centric future – and if the research is anything to go by, they're already considering the architecture necessary to make the mass adoption of AI a success.”
Explore the latest edition of AI Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.
Discover all our upcoming events and secure your tickets today.
Also sign up to our free weekly newsletter for the latest insights and stories straight into your inbox.
AI Magazine is a BizClik brand