What impact are AI investments expected to have on network capacity demand in Australia and also in Asia?

The consensus is that these significant investments in AI-driven data centers will have a massive and accelerating impact on network capacity demand across Australia and Asia.

The fundamental reason is the nature of Artificial Intelligence (AI) workloads, especially the training and inference of Large Language Models (LLMs) and other advanced AI applications, which require:

1. Enormous Processing Power and Storage: This is the core driver of the data center capacity growth (measured in Gigawatts, or GW).
2. Low-Latency and High-Bandwidth Connections: This is the direct driver of network capacity demand, particularly in Data Center Interconnect (DCI).

Here is a breakdown of the expected impacts:

Australia

• Massive Capacity Growth and Supply Gap:
  • Australia’s live data center capacity is forecasted to increase significantly (e.g., from about 1.4 GW in 2025 to around 1.8 GW within three years, and the investable universe is expected to grow to circa A$46 \text{billion} by 2029).
  • Despite this growth, demand is outpacing supply, with an estimated supply gap of 0.7 to 1.7 GW projected by 2028, highlighting the intense pressure on infrastructure.
• Surge in High-Density Workloads:
  • The computational intensity of AI is pushing rack densities (power per rack) to extremely high levels, often rising from 30-40kW to over 80kW for modern AI deployments. This concentrated power requires highly robust internal and external network connections.
• Infrastructure Stress and New Hubs:
  • The demand for power and connectivity is straining existing infrastructure, with grid capacity constraints emerging in primary hubs like Sydney and Melbourne.
  • This is driving the expansion of data centers into secondary/emerging hubs like Melbourne (whose share is forecast to rise), Brisbane, and Perth, which in turn necessitates new and upgraded high-capacity network links connecting these sites both nationally and internationally.
• Strategic Hub for Asia-Pacific: Australia’s stability, regulatory environment, and access to subsea cables position it as an ideal launchpad for AI workloads across the Indo-Pacific, which will further increase international network traffic flowing through the country.¹⁴

Asia-Pacific (APAC)

• Exponential Network Traffic Growth: Global data center experts predict at least a 6X increase in Data Center Interconnect (DCI) bandwidth demand over the next five years, with AI workloads believed to place the biggest demand on DCI.
• Dominant Driver of Global Growth: AI adoption is driving massive capacity additions across the region.
  • Generative AI is expected to account for a significant portion of the increase in data demand.
  • The Asia-Pacific region is predicted to double its data center capacity within the next five years, with an additional 2 GW coming online each year, driven by countries like Japan, South Korea, Indonesia, Malaysia, and Australia.
• Decentralisation and Edge Computing: As AI workloads like training Large Language Models (LLMs) become more distributed across different facilities, and inference workloads are deployed closer to users at the “edge” to reduce latency, the need for Managed Optical Fiber Networks (MOFN) and high-capacity fiber connectivity between these distributed centers will accelerate dramatically.
• New Regional Hotspots: High demand and limited resources in traditional hubs (like Singapore and Hong Kong) are pushing investment into secondary markets in Southeast Asia (e.g., Indonesia and Malaysia, which have projected capacity growth rates between 32% and 56% CAGR), leading to new subsea cable and terrestrial network routes being developed to support this infrastructure.
• Bandwidth Intensity: To cope with the data demands, 87% of data centre experts anticipate needing a fibre optic capacity of 800 Gb/s or higher per wavelength for DCI.

In summary, the AI investment boom is creating a virtuous cycle of massive capacity expansion, exponential traffic growth, and an urgent need for network upgrades that support ultra-high density and low-latency connections both within and between data center hubs across Australia and the broader Asia-Pacific region.²⁷