Data centres are the engines of the internet. These large, high-security facilities host racks of servers that store and process our digital data, 24 hours a day, seven days a week.

There are already more than 250 data centres across Australia. But there are set to be more, as the federal government’s plans for digital infrastructure expansion gains traction. We recently saw tech giant Amazon’s recent pledge to invest an additional A$20 billion in new data centres across Sydney and Melbourne, alongside the development of three solar farms in Victoria and Queensland to help power them.

The New South Wales government also recently launched a new authority to fast-track approvals for major infrastructure projects.

These developments will help cater to the surging demand for generative artificial intelligence (AI). They will also boost the national economy and increase Australia’s digital sovereignty – a global shift toward storing and managing data domestically under national laws.

But the everyday realities of communities living near these data centres aren’t as optimistic. And one key step toward mitigating these impacts is ensuring genuine community participation in shaping how Australia’s data-centre future is developed.

The sensory experience of data centres

Data centres are large, warehouse-like facilities. Their footprint typically ranges from 10,000 to 100,000 square metres. They are set on sites with backup generators and thousands of litres of stored diesel and enclosed by high-security fencing. Fluorescent lighting illuminates them every hour of the day.

A data centre can emanate temperatures of 35°C to 45°C. To prevent the servers from overheating, air conditioners are continuously humming. In water-cooled facilities, water pipes transport gigalitres of cool water through the data centre each day to absorb the heat produced.

Data centres can place substantial strain on the local energy grid and water supply.

In some places where many data centres have been built, such as Northern Virginia in the United States and Dublin in Ireland, communities have reported rising energy and water prices. They have also reported water shortages and the degradation of valued natural and historical sites.

They have also experienced economic impacts. While data centre construction generates high levels of employment, these facilities tend to employ a relatively small number of staff when they are operating.

These impacts have prompted some communities to push back against new data centre developments. Some communities have even filed lawsuits to halt proposed projects due to concerns about water security, environmental harm and heavy reliance on fossil fuels.

A unique opportunity

To date, communities in Australia have been buffered from the impacts of data centres. This is largely because Australia has outsourced most of its digital storage and processing needs (and associated impacts) to data centres overseas.

But this is now changing. As Australia rapidly expands its digital infrastructure, the question of who gets to shape its future becomes increasingly important.

To avoid amplifying the social inequities and environmental challenges of data centres, the tech industry and governments across Australia need to include the communities who will live alongside these crucial pieces of digital infrastructure.

This presents Australia with a unique opportunity to set the standard for creating a sustainable and inclusive digital future.

A path to authentic community participation

Current planning protocols for data centres limit community input. But there are three key steps data centre developers and governments can take to ensure individual developments – and the broader data centre industry – reflect the values, priorities and aspirations of local communities.

1. Developing critical awareness about data centres

People want a greater understanding of what data centres are, and how they will affect their everyday lives.

For example, what will data centres look, sound and feel like to live alongside? How will they affect access to drinking water during the next drought? Or water and energy prices during the peak of summer or winter?

Genuinely engaging with these questions is a crucial step toward empowering communities to take part in informed conversations about data centre developments in their neighbourhoods.

2. Involving communities early in the planning process

Data centres are often designed using generic templates, with minimal adaptation to local conditions or concerns. Yet each development site has a unique social and ecological context.

By involving communities early in the planning process, developers can access invaluable local knowledge about culturally significant sites, biodiversity corridors, water-sensitive areas and existing sustainability strategies that may be overlooked in state-level planning frameworks.

This kind of local insight can help tailor developments to reduce harm, enhance benefits, and ensure local priorities are not just heard, but built into the infrastructure itself.

3. Creating more inclusive visions of Australia’s data centre industry

Communities understand the importance of digital infrastructure and are generally supportive of equitable digital access. But they want to see the data centre industry grow in ways that acknowledges their everyday lives, values and priorities.

To create a more inclusive future, governments and industry can work with communities to broaden their “clean” visions of digital innovation and economic prosperity to include the “messy” realities, uncertainties and everyday aspirations of those living alongside data centre developments.

This approach will foster greater community trust and is essential for building more complex, human-centred visions of the tech industry’s future.

Google Launches Lightweight Gemma 3n, Expanding Edge AI Efforts

• By John K. Waters
• 07/07/25

Google DeepMind has officially launched Gemma 3n, the latest version of its lightweight generative AI model designed specifically for mobile and edge devices — a move that reinforces the company’s emphasis on on-device computing.

The new model builds on the momentum of the original Gemma family, which has seen more than 160 million cumulative downloads since its launch last year. Gemma 3n introduces expanded multimodal support, a more efficient architecture, and new tools for developers targeting low-latency applications across smartphones, wearables, and other embedded systems.

“This release unlocks the full power of a mobile-first architecture,” said Omar Sanseviero and Ian Ballantyne, Google developer relations engineers, in a recent blog post.

Multimodal and Memory-Efficient by Design

Gemma 3n is available in two model sizes, E2B (5 billion parameters) and E4B (8 billion), with effective memory footprints similar to much smaller models — 2GB and 3GB respectively. Both versions natively support text, image, audio, and video inputs, enabling complex inference tasks to run directly on hardware with limited memory resources.

A core innovation in Gemma 3n is its MatFormer (Matryoshka Transformer) architecture, which allows developers to extract smaller sub-models or dynamically adjust model size during inference. This modular approach, combined with Mix-n-Match configuration tools, gives users granular control over performance and memory usage.

Google also introduced Per-Layer Embeddings (PLE), a technique that offloads part of the model to CPUs, reducing reliance on high-speed accelerator memory. This enables improved model quality without increasing the VRAM requirements.

Competitive Benchmarks and Performance

Gemma 3n E4B achieved an LMArena score exceeding 1300, the first model under 10 billion parameters to do so. The company attributes this to architectural innovations and enhanced inference techniques, including KV Cache Sharing, which speeds up long-context processing by reusing attention layer data.

Benchmark tests show up to a twofold improvement in prefill latency over the previous Gemma 3 model.

In speech applications, the model supports on-device speech-to-text and speech translation via a Universal Speech Model-based encoder, while a new MobileNet-V5 vision module offers real-time video comprehension on hardware such as Google Pixel devices.

Broader Ecosystem Support and Developer Focus

Google emphasized the model’s compatibility with widely used developer tools and platforms, including Hugging Face Transformers, llama.cpp, Ollama, Docker, and Apple’s MLX framework. The company also launched a MatFormer Lab to help developers fine-tune sub-models using custom parameter configurations.

“From Hugging Face to MLX to NVIDIA NeMo, we’re focused on making Gemma accessible across the ecosystem,” the authors wrote.

As part of its community outreach, Google introduced the Gemma 3n Impact Challenge, a developer contest offering $150,000 in prizes for real-world applications built on the platform.

Industry Context

Gemma 3n reflects a broader trend in AI development: a shift from cloud-based inference to edge computing as hardware improves and developers seek greater control over performance, latency, and privacy. Major tech firms are increasingly competing not just on raw power, but on deployment flexibility.

Although models such as Meta’s LLaMA and Alibaba’s Qwen3 series have gained traction in the open source domain, Gemma 3n signals Google’s intent to dominate the mobile inference space by balancing performance with efficiency and integration depth.

Developers can access the models through Google AI Studio, Hugging Face, or Kaggle, and deploy them via Vertex AI, Cloud Run, and other infrastructure services.

For more information, visit the Google site.