In July 2026, over 200 community members came together at Mandoon Bilya in Perth’s eastern suburbs to plant more than 3,000 native plants as part of a wetland restoration project. This scene stood in contrast to what the area might have become if a major AI data center had gone ahead. The site was originally planned to host one of Australia’s largest hyperscale artificial intelligence facilities. Local Aboriginal groups, environmental activists, and community organizations successfully blocked the development. Instead of a 120-megawatt AI facility, the land is now being used for ecological restoration and protecting traditional Country.
The campaign against the Mandoon Bilya project reflects broader concerns about how AI infrastructure decisions are made in Australia, according to Phys.org. While policymakers focus on regulating AI technologies themselves, many Indigenous communities have had little say in planning the physical systems that support them. This includes how resources and infrastructure are allocated for AI development, raising questions about future Indigenous relations with government on digital matters.
The Mandoon Bilya campaign was led by the Bibbul Ngarma Aboriginal Association and supported by environmental and community groups. Developers had planned to build on land that contained important Aboriginal heritage sites. Opponents argued the project would threaten a river system with deep cultural significance for local Aboriginal groups. The river also plays a role in Perth’s drinking water supply and supports local ecosystems during hot weather. These concerns raised questions about whether free, prior, and informed consent was properly obtained for the development.
In May 2026, the proposal was canceled before it could be reviewed by Western Australia’s Development Assessment Panel. Campaigners welcomed the decision and have since redirected their efforts toward broader ecological restoration plans.
Similar debates are emerging in other parts of Australia, such as Victoria, where opposition to a proposed AI Hub has raised concerns about cultural heritage and environmental impacts. AI is often seen as an abstract concept, but it relies heavily on physical resources like land, water, and energy. Data centers require large amounts of these resources, creating environmental pressures that many Aboriginal communities already face.
Internationally, Indigenous groups are also challenging proposed data center projects on similar grounds. They argue such developments compromise Indigenous rights, culturally significant sites, and water systems. These challenges reflect principles found in the United Nations Declaration on the Rights of Indigenous Peoples.
More people now recognize the importance of including Indigenous leadership in data and digital technologies. This includes support for Indigenous data sovereignty, which means Indigenous people control access to their own data. Indigenous leadership should also be central in decisions about the physical systems that enable AI technologies.
Some projects are beginning to include Indigenous participation and ownership in their planning. Western Australia’s largest data center is being developed through an Indigenous Land Use Agreement with the Karajarri People. They are also co-owners of that project. In the Northern Territory, Energy North is working with Traditional Owners as part of its assessment process for a proposed hyperscale data center.
However, these examples are still rare in Australia. While the country is focusing on regulating AI infrastructure, few policies require Aboriginal leadership in decision-making.
Kaley Nicholson, chief executive of the Federation of Victorian Traditional Owner Corporations, said this confirms a gap in current expectations. She noted that these expectations do not include collaboration with Traditional Owner groups, despite lasting impacts on Country. This raises a key question: Who gets to decide about digital systems that directly affect Indigenous lands, rights, and livelihoods?
Conflicts over AI infrastructure echo long-standing debates about resource extraction in Australia. Aboriginal peoples have often been left out of decisions about such developments on or near their Country. Yet they frequently bear the environmental, cultural, social, and personal consequences.
Research with Aboriginal communities shows that consultation processes don’t always lead to real influence. Communities often do the work of engagement and negotiation while key decisions are made elsewhere. Agreements don’t always result in better outcomes due to the challenges of culturally sensitive participation. Participation requires a lot of time and unpaid work, often falling on Elders and leaders who manage logistical burdens.
As AI systems grow, it’s important that partnerships are not just symbolic but meaningful and not overly demanding.
Meanwhile, the rapid growth of AI data centers is creating a shortage of memory chips used in everyday devices like laptops, phones, and gaming consoles, according to a report from govinfosecurity.com. The world’s three largest memory chipmakers are shifting production to High Bandwidth Memory for AI data centers. This shift has reduced the supply of traditional DRAM and NAND chips used in consumer electronics. As a result, tech manufacturers are paying more for components and passing those costs to customers through higher prices.
The expansion of AI is driving up costs across the technology industry. An EY survey found that while 82% of leaders are concerned about AI token costs, only 64% actively monitor usage and set clear budgetary guardrails. CIOs must connect AI investments to business outcomes to ensure spending delivers real value. To manage costs effectively, companies should route routine tasks to cheaper models and reserve the most expensive ones for high-impact work. Organizations also need to fund cross-functional AI projects rather than siloed departmental budgets. This strategic approach helps ensure AI spending drives actual business growth instead of just increasing expenses.
CIOs should also build feedback loops that track the cost of each outcome against new revenue or operational improvements. By reinvesting those savings into future AI initiatives, companies can create a value flywheel that supports long-term growth.
Despite rising costs, 37% of leaders are still expanding their AI rollouts. The difference between successful and unsuccessful AI adoption lies in focus and strategic planning. Market leaders use cost pressure to prioritize what actually works rather than scaling expensive experiments. They also build strict budgets and guardrails into their AI projects from the start. By tying spend to core business goals, they see measurable returns on their AI investments.
Nearly 90% of organizations are piloting or deploying AI-built internal software, but 72% say these efforts are already encountering obstacles. CIOs should consider whether an application is core to the business and unique IP or if a vendor already offers that capability well. They also need to evaluate total cost of ownership including maintenance, security, and talent requirements. Building software just because it’s easy is not enough reason — there must be clear business value beyond upfront savings. Success in AI depends more on how CIOs use available tools with their own data than on the tools themselves.
IMAGE: Christopher Bowns / Wikimedia Commons (CC BY-SA 2.0)