Threat intelligence company, Alethea identified campaigns attributed to China, Russia and Iran targeting allied data centre projects in research published July 2026. According to researchers, fake local news sites were used to amplify conversation around data centre opposition, and state media published satellite imagery of construction sites, framing them as strategic military infrastructure.
The reporting illustrated that while governments debate how to govern strategic computer infrastructure, adversaries are already treating it as contested terrain and exploiting domestic fissures.
Photo by Annie Spratt on Unsplash
Beyond the Build
The dominant narrative around artificial intelligence infrastructure is one-dimensional: who builds the biggest, who builds the fastest. In Washington, Senate hearings have focused on acceleration, while in Beijing, AI is framed as a global leadership opportunity with emphasis on embodied intelligence and robotics.
Unlike many technological advancements, which are abstract, data centres offer a tangible representation of progress and competition. Measured in metrics like gigawatts, capital expenditure and chip counts, it becomes easier to compare and contest the merits.
One of the most consequential questions is strategic: it isn’t who builds the data centre, but who controls the infrastructure. Data centres impose costs on the local economy, vacuuming up electricity, water and land. They demand human resources to build, a labour force to power it, planning approvals, community consent and long-term grid investment. The host country absorbs the physical and social burdens in the short, medium and long-term, but control can remain offshore indefinitely.
Australia’s Position
Australia provides an intriguing illustration of the data centre dilemma. As a Five Eyes member and long-term US ally, it has become an attractive destination for hyperscale investment thanks to political stability, a reliable legal system, proximity to Asian markets, and a large geographic footprint.
However, despite continued investment in infrastructure and startups like Sovereign Australia AI, which is developing the ‘Australis’ foundational Australian Large Language Model, Australia remains dependent on overseas technology for much of its advanced AI compute.
Recent comments by Dr Andrew Charlton, Assistant Minister for Science, Technology and the Digital Economy, are particularly revealing. Charlton argued that Australia risks “renting its future,” framing a choice between being “tenants in someone else’s digital future” and a country that “helps build it.” The tenant/owner framing assumes hosting confers ownership, but in practice, it’s possible to host the shed and still rent the stack.
Two Empires, Same Outcome
The AI cold war has exposed a new era of superpower competition, with two vastly different approaches.
The American model is reliant on legal and contractual reach. The US CLOUD Act allows American authorities to compel US service providers to produce data within their possession, custody or control, regardless of where it is stored.
The reach isn’t only statutory. Hyperscale contracts reserve operational levers for the providers, so although a host government can regulate the building, regulating a terms-of-service update is somewhat trickier. Add export controls – what chips or model weights an offshore facility may run – and the host’s ‘sovereign’ infrastructure is governed day-to-day from somewhere else.
’This is another way we can think about it: maybe the chips only go to allies of the US, where they eventually also can enforce these controls. And then the US can enforce this, particularly with all of the allies within a semiconductor supply chain, to just make sure, like, “We have these new rules; how are we going to use AI chips responsibly?” — and you only get these chips if you follow these rules. That’s one way of going about it. Otherwise the chips are not going to go there.’
China’s approach centres on infrastructure ownership and state-linked investment abroad. Papua New Guinea’s National Data Centre, built by Huawei and financed by a US$53 million China Exim Bank loan, was designed to store the government’s data archive. The strategy has been labelled ‘debt-trap diplomacy’, but beyond the issue of debt, security considerations remain at the forefront. A 2020 technical review, partly funded by Australia, found obsolete encryption and core switches outside the firewalls. PNG’s communications minister called it a “failed investment” and sought to cancel the loan.
Southeast Asia hosts more Chinese-owned data centres than anywhere outside China. Malaysia provides a fascinating case study of the push-pull between the US and China. Viewed as a neutral hub with presence from both US and Chinese firms, the low cost of infrastructure and its geography led to over $30 billion in US investment in 2024. In 2025, Malaysia began reining in data-centre expansion following concerns about resource use and how the infrastructure had become a workaround for Chinese access to restricted AI chips.
Two empires, same outcome.
Why Influence Operations Work
Infrastructure operations are often understood as sabotage rather than espionage. But from Huawei’s build in PNG to the campaigns targeting data centre debates, the pattern is the same – trust in critical infrastructure becomes a tool of statecraft.
These operations don’t depend solely on convincing communities that data centres are physically dangerous. They exploit a more credible tension: communities experience the local costs directly – higher electricity demand, industrial land use, water consumption, a changing landscape – while the strategic benefits are less tangible. Adversaries don’t need to invent the grievance; they only need to amplify it, and each success emboldens the next attempt.
The vulnerability is not the facility; it’s the unresolved governance question surrounding it. Adversaries identified this gap well before many regulators did.
Thanks for reading Seeline Cyber! This post is public so feel free to share it.
Mine the Gap
Australia already regulates part of this landscape via The Security of Critical Infrastructure framework, which helps govern resilience and operational security for designated assets. The Foreign Investment Review Board assesses foreign acquisitions at the point of investment. While both are important, neither fully addresses ongoing strategic control after infrastructure is built.
Assessing a data centre currently means asking two questions: who owns it (once, at investment) and will it remain running (continuously, under SoCI). But strategic control is multidimensional, and most of its dimensions sit outside both frames. These include legal reach over the data (CLOUD ACT), operational reach over the stack (contract terms, software dependencies), supply reach over what the facility can run (export controls), and the local ledger of water, power and land that communities carry while control accrues elsewhere. No Australian framework is responsible for assessing these together, or for reassessing them as they shift. Alethea’s findings suggest adversaries already assess them this way. The environment debate and the sovereignty debate are usually run separately, and that separation is the vulnerability.
Leave a Reply