Australia’s approach to artificial intelligence infrastructure must avoid both uncritical enthusiasm and blanket restrictions, according to Labor’s AI policy advisor Andrew Charlton, as U.S. states impose sweeping bans on data center construction.
Charlton’s comments come as at least 12 U.S. states—including Texas, Florida, and Georgia—have enacted laws or executive orders prohibiting the construction of data centers for AI training without explicit state approval. The measures follow high-profile opposition from local governments and environmental groups concerned about energy consumption, land use, and potential job displacement. In Texas, Governor Greg Abbott signed an executive order in May requiring state approval for any new data center exceeding 50 megawatts of capacity, while Florida’s Senate passed a bill in June mandating environmental impact assessments for AI infrastructure projects.
“The Australian government should steer a middle course—neither boosterism nor alarmism,” Charlton told The Australian Financial Review on Monday. “We need to recognize the economic benefits of AI infrastructure while ensuring it’s deployed responsibly, with clear safeguards on energy use, local community impacts, and national security.” His remarks align with growing tensions between tech companies seeking to expand AI capabilities and regulators grappling with the sector’s rapid growth.
Why Are U.S. States Blocking AI Data Centers?
Opposition to AI data centers in the U.S. stems from three primary concerns, according to state officials and industry reports. First, energy demands: A single AI training facility can consume as much power as a small city. In Texas, the Electric Reliability Council of Texas (ERCOT) warned in April that unchecked data center growth could strain the grid during peak demand periods. Second, land-use conflicts have erupted in communities like Clark County, Nevada, where Meta’s planned $3.6 billion AI campus faced lawsuits from local residents over traffic and water usage. Third, national security risks have prompted scrutiny of foreign-owned facilities, particularly those linked to Chinese tech firms, which now account for nearly 40% of global AI chip production.
“This isn’t just about NIMBYism—it’s about ensuring these projects align with state priorities,” said Florida Senator Jeff Brandes, sponsor of the state’s AI data center bill. “We’re not anti-tech, but we won’t sacrifice our communities or energy reliability for short-term economic gains.” The legislation passed with bipartisan support, reflecting a shift from earlier pro-business stances in states like Georgia, which had aggressively courted tech investments.
How Is Australia Positioning Itself?
Unlike the U.S., Australia has not imposed blanket bans on AI infrastructure. However, the federal government is tightening oversight through two parallel strategies. First, the Critical Infrastructure Centre (CIC), established under the Security of Critical Infrastructure Act 2018, has expanded its mandate to include AI data centers deemed “systemically important.” The CIC identified 11 high-risk facilities in a December 2023 report, including NVIDIA’s Melbourne-based AI research hub and Google’s Sydney data center. Second, the Australian Energy Market Operator (AEMO) has introduced voluntary energy-efficiency benchmarks for AI operations, though enforcement remains limited.
Charlton’s call for a balanced approach reflects internal debates within the Labor government. While Treasurer Jim Chalmers has emphasized AI’s role in driving productivity growth—citing a 2024 Deloitte report projecting AI could add A$315 billion to Australia’s economy by 2030—other ministers, including Climate Change Minister Chris Bowen, have warned against repeating the U.S. states’ “patchwork” regulatory failures. “We need a national framework, not a free-for-all,” Bowen told a Senate committee in May.
What Happens Next for Australia’s AI Infrastructure?
The next critical step is the release of the federal government’s AI Infrastructure Strategy, expected in late July. Leaked drafts suggest three potential measures: mandatory energy audits for facilities exceeding 100 megawatts, a “fast-track” approval process for projects meeting sustainability criteria, and expanded CIC oversight for foreign-owned AI operations. However, industry groups warn that overly restrictive rules could deter investment, pointing to Canada’s experience, where AI firms have shifted projects to the U.S. due to regulatory uncertainty.
“Australia risks falling behind if it doesn’t get this right,” said James Pearson, CEO of the Australian Information Industry Association. “The U.S. states’ approach shows what happens when policy lags behind technology—but we also can’t afford to repeat their mistakes.” The strategy’s final form will hinge on negotiations between the Treasury, the Department of Climate Change, and the National Security Committee, with a decision anticipated before the next federal budget in October.
Meanwhile, state-level actions are already emerging. Victoria’s Labor government announced in June it would require all new AI data centers to submit climate impact assessments, while Queensland’s Resources Minister, Scott Stewart, signaled openness to tax incentives for low-emission AI facilities. The divergence between states could create a fragmented regulatory landscape, complicating Australia’s efforts to attract AI investment while maintaining oversight.
