AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 568

Economics with Heart

Submitter group: Companies.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues that AI's physical footprint, particularly large data centres, can raise electricity and network costs for households and regional small businesses, eroding AI's productivity gains. Drawing on one field visit and desk-based case studies, with cost estimates described as illustrative, it makes nine recommendations on National Electricity Rules reform, cost recovery, demand response, local government capacity and cost-shift modelling.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. Argues that infrastructure-driven living costs can offset AI productivity gains, noting that network costs form a substantial part of household electricity bills.

    Any productivity gain from AI software can be offset if the infrastructure behind it raises the cost of living. Network costs are a substantial part of household electricity bills, and new transmission lines, substations and water trunk works for large loads must be paid for by someone.

  2. The submission states that regions host the physical footprint of AI while the returns, skilled jobs and digital capability are largely captured elsewhere.

    Regions host the physical footprint of AI: substations, transmission corridors, water and land. The returns, skilled jobs and digital capability are largely captured elsewhere.

  3. The submission argues that utility cost stability is a precondition for small business AI adoption and that full cost recovery from large loads protects it.

    Utility cost stability is a precondition for small business AI adoption, and full cost recovery from large loads protects it (Recommendations 1 and 4).

  4. Argues that WA’s capacity market passes large new loads’ capacity costs to all consumers unless those loads must pay them.

    WA’s main grid, the SWIS, is an island grid funded through a capacity market, the Reserve Capacity Mechanism. A large new load raises the capacity the market must buy, and that cost is passed through to all consumers unless large loads are required to pay it.

  5. Argues AI tools combining operator and community data could estimate community costs before closures, rather than afterward.

    On our base case, that gap cost the community about $29,200, or about $39,900 a year if it had become permanent (Case Study, Appendix A). AI tools that combine operator and community data could estimate such costs before a closure, not after (Recommendation 8).

  6. The submission states that national electricity reforms are built around the NEM and do not automatically reach WA or the NT.

    National reforms are built around the NEM and do not automatically reach WA or the NT.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Reform the National Electricity Rules before the 2027 national legislation so large data centres are visible to the grid, accountable to the public and pay their own way.

    Reform the National Electricity Rules before the 2027 national legislation, so that large data centres are visible to the grid, accountable to the public and pay their own way (Recommendation 1).

  2. Make large data centres pay the full capacity and network costs they trigger, including on isolated grids such as WA’s SWIS.

    Make large data centres pay the full capacity and network costs they trigger, including on isolated grids such as WA’s SWIS (Recommendations 3, 4 and 6).

  3. Make demand response a condition of connection for very large loads.

    Make demand response a condition of connection for very large loads (Recommendation 5).

  4. Use AI-enabled modelling to measure the cost shifted onto regional communities when services close.

    Use AI-enabled modelling to measure the cost shifted onto regional communities when services close (Recommendation 8).

  5. Provide host regions with enforceable community benefit, local procurement and grid upgrades that also serve local enterprise.

    To spread AI’s benefits, host regions need enforceable community benefit, local procurement and grid upgrades that also serve local enterprise (Recommendations 6 and 7).

  6. Where public money underwrites a large load's flexibility, make that flexibility a contracted, visible condition of connection.

    The lesson: if public money underwrites a large load’s flexibility, that flexibility should be a contracted, visible condition of connection.

  7. Use term of reference (n) to examine AI infrastructure, including data centres, while coordinating findings with the Senate inquiry and planned 2027 legislation.

    We ask the Committee to use term (n) to examine AI infrastructure, including data centres, and to coordinate its findings with the Senate inquiry into Artificial intelligence and data centres and with the planned 2027 legislation (Recommendation 9).

  8. Include in national standards specific provisions for isolated grids (the SWIS and the NT), rather than assuming NEM conditions.

    National standards include specific provisions for isolated grids (the SWIS and the NT), rather than assuming NEM conditions.

  9. Pair any registration threshold with anti-fragmentation rules, and lodge energy agreements of large loads on a public register with no size threshold.

    Any registration threshold is paired with anti-fragmentation rules, and energy agreements of large loads are lodged on a public register with no size threshold.

  10. Require large data centres connecting to the SWIS to pay the full capacity and network augmentation costs they trigger, so these are not passed to households and regional businesses.

    Large data centres connecting to the SWIS pay the full capacity and network augmentation costs they trigger, so these are not passed to households and regional businesses.

  11. Apply equivalent protections in Western Australia and the Northern Territory.

    Equivalent protections apply in Western Australia and the Northern Territory.

  12. Have large data centres register on-site storage and curtailable load as demand response, available during reserve shortfalls, including overnight.

    Large data centres register on-site storage and curtailable load as demand response, available during reserve shortfalls, including overnight.

  13. Give host local governments assessment capacity, cost recovery and enforceable community benefit provisions.

    Host local governments receive assessment capacity, cost recovery and enforceable community benefit provisions.

  14. Treat AI-enabled regional service-viability and cost-shift modelling as public-interest infrastructure.

    AI-enabled regional service-viability and cost-shift modelling is treated as public-interest infrastructure.

  15. Coordinate the Committee's findings with the Senate inquiry into AI and data centres and the planned 2027 legislation.

    The Committee coordinates its findings with the Senate inquiry into Artificial intelligence and data centres and the planned 2027 legislation.

  16. Provide Committee or Australia Post data on CPA payments, transaction volumes, PO-box and bill-payment users, and the LPO opening date to replace the study’s assumptions.

    We would welcome the Committee’s or Australia Post’s data to replace our assumptions: the CPA payment, Coleambally transaction volumes, PO box and bill payment user counts, and the LPO opening date.

Read the original

Open in the interactive guide