AI infrastructure that benefits Tasmanian businesses

AI infrastructure that benefits Tasmanian businesses

Personal submission to Building Tasmania

Alex Carpenter | 01/10/2026

Consultation: Draft Expectations for Data Centres and AI Infrastructure Developers

Purpose of this submission

Thank you for the opportunity to contribute. I am a PhD researcher studying how Australian small and medium enterprise leadership teams frame and respond to AI. I make this submission in a personal capacity.

I support developing AI infrastructure in Tasmania. My main recommendation is to connect that investment to capability that Tasmanian businesses can actually use, supported by new renewable generation, proportionate water requirements and timely approvals. Government should also retain the option to invest substantially alongside private providers where a project delivers a clear public benefit.

The draft is non-statutory guidance for infrastructure development, and broader AI policy is outside this consultation's scope (Building Tasmania, 2026). My comments concern the benefits and conditions of infrastructure investment, particularly its contribution to local business capability and Australian control of AI services.

Make business capability a practical local benefit

Under Expectation 1, I would like the assessment of local benefits to include whether Tasmanian businesses gain useful access to AI services and the skills to put them to work. Construction activity and operating jobs matter. The wider opportunity is to help businesses improve their everyday work using the capability that this infrastructure supports.

Access to computing resources alone is insufficient. People need help to develop useful workflows, judge outputs and understand their responsibilities. Training and implementation support should be treated as substantive contributions from infrastructure providers and their partners. Those contributions should be assessed by what businesses can do with the support, rather than by a nominal allocation of compute or the number of introductory sessions delivered.

The OECD's 2025 study of SMEs found that a third or fewer of businesses using generative AI had taken measures such as staff training, internal guidelines or investigating legal issues. Respondents welcomed training support. The survey covered seven countries and did not include Australia, so these findings are not an estimate of Tasmanian practice (OECD, 2025). They nevertheless support giving implementation capability serious attention alongside infrastructure access.

Support should be useful to small businesses with limited time and technical capacity. For example, a provider could work with local training and business organisations to help firms build and assess workflows relevant to their operations. The objective should be sustained capability, with businesses able to make informed choices about the services they use.

Make claims of sovereign capability meaningful

Tasmanian infrastructure could contribute to an Australian inference option that businesses can depend on. Inference is the use of a trained model to produce an output. My priority is that every Australian business, including those in Tasmania, should have a practical option to use capable models served within Australia under Australian control.

Where a development claims sovereignty as a public benefit, that claim should be assessed through the legal and practical ability of an Australian operator to continue running an agreed model if its overseas partner withdraws access. A local building and local data storage do not establish that ability by themselves. Training a model in Australia may bring other benefits, but the location of training does not settle who controls its subsequent use.

The expectations should leave the means of achieving this outcome open. International partnerships can help deliver Australian capability, and businesses should remain free to choose overseas services. Assess the rights, operating arrangements and security of the particular proposal. This would make sovereign capability a meaningful infrastructure benefit without requiring every facility to provide the same services or every business to use them.

Retain a role for strategic public investment

The draft expects commercial viability without government financial support and seeks to protect existing customers from inappropriate cost shifting through utility infrastructure (Building Tasmania, 2026, pp. 7, 9). I support scrutiny of project economics and a fair contribution from developers. I would reconsider any absolute exclusion of public investment.

If government wants infrastructure that delivers lasting Australian control and benefits for Tasmanian businesses, it should be prepared to contribute to that outcome. Substantial public investment alongside private investment can be justified where the public benefit is clear. This may include shared electricity and water infrastructure that supports the capability.

I accept that some costs may be shared through household and business electricity and water bills. Those contributions should be explicit and justified. People should be able to understand what they are paying for, what the developer contributes and what benefit the public receives. Protecting consumers requires that scrutiny; it need not prohibit every shared contribution.

I suggest revising the financial support expectation to allow a transparent case for strategic public investment. The assessment should explain the capability being secured, who bears the costs and risks, and why the proposed contribution is warranted. Government should be able to distinguish projects seeking ordinary commercial returns from investments that also deliver a capability it has deliberately chosen to support.

I do not support general subsidies for AI consumption. Businesses and other users should pay for the services they use, including a possible premium for Australian control and continuity. Investment in infrastructure and practical skills should help establish useful capability without becoming a general discount on users' AI bills.

Expand renewable supply with new demand

I strongly support new renewable generation to match data-centre electricity demand. The draft's allowance for managing differences between the timing of new load and new generation is a useful practical feature (Building Tasmania, 2026, pp. 10–11).

Government, utilities and developers should use AI infrastructure investment to accelerate renewable generation and the network infrastructure needed to connect it. The assessment should consider how the proposed demand and additional supply can be brought into operation together, with credible arrangements for any transition.

The expectations should provide a workable route through those decisions. Where a connection or supply constraint exists, the parties should examine whether changes to location, project staging or shared infrastructure can resolve it. Clear responsibilities and timely decisions would help proponents commit to the generation and network investment required.

I favour practical flexibility about how projects achieve the energy outcome. The central objective is additional renewable supply that supports the new demand while maintaining a dependable electricity system.

Assess water use in its local context

Water requirements should reflect the facility's actual demand, cooling design and local supply conditions. Average use, peak requirements and performance during dry periods are relevant to deciding whether a proposal is suitable for its location. The assessment should recognise the differences between cooling systems and available water sources.

Low-water designs and suitable recycled-water supplies should help enable timely development. Where a genuine constraint exists, government, the utility and the developer should work through practical options, including shared infrastructure or a more suitable location. The level of assessment should remain proportionate to the proposed use and the local risk.

I support protecting existing water security. I also want the expectations to make clear how a proponent can demonstrate that a concern has been resolved. Once the evidence supports a workable arrangement, water concerns should not become a reason for open-ended delay.

Coordinate decisions and make commitments visible

The Office of the Coordinator-General's coordinating role and the use of existing approval pathways provide a sensible basis for implementation (Building Tasmania, 2026, pp. 9, 14). That coordination should help proponents identify genuine constraints early, bring the relevant agencies together and obtain timely decisions.

The economic cost of waiting belongs in the assessment. In my view, the opportunities forgone through unnecessary delay are a larger concern than continued overseas dependence alone. Responsible development needs clear decisions about whether a project can proceed and what would make it workable. Where moving or changing a project resolves a real constraint, that option should be pursued promptly.

Community engagement should explain what is proposed, what it will require and how people may benefit. If costs will be shared, say so plainly. People should be able to distinguish an expected benefit from a firm commitment and see how delivery compares with what was promised.

Public reporting should focus on material commitments and actual performance, with appropriate protection for confidential and security-sensitive information. Reusing information already required through approval and utility processes would help keep reporting proportionate. The purpose should be to make the project's effects and contributions understandable.

Recommendations

I recommend that Building Tasmania:

• Give practical AI implementation support and skills for Tasmanian businesses a clear place in the assessment of local benefits.
• Assess any claim of sovereign capability through the legal and practical ability to continue operating agreed models under Australian control, while preserving international partnerships and business choice.
• Allow substantial strategic public investment alongside private investment, including transparent contributions to shared utility infrastructure, while leaving users responsible for the AI services they consume.
• Support new renewable generation to match demand, with practical coordination of generation, connections and project timing.
• Apply proportionate water requirements that recognise actual demand, cooling design and local supply, and help resolve genuine constraints.
• Use coordinated approvals, clear community information and proportionate reporting to support timely, responsible development.

References

Building Tasmania. (2026). Draft expectations for data centres and AI infrastructure developers: Tasmania.

OECD. (2025). Generative AI and the SME workforce: New survey evidence. OECD Publishing.