Firmus lab approved without council vote, sparking calls for reform

A research and development facility for AI has been approved by a Tasmanian council, reviving calls to overhaul planning rules criticised as not fit for purpose.

AI data centre builder Firmus will convert an existing warehouse at the former Wesley Vale pulp mill, in the state's north-west, into a 26-megawatt facility for AI computing operations and research into alternative power sources during interruptions.

"Our Wesley Vale facility will allow Tasmanian engineers to test novel operating strategies for AI infrastructure and help solve a challenge facing the industry globally," co-founder and principal engineer for research and development (R&D) Jonathan Levee said.

"Building on software we developed and patented in Tasmania, we're exploring how AI computing can be more flexible and respond to electricity grid conditions.

"And how technologies such as batteries could help reduce reliance on diesel generators."

Separate to the R&D facility, Firmus remains committed to building an AI data centre at the site, also 26 megawatts, which the Latrobe Council has sought the state government to assess as a major project.

Last month, Firmus's highly anticipated public offering on the Australian stock market was abandoned after investor interest cooled. 

The company, which has plans to build three AI data centres in Tasmania, had hoped to go public at a valuation of $44 billion but withdrew its plans after interest failed to eventuate. 

Just 5 per cent of the company’s contracted data centre capacity was online when the company was trying to float.

Residents feel locked out of process

Across Australia, AI ventures have met opposition from residents, with one Melbourne community using cars and wheelie bins to obstruct the installation of power poles which would have increased energy supply to a nearby centre.

In the unconfirmed minutes from the October meeting, Latrobe Council received several community questions about the R&D facility.

The questions related to the building's footprint and location, how it would be powered, its planned diesel generator use, and the council's capacity to assess the project's scale and complexity.

Local resident Erin Woodcock said she felt locked out by the process.

"We don't really know what's happening right next door to us and what risks can be involved in that,"

she said.

"It's just all very unclear and worrying without the transparency of details."

Emerging tech slipping through gaps

Under Tasmania's planning scheme, the R&D application was not required to be advertised or voted on by councillors.

Instead, it was approved by council staff and formalised by the general manager, a process that outgoing mayor Peter Freshney said was standard for applications that fit the appropriate use class.

"There were no discretions to be considered, and if there had been, then yes, it would have come to council, and it would have undertaken the normal process and been advertised," Mr Freshney told 936 ABC Radio Hobart.

But Mr Freshney said it highlighted the constraints of the planning scheme, legislated in 1993, and demonstrated it wasn't fit for purpose.

"It doesn't provide a rigorous and robust assessment of these emerging technology development applications,"

he said.

"We're still strongly of the view that there needs to be some work put into the planning scheme to allow councils, acting as planning authorities, to adequately and appropriately assess these applications.

"It does seem rushed."

Process 'entirely appropriate', minister says

Planning and Housing Minister Kerry Vincent defended the process.

"It is entirely appropriate for developers to rely on the planning system when determining where to invest, and for local planning authorities to remain true to that system when assessing applications," he said.

Mr Vincent said the government had been clear that major AI developments must deliver for the local community, and had released its draft statement of expectations for data centres and AI infrastructure.

"The planning system shouldn't be politicised in the process,"

he said.

Firmus expects the R&D facility to create about 70 jobs during construction, with 20 ongoing roles once operational.

The company has stated it will use dry cooling for most of the year, with "little or no water required for cooling under typical operating conditions".

It estimates annual water use for cooling to be approximately 224,000 litres, subject to final design and operation conditions, with water to come primarily from rainwater collected on-site.

Construction will occur seven days a week, with reduced weekend hours.