Luke Coleman, CEO ATA Speech to CommsDay Wholesale Congress
Delivering the digital infrastructure to meet Australia’s AI ambitions
14 October 2026
Good morning everyone, and thank you the CommsDay team for the opportunity to speak.
This year’s CommsDay Wholesale Congress seems to have landed as political interest in datacentres and AI hits fever-pitch.
Over the past month alone, the ATA has provided submissions to no fewer than four different Parliamentary inquiries into AI and datacentres.
There’s the Senate Communications Committee inquiry into AI and data centres.
The Joint Select Committee Inquiry into Artificial Intelligence.
The Tasmanian Parliament’s inquiry into the construction and operation of AI data centres.
And on top of these inquiries, there’s National Cabinet – the group of Commonwealth, State, and Territory Leaders – which has agreed to develop nationally-consistent laws and mandatory standards for datacentre deployments.
And, if that wasn’t enough, the South Australian Parliament has announced a Royal Commission into Artificial Intelligence.
You would think that with this intense political focus on datacentre deployments to enable AI, that Governments would place equally high importance on the connectivity required to connect them.
Water and energy infrastructure are always in scope for these inquiries, and we see regular headlines about whether or not there’s sufficient power and water to meet datacentre demand.
But there seems to be an expectation, across all levels of Government, that telecommunications infrastructure will simply materialise out of thin air.
As the people in this room know better than anyone, telecoms infrastructure does not appear with the wave of a magic wand.
So today, I’d like to speak about the role the telecoms sector needs to play in the Government’s AI and datacentre policy agenda, which the Prime Minister calls ‘AI in Australia’s interests’.
First, I’ll talk to the foundational role that fibre networks play in enabling AI.
Second, I’ll talk about the role of mobile networks in AI, and the spectrum required to meet future demand.
And third, I’ll focus on the significant reforms required to enable the telco sector to deliver ‘AI in Australia’s interests’.
So let’s begin with the foundational role that telco plays in enabling AI.
In December last year, the Commonwealth released its ‘National AI Plan’.
This plan recognises that AI is dependent on two critical components: data centres, and high-speed telecoms networks[1].
Without connectivity to submarine cables, fibre-optic networks, and 5G networks, data centres are stranded assets.
Just a few months ago, the Prime Minister delivered a speech titled ‘AI in Australia’s interests’, where he made the case for localised AI development.
He said, “We want more Australian companies and global firms developing AI here. To boost our sovereign capability. To strengthen our national security. And build our economic resilience.”[2]
If the Government has these ambitions for AI, then there is a strong argument that these same ambitions should be applied to telco networks.
We should want more Australian companies and global firms developing digital infrastructure here.
To boost our sovereign capability.
To strengthen our national security.
And build our economic resilience.
The generational investment boom in the datacentres that house AI applications should not take investment in telecoms networks for granted.
Telco industry analyst David Kennedy from Venture Insights has called out that on average, Australia’s three largest telcos are making returns on invested capital below their cost of capital.
In other words – they’re losing money on new network investments[3].
He pointed out a ‘decoupled trend’, where rapid data traffic expansion is not translating into revenue growth.
This is a fundamental problem – not just for our industry, but for our nation.
The digital infrastructure that provides the connective tissue to AI datacentres should be a key part of Australia’s National AI plan.
If we want to boost our sovereign capability,
To strengthen our national security,
And to build our economic resilience,
Then we need investment in the networks that enable it.
The role played by digital infrastructure in our national security was spelled out in detail by the defence minister in a recent speech.
Speaking on the topic of maritime security in the Indo-Pacific, the defence minister dedicated the majority of the speech to submarine cable security.
That’s significant.
When the Government is spending $368 billion on nuclear submarines[4], the defence minister chose to speak about submarine cables.
He said that, globally, we were witnessing “attacks against subsea critical infrastructure at a scale and frequency that is historically unprecedented.”
Australia was singled out as “among the most exposed nations in the world to this threat.”
He said that “Around ninety-nine percent of Australia’s internet traffic flows through just fifteen subsea cables. Consider that number for a moment. Fifteen cables — fifteen physical assets on the ocean floor — carry essentially the entirety of our international digital connectivity.”
“Our financial systems, our health systems, our communications, our intelligence partnerships, our ability to operate as a modern economy and a functioning state: all of it is critically dependent on infrastructure that is exposed, that cannot move and – as we have now seen demonstrated in the Baltic – can be cut with an anchor in the middle of the night.”[5]
The core argument here does not just apply to submarine cables.
All telecommunications networks are the critical infrastructure required to operate “our financial systems, our health systems, our communications, our intelligence partnerships, our ability to operate as a modern economy and a functioning state.”
As more and more sectors of the Australian economy utilise AI, this dependence on digital infrastructure will only grow.
Which brings me to my second point: meeting AI demand will take more than just fibre and submarine cables.
6G and satellites will be equally important for consumers accessing AI applications, and it will require a lot of new spectrum to enable that to happen.
The ATA recently merged with the Australian Mobile Telecoms Association, AMTA, which has commissioned a report by Deloitte Access Economics that considers the importance of mobile connectivity for accessing AI.
While the report is still in development, Deloitte has undertaken consumer research which found that almost two-thirds of Australians already use AI-powered mobile tools.
Of that figure, almost 60% said that mobile AI tools made them more productive, saving them 1 hour and 45 minutes per week, on average.
New smartphones are increasingly marketed on their new AI capabilities.
The latest Samsung Galaxy and Google Pixel smartphones are advertised for their AI services, which are built into applications like phone calls, messaging, photo editing, live voice translation, calendars, and more.
Similarly, Apple’s latest iPhone models market their AI capabilities, promoting a revamped Siri as “Your AI assistant”, integrating AI with smart home devices, fitness applications, and enabling AI agents to update compromised passwords and create automated shortcuts.
On-device AI capabilities are complemented by cloud-connected AI processing, driving more data across 4G and 5G networks.
According to AMTA’s latest ‘Future of Mobile’ report, mobile data downloads are increasing at a rate of around 33 per cent annually, with no sign of slowing.
But mobile AI applications will shift network requirements towards greater upload capacity.
According to Ericsson’s projections, a ‘medium’ AI adoption scenario would drive a 3-fold increase in uplink traffic by 2031, while ‘high’ adoption scenarios could push that to a 5-fold increase[6].
Analysis of mobile network traffic growth across 55 telcos globally showed that uplink traffic is growing faster than downlink – and in many cases, significantly faster.
43 out of 55 operators experienced a higher uplink growth rate than downlink. And 17 out of 55 operators experienced a more than 1.5 times higher uplink growth rate than downlink[7].
Mobile networks built to handle AI traffic will require a lot more than just new RAN equipment.
They will require new spectrum.
For Australia to be prepared for this, four priorities stand out – and we only have a short window of time to address them.
The first two are about enabling Australia’s successful transition to 6G:
- Three contiguous blocks of 200 MHz spectrum in the upper 6 GHz and lower 7 GHz bands to enable the launch of 6G from 2029,
- And freeing up low-band 600MHz spectrum for 6G in regional areas, once policy decisions have been made on the future of broadcast television services which currently occupy the band.
The second two priorities are about enabling the Universal Outdoor Mobile Obligation, or UOMO, and Direct-to-Device services:
- Enabling access to harmonised 2 GHz MSS spectrum for ubiquitous connectivity, and,
- Developing new spectrum options to support future D2D growth.
If Australia wants to pursue a policy objective of ‘AI in Australia’s interests’, we need to quickly make the spectrum available for future 6G and direct-to-device networks.
Because Australians are increasingly accessing AI applications in the palm of their hand, as much as in the datacentre.
And that brings me to the third part of my speech: major reforms are required to enable the telco sector to deliver the connectivity needed to support Australia’s AI ambitions.
Faster deployment of digital infrastructure must be a national reform priority.
Shortly after the last federal election, the must-read book for anyone into politics was ‘Abundance’ by Ezra Klein and Derek Thompson.
The Treasurer read it. The Chair of the Productivity Commission read it. The Assistant Minister for Productivity read it.
The Treasurer said the book was doing the rounds with his colleagues[8].
So being a political tragic myself, I read it too.
If I could summarise it in one sentence, the core message of the book is: we need to make it easier to build stuff.
Less red tape. Fewer hoops to jump through. Less bureaucratic meddling. More building.
In the Treasurer’s own words, “we’ve got to stop getting in our own way.”[9]
I hope that it becomes mandatory reading for anyone involved in telco planning approvals.
Because we have to make it easier to build digital infrastructure in this country.
Three structural barriers are actively limiting the telco sector’s capacity to further improve productivity across the economy:
1. Access to key inputs, particularly spectrum,
2. Excessive regulatory cost and complexity, and
3. Multi-jurisdictional deployment roadblocks.
We need to stop diverting millions into unnecessary regulatory costs, and turn that investment towards critical digital infrastructure.
Mobile towers can be constructed in just a week or two – yet council assessments take an average of 211 days.
In some jurisdictions new builds can take up to two years to be approved, despite the community crying out for better mobile coverage and capacity.
Here’s just a few real-world examples:
- An application for one new mobile site took 664 days before it was ultimately refused.
- Another site took 377 days tangled in bureaucratic red tape before it was refused.
- Applications in one local government area routinely take more than 450 days to be reviewed.
To put that in perspective: the Empire State Building was built in 410 days.
That’s right: the most famous skyscraper in the world – 103 stories high – was built faster than the time it takes some councils to review an application for a new mobile site.
We have to make it easier to build digital infrastructure in this country.
It’s a similar situation for fibre optic networks, the backbone of the internet.
The deployment of one inter-capital fibre route required:
- More than 3,000 land access activity notices,
- More than 1,100 construction certificates,
- More than 1,700 land access surveys, and
- More than 170 cultural heritage and environment assessments.
Another project required engagement with more than 200 different entities across State and local government, native title holders, tenement title holders, and private land title holders.
Construction through one segment just 30km long required 16 different approvals, across four government departments, and 12 different title holders.
This excessive red tape resulted the project being delayed by a year and half.
I mentioned earlier that AMTA had commissioned a report by Deloitte Access Economics on Reforms to Modernise Australia’s Telecommunications.
Deloitte found that small reforms would produce disproportionately large economic gains – especially in new growth areas and regional Australia.
A 10% improvement in regulatory practices could unlock around $430 million – which could then be channelled into additional telecommunications investment annually.
A 25% reduction in approval timeframes would allow industry to deliver between 150 and 200 accelerated mobile sites every year, bringing improved connectivity to up to 250,000 Australians sooner.
And a 10% reduction in regulatory complexity with nationally consistent planning practices would deliver around $150 million in annual productivity gains.
The report identified 3 short term opportunities:
First, establishing a national Digital Infrastructure Coordinator General, to support streamlined governance to accelerate significant deployments.
Second, create a harmonised planning framework with incentives for adoption by local councils.
And third, amend Schedule 3 of the Telecommunications Act to adopt a risk-based approach to fast-track low-impact builds.
We were pleased to see the Government commit to accelerated planning approvals announced as part of the Federal Budget.
It can’t happen fast enough.
But it’s also nowhere near ambitious enough.
The report also identified 3 longer term opportunities:
To develop a National Spectrum Strategy to provide investment certainty, drive efficient allocation and lay a pathway for the rollout of 6G.
Creating targeted planning and land-access relief for Government co-funded mobile sites.
And a process for streamlining the Tel Act over time, so it can be flexible in managing the requirements of a rapidly-changing sector.
We have to make it easier to build digital infrastructure in this country.
So I’ll conclude by saying this:
Australia’s AI future depends on telecommunications infrastructure as much as it depends on compute infrastructure.
‘AI in Australia’s interests’ ultimately requires Australian digital infrastructure – fibre networks, mobile networks, and satellite networks alike.
We need them to boost our sovereign capability.
To strengthen our national security.
And to build our economic resilience. As a nation, if we want these things, then we have to make it easier to build digital infrastructure in this country.
[1] https://www.industry.gov.au/publications/national-ai-plan
[2] https://www.pm.gov.au/media/ai-australias-interests-0
[3] https://ventureinsights.com.au/research/telecom/state-australian-telecommunications-industry-crossroads
[4] https://www.aspistrategist.org.au/budgeting-for-australias-nuclear-powered-submarines/
[5] https://www.minister.defence.gov.au/speeches/2026-05-30/address-2026-shangri-la-dialogue-plenary-session-three
[6] https://www.ericsson.com/en/reports-and-papers/mobility-report/articles/rising-uplink-demand-ai-driven-mobile-networks
[7] https://www.ericsson.com/en/reports-and-papers/mobility-report/articles/rising-uplink-demand-ai-driven-mobile-networks
[8] https://www.theguardian.com/australia-news/2025/jul/13/abundance-the-us-book-is-a-sensation-among-our-progressive-mps-but-can-it-spur-action-in-canberra
[9] https://www.lowyinstitute.org/external-publications/are-we-getting-our-own-way
