The Infrastructure Nobody Talks About: Why Internet Will Determine the AI Future

By Daniel Filmer, Move Up Internet Founder & Managing Director.

Artificial intelligence is transforming how we work. From ChatGPT writing emails to automated customer service, from smart analytics to real-time translation, AI tools are everywhere.

But there’s a fundamental requirement that almost nobody discusses: AI runs on internet infrastructure. And most of Australia’s internet simply wasn’t built for it.

The Problem Hiding in Plain Sight

I’ve had countless conversations with people frustrated that AI tools don’t work as advertised. Video calls that freeze. Cloud applications that timeout. File uploads that take forever. AI assistants that lag so badly they’re unusable.

The technology isn’t broken. The internet connection is.

We built networks for a world of web browsing and email. AI needs something completely different: constant, heavy data flow in both directions. When working with cloud-based AI, users aren’t just downloading information. They’re continuously uploading data, syncing files, and pushing information back to remote servers.

Most Australian internet infrastructure was never designed for this.

What Changed

Ten years ago, home internet was mostly about downloading. Streaming Netflix, browsing websites, checking email. Upload speeds didn’t matter much because people weren’t sending large amounts of data anywhere.

AI flipped that model entirely.

Now, people are recording video meetings that get transcribed in real-time. Backing up entire photo libraries to cloud storage for AI organisations. Using design tools that process everything on remote servers. Running smart home devices that constantly communicate with cloud platforms.

Every one of these activities requires substantial upload bandwidth, exactly what most internet connections can’t reliably provide.

What Actually Matters Now

The old metrics for evaluating the internet don’t tell the full story anymore. Here’s what actually matters for AI-era connectivity:

Upload speeds that match downloads: Most plans advertise download speeds but bury upload speeds in fine print. For AI tools, upload matters just as much.

Consistent performance: Speeds that vary wildly throughout the day make cloud-based AI tools unreliable when needed most.

Genuine unlimited data: AI workflows use far more data than traditional internet activities. Artificial caps and throttling break the experience.

Low latency: The delay between devices and cloud servers determines whether real-time AI applications actually work.

The Infrastructure Gap

The major telcos built networks optimised for a different era. They’re upgrading, but slowly, and often only in select areas. Meanwhile, AI adoption is accelerating quickly.

This creates a strange situation where the technology is ready, willing to transform how we work and live, but the basic infrastructure to support it isn’t universally available.

For apartment buildings in particular, the problem is acute. Many MDUs are stuck with shared infrastructure that was barely adequate five years ago. As more residents try to use AI tools simultaneously, networks become even more congested.

Looking Forward

The conversation about AI usually focuses on ethics, job displacement, and which tools are most powerful. Those are important discussions.

But we’re overlooking something more fundamental: AI transformation requires infrastructure transformation. The regions and communities that invest in proper connectivity will access opportunities that others simply can’t reach.

This isn’t about being early adopters or tech enthusiasts anymore. AI tools are becoming essential for work, education, and daily life. Access to them shouldn’t depend on whether a building happened to get proper internet infrastructure.

The AI revolution everyone’s talking about? It’s already here. But it only works if the internet can support it.