AI is beginning to starve; its next meal should be in your portfolio

Picture AI as a ravenous beast, constantly prowling for its next meal. Every tweet, purchase, or smartwatch heartbeat fuelling the AI machine.

But here's the kicker: most of that data is either locked away or a jumbled mess. Junk food for AI models that hunger for the finer things in life.

Human-labelled data. High-quality written text. Accurate and up-to-date information — that’s the caviar.

In today’s world, the most valuable commodity isn’t oil or gold — it’s quality data.

Former CEO of Google, Eric Schmidt, wowed audiences back in 2010 by saying:

‘There were 5 exabytes of information created between the dawn of civilization through 2003, but that much information is now created every two days.’

Although some criticised this generalisation at the time, today's reality far surpasses it.

We generate around 402.7 exabytes of data daily — and it’s still not enough.

As artificial intelligence (AI) systems grow larger and more sophisticated, they're gulping down data at an alarming rate. And the well is running dry.

Until recently, the well has been the world’s great ‘data commons’—the Internet. However, these models already incorporate most of the freely available information.

The next place they are looking should be in your portfolio.

The Insatiable Beast and its Leash

A study released by research group Epoch AI projects that we will exhaust public training data for AI sometime between 2026 and 2032.

That’s not just because AI’s appetite is growing. We’ve also seen a rapid rise of a much more guarded internet.

Many industries and websites now enforce AI-specific restrictions on web crawlers. In the past year, around 25% of the internet’s high-quality data was restricted from AI.

That’s according to researchers at the Data Provenance Initiative. Their recent paper’s title, ‘Consent in Crisis: The Rapid Decline of the AI Data Commons,’ tells you all you need to know.

Here’s a chart from that paper showing the steady rise of those restrictions in orange and red:

Despite the short history shown, the trend is clear. So, while one potential future is an internet tapped out of useful information, another is where that info is locked away.

Now, you could dismiss this as something you don’t have to worry about. But for forward-thinking investors, this problem could be an opportunity.

As the internet’s available data supply dwindles, private companies are becoming the next target for big tech’s AI.

Investors should consider future-proofing their portfolios with companies that have or create quality data.

But where to look?

The New Data Barons

While this wave of restrictions grows, AI companies haven’t sat idly by. Major licensing deals with AI companies have become the norm.

In May, OpenAI paid News Corp (ASX: NWS) $250 million in ‘cash and credits’ to access its content for five years. But that’s not the only deal it’s struck recently:

In this data-parched landscape, a new breed of winners is emerging. That’s companies with high-quality data.

These aren't just traditional tech giants. Many companies are sitting on goldmines of proprietary data, often in industries you'd least expect.

Here are some ASX companies that could fit this bill:

Take Life360 (ASX: 360), for instance. This family tracking app isn't just a place for tracking the location of your teenagers; it's a vast repository of human behaviour, preferences, and trends.

Driving habits, retail traffic, and daily movements are already tracked and sold to partners. One such deal, with Placer.ai, provides retailers with unparalleled insights into consumer habits.

Or consider Kogan (ASX: KGN), the e-commerce retailer. It's not just selling products; it's amassing an unparalleled dataset on consumer behaviour. Every search, purchase, and abandoned cart tells a story that AI can turn into actionable insights.

In the financial sector, Commonwealth Bank (ASX: CBA) is sitting on a treasure trove of transaction data. This isn't just about knowing how much money is moving around — it's about understanding the pulse of the economy in real time, a power that AI can amplify.

Healthcare is another frontier where data has become king. Companies like Pro Medicus (ASX: PME) aren’t just helping patients; it's collecting invaluable medical data that could revolutionise AI diagnostics.

It’s already signing deals with partners around the globe who are using AI to improve clinical outcomes and its own revenues.

Don't overlook the industrial sector. engineering support company Worley (ASX: WOR) is no slouch. In its most recent earnings report, CEO Chris Ashton singled out AI as a top priority.

It has already expanded its Mumbai office, employing around 200 data scientists. This team is exploring 200 potential AI applications, from automated tendering to improved digital twins.

These twins are virtual replicas of facilities or assets that help engineers test and optimise products before they are built.

They also work with the ‘Industrial Internet of Things’ (IIOT), where thousands of sensors capture real-time data and build live virtual models.

All this data can be fed into AI systems that can optimise supply chains, predict equipment failures and boost manufacturing efficiency.

These are just a few examples of companies with very different data sources. But they all share a commonality — they all hold or make high-quality data.

That could mean that their future profits could be boosted through similar licensing deals. Or, if the AI bull case becomes a reality, they could leverage their own AI to boost profits.

But before you invest in every company that holds ‘data’, let's consider what makes it quality.

The Quality Conundrum

Data may be the new oil, but not all data holds the same value.

Predictions are that by 2030, between 50–90% of the internet’s textual data could be AI-generated.

Today, OpenAI alone reportedly generates 100 billion words per day. That’s 36.5 trillion words a year!

This could be a massive problem if we expect the next generation of AI to try to hold up-to-date information.

With so much potential rubbish data, finding high-quality sources will be the difference between good and bad AI models. And good and bad investments.

We don't have to guess whether these companies will use the data themselves in the future or share it with AI companies. Simply having it gives them flexibility.

Forrester, the research firm, advises that ‘Data quality is now the primary factor limiting Gen AI adoption [in businesses]’.

So, what makes data ‘high-quality’?

There are too many factors to highlight here as data comes in many ways, but the broad strokes to look for are:

• Accuracy and consistency
• Thorough and up-to-date
• Unique and trustworthy
• Decent labelling or identifiers
• Diversity to prevent bias
• Ethically sourced and privacy-compliant

The companies that will truly strike gold in this future aren't just those with the most data — they're the ones with the cleanest, fullest, and most diverse data.

It's not about quantity - it's about quality.

As if these technical challenges weren't enough, we're also navigating an ethical minefield. The European Parliament has already passed rigorous regulations countries like Australia are looking to.

These restrict the more Orwellian types of data collection, including face, emotions and social scoring. They also require AI companies to disclose their data sources.

Thankfully, a modicum of our privacy will remain intact. But this isn't just red tape — it's a fundamental shift that will make ethically sourced, high-quality data even more valuable.

Use healthcare, for example. Many medical service companies may appear to have valuable data, but much of that could be restricted due to patient confidentiality. Biomedical companies, however, could be a better target for investors.

The Investment Frontier in a New Cycle

For forward-thinking investors, this quality data drought spells opportunity.

Now, I'm not arguing that all your investments need to follow this theme. But next time you consider a stock for investment, add ‘quality data’ to your checklist.

This will help strengthen your portfolio from future disruptions and capture the upside of any AI growth.

Regardless of your views on AI, this investment mindset is worth considering.

In prior technological cycles, massive disruptions occurred every 50–60 years. A large explosion of effort and energy went into the change. New businesses prospered while older ways collapsed.

However, there is also a typical early cycle of overinvestment and bubble-like behaviour as investors expect outsized returns.

Using a data-themed investment could allow you to diversify from expensive AI stocks while capturing the potential upside of an AI future.

If AI continues at this pace of progress, it's not difficult to imagine a world in which labour and capital have fundamentally changed.

However, the next decade’s leaders may not be those building AI models but those controlling the data behind them.

Imagine getting in on the ground floor of Standard Oil just as the automobile was taking off.

That's the kind of opportunity we're looking at with these data-rich companies. They're not just participants in the AI revolution. They're the landowners on which this future will be built.

Of course, predictions about the future are often wrong.

Regardless of whether AI sparks the next industrial revolution, the shift to an information-driven economy is undeniable.

The right path is to be open-minded and invest in this future with a healthy dose of scepticism. A data approach can give you the best of both worlds.

If you want to read more articles like this, sign up at Fat Tail Investment Research to see more market-leading topics on gold, commodities, and macro trends.