AI – FROM HYPE TO HANGOVER – ROGER MONTGOMERY

As discussion mounted around the effects on contemporary and future society of Artificial Intelligence (AI), The Sydney Institute invited Justin Tauber, General Manager, Agentic Technology, Trust & Adoption, Salesforce ANZ and Roger Montgomery – founder and Chairman of Montgomery Investment Management Pty Limited to give their different views of the influence and change to come with AI. Offering both positive and negative impacts, Justin Tauber and Roger Montgomery addressed The Sydney Institute on Wednesday 29 April 2026.

AI – FROM HYPE TO HANGOVER

ROGER MONTGOMERY

There are many lenses through which the arrival, scaling and adoption of AI tools can be seen. Let me begin as an investor.

For all its earth shattering, life transforming and human history-changing promise, AI is but another General-Purpose Technology (or GPT) in a long list of technologies throughout history that have inspired inventors, consumers and investors. These include the printing press, the steam engine, electricity, the railroad, the internal combustion engine, the computer, the internet and nanotechnology.

The characteristics of a GPT include pervasiveness and are used in many diGerent sectors. The technical performance improves over time and they make it easier to produce new technologies. Eventually, they bring about widespread productivity gains. I’ll describe how GPT booms develop and die, with reference to the current AI boom.

Before I do, let me tell you some of the promises that have been made by the proponents of AI. It has been claimed AI will help in the eradication of disease, personalised medicine, decoding the mechanism of aging and solve the problem of limitless energy.

Open AI’s Sam Altman has been quoted claiming AI will solve the 1) housing crisis, 2) cancer, 3) poverty, 4) climate change, 5) mental health, 6) democracy, 7) universal basic income, 8) a bunch of diseases, 9) heart disease, as well as 9) help you accomplish your goals and be your best, 10) provide very high quality healthcare, 11) make important new scientific discoveries, 12) bring the marginal cost of energy rapidly towards zero, and 13) create a more equal world, with its inhabitants enjoying, 14) universal extreme wealth.

The first requirement for a GPT boom/bubble is hype. This hype – or excitement about the new invention – drives super-exponential growth in share prices. The share prices of the likes of Nvidia have displayed this super exponential path. Nvidia’s market value now hovers around US$5t, and Alphabet is at $4.1t. The rapid, faster-than exponential appreciation is, for some investors, the very definition of a bubble.

The accelerating price, fueled by herding behaviour as followers hop on board, often out of a fear of missing out, creates a self-reinforcing cycle. As the same price behaviour is widespread currently, let’s tick that as a requirement met.

The next phase is for bubble participants to do deals with their overpriced shares. There are a couple of motivations for this. The first and more obvious one is that when your shares are expensive, the deals you do are cheap. And, secondly, the activity itself gives validity to the boom.

There has been a wave of deals done with shares, including OpenAI and AMD, OpenAI and Coreweave, and Nvidia and Intel. What might quite rightly worry investors is that the cash, the compute and the equity are all flowing in a loop with multiple players reporting the same wad of cash as their own revenue. The next stage is the existence of bubble participants, or “insiders” buying each other’s products in an eGort to demonstrate “faith” in the technology and in its structural status as history-making. We’ve seen that too.

The worry, however, is that while the pace of buying is accelerating, so too are the receivables. NVIDIA’s quarterly balance sheets reveal “receivables” – that is, sales on credit – are accelerating. And the rate of increase in receivables is faster than the growth in revenue. Therefore, a rising proportion of sales is on a promise.

Next, is the presence of vendor financing, which is used to sustain the appearance of growth. NVIDIA sells graphic processing units (GPUs) to a hyperscaler, who then pays cash to NVIDIA. But not really, since that cash is sent right back by NVIDIA to the hyperscalers in return for shares. So, ultimately, Nvidia is vendor financing its own sales in return for equity in its customers.

The law of large numbers ensures an attractive rate of growth is unsustainable. Indeed, Wall Street analysts estimate Nvidia’s revenue growth will be 31 per cent next year. Which is still impressive, but less than half this year’s projected 72 per cent growth.

Finally, you see a massive jump in capital expenditure and physical build-out to scale the new technology. This is all possible because there’s so much hype, around the tech, that share prices have surged, bringing down the cost of capital for the bubble participants. With that lower cost of capital, and after all the equity deals have been done, companies, in a final wave of “pretend to extend”, start borrowing money.

In the railroad boom of the 1800s, US railroad corporations issued bonds on everything – on their equipment, on their track, on their land grants from the federal government, and even on future profits. Today, the AI boom has triggered a massive debt spree among hyperscalers. In just seven weeks, late last year, Amazon, Google, Meta, Microsoft, and Oracle raised a staggering $US120 billion in bonds, suggesting cash flows are insuGicient to cover exploding capital expenditures, projected to reach $US534 billion next year and consume 80 per cent of the group’s forecasted cash flow. We know that because we’ve already seen Nvidia’s rising receivables (or lack of cash receipts). Most recently, debt is being raised with GPUs as collateral – which is something that should cause us to sit up and take notice.

CoreWeave’s total GPU-backed debt has exceeded US$10 billion, Fluidstack (now a part of CoreWeave) received approval for over US$10 billion in financing using its Nvidia GPUs as collateral, Lambda secured a $500 million loan backed by its existing Nvidia chips, and Crusoe (formerly Crusoe Energy) has taken on hundreds of millions in GPU-backed debt, including a US$200 million deal in late 2023 and another US$225 million last year.

The Railroad bubbles and busts of 1873 and 1893, each demonstrated unsustainable growth and speculative lending that weakened the financial system and were major factors in the subsequent economic depressions, partly because the railroads were built on mountains of debt. Today’s artificial intelligence (AI) buildout and data centre construction has, hitherto, been financed by the free cash flows and cash balances of the major US technology companies, and with private capital from Private Equity firms like Apollo and Blackstone.

But changes are afoot. Back in October, Bank of America Global Research noted: “borrowing to fund AI datacentre spending exploded in September and, so far, in October.” Historically, GPT booms have all evolved along similar lines. While AI may be the latest groundbreaking tech, the development of the associated bubble might be remembered for breaking bank accounts. This happens because the consequence of the build-out, the capital expenditure, the massive land grab is over capacity.

When over capacity occurs, investment returns disappoint. Disappointing returns transpire because even though “this technology-is-going-to-change-history”, it usually isn’t going to right now. The perception or assumption today is that the AI theme is structural and uninterruptable.

Few question whether the tech will change the course of human history. I believe it will. But I also believe that the suppliers of this tech must ultimately deal with customers whose demand isn’t structural – theirs is cyclical. When a structural theme meets a cyclical commerciality, it’s the cyclical commerciality that wins.

What’s next is a period of Creative Destruction. Those of you who remember the tech wreck of 2000 will recall that term – Creative Destruction. As price gains unwind, everyone other than the earliest investors loses between 60 and 95 per cent. Many are wiped out. Now, at this point, investors cry out – “But nothing’s changed”. Why are shares collapsing if the theme remains true? What’s happened is a reality check – the time frames are wrong. As the share prices plunge and volatility increases, the cost of capital rises and all that debt that was amassed at the end of the boom sends some companies to the wall. That’s what’s changed. For bulls and optimists, artificial intelligence is seen as a productivity booster that will usher in a period not unlike the Roaring 1920s.

That narrative is rapidly morphing into an existential threat, and the stock market is beginning to price in the possibility that AI becomes a monster that turns on the world. Anyone who’s watched any of the short movies created by AI agents in the last two months, with just two or three prompts, must wonder what happens to the thousands of VFX experts who typically dominate old-school movie credits. They’re the first victims, followed by the white-collar workers in information technology, where payroll employment has remained flat since the release of ChatGPT in late 2022 and where companies such as Spotify have disclosed their senior engineers haven’t written a single line of code since last year.

Microsoft’s head of AI, Mustafa Suleyman, predicts that white collar work will be fully automated within 12-18 months. If high-earning professionals are forced into lower-paying roles or into unemployment, the resulting hit to consumer spending would trigger a broader economic crisis. Maybe that explains Louis Vuitton’s 30 per cent share price crash, Kering’s 70 per cent fall and Estée Lauder’s 81 per cent plunge. Unemployed (or anxious) high-earners don’t buy $50,000 handbags or even $5000 handbags.

The “wealth eGect” is reversing because the AI benefits aren’t evenly distributed. Richard White wrote in his epic history of US railroad companies – entitled The Transcontinentals and the Making of Modern America – “They created modernity [but they also left behind] a legacy of bankruptcies, two depressions, environmental harm, financial crises, and social upheaval.”

It might be worth remembering the words of journalist Derek Thompson, who wrote in his November newsletter last year, AI Could Be the Railroad of the 21st Century. Brace Yourself, “Memories are short, and prudence and natural risk aversion are no match for the dream of getting rich on the back of a revolutionary technology that ‘everyone knows’ will change the world.”

The global railway mania of the 1800s bankrupted thousands of investors, wiped out hundreds of companies, but left nations with a rail network that powered a century of industrial dominance. Similarly, when the fibre-optics/internet/broadband boom of 1999 crashed in early 2000, it vaporised $US5 trillion of market value, but it also laid the wiring on which the Internet Age began.

Similar stores can be told about the electricity bubble, aviation, broadcast radio and automobiles. Indeed, about three-quarters of General-Purpose-Technology (GPT) booms share a similar pattern: colossal over-investment, financial carnage through creative destruction, then lower prices and decades of productivity gains built on the residual infrastructure.

At the core of today’s AI boom are General Purpose Technologies (GPTs) – on the scale of electricity or the internet – and the infrastructure being built will support decades of economic activity. And unlike the Dot.Com bubble, the investment isn’t nine billions in speculative debt being loaned to “pre-revenue” companies with no path to profitability. This time, the investment is coming (so far) mostly from companies with strong balance sheets. But beyond that core, the classic signs of every bubble are on display.

It would be unusual, given the great enthusiasm and hype, the high asset prices, the overbuilding, the uncertainty of future demand, indeed the many elements that conform to past bubbles, if this boom doesn’t also conform to the pattern. It would indeed be a first. And the most dangerous four words in investing are: This time is DiGerent.