The trillion-dollar gamble
Every market cycle has its defining narrative. In the late 1990s it was the internet. Before the Global Financial Crisis, it was structured credit. Today, it is artificial intelligence.
The challenge for investors is that the most dangerous investment bubbles are rarely built on fiction. They are usually built on powerful truths taken to excess. The internet transformed the global economy; credit expanded growth and consumption. Artificial intelligence will almost certainly reshape productivity, business models and economic output.
The question is not whether AI matters; it clearly does. The question is whether markets have become too optimistic about the returns that will ultimately flow from the enormous capital being deployed.
There are reasons for caution.
A familiar warning signal
Historically, major market peaks have often been characterised by a shift from companies buying-back their own shares to issuing new equity. When management teams decide it is attractive to sell stock to investors rather than repurchase it, it can be a powerful signal that valuations have become stretched. We saw this during the technology boom of 1999 and 2000 and again during parts of the post-pandemic investment frenzy. There are now early signs that this pattern is beginning to re-emerge among some of the largest technology companies.
At the same time, the scale of investment flowing into AI infrastructure is extraordinary. What began as hundreds of billions of dollars of annual spending is rapidly moving towards a potential trillion-dollar run rate. Technology companies are caught in a strategic arms race. Underinvest and risk irrelevance; overinvest and risk destroying shareholder value.
No executive wants to be remembered as the leader who missed the most important technological shift of a generation. Yet investors should not assume every dollar committed to data centres, semiconductor capacity and energy infrastructure will ultimately earn acceptable returns.
This is the critical distinction often lost in today's market enthusiasm. AI can be transformative for the economy while still producing disappointing outcomes for investors who pay too much for exposure.
Beyond Silicon Valley
The AI investment cycle is also far broader than many appreciate. Most attention remains focused on software businesses and semiconductor manufacturers, but the beneficiaries extend throughout the economy. Building the next generation of AI infrastructure requires enormous quantities of steel, copper, power generation, electricity networks and construction capacity.
The winners are not confined to Silicon Valley. They include industrial companies, commodity producers, infrastructure owners and energy providers across public and private markets.
To put this in perspective, the Sydney Harbour Bridge contains approximately 52,800 tonnes of steel and a one-gigawatt data centre contains over 200,000 tonnes of steel, Anthropic alone recently announced it was interested in 1.4 gigawatts of data centre capacity in Australia.
Risks extend into the broader economy
The opportunity is significant, but so is the risk.
The interconnected nature of this investment theme means a disappointment in the AI cycle could reverberate well beyond technology stocks. If returns fail to justify the unprecedented capital expenditure currently underway, the consequences could be felt across infrastructure, private markets, industrials, energy and credit. Investors increasingly view AI as a technology story. In reality, it has become an economy-wide investment phenomenon.
Importantly, this does not mean the broader economy is approaching recession. The United States remains remarkably resilient. AI investment is supporting employment, construction activity and productivity growth. Labour markets remain strong, and inflation pressures remain persistent. In such an environment, central banks have little reason to aggressively cut interest rates. A healthy economy and a vulnerable share market are not mutually exclusive outcomes.
Indeed, one of the more plausible scenarios for investors is that economic growth remains reasonably strong while equity valuations come under pressure. Markets and economies do not always move together. The real economy may continue benefiting from AI-driven productivity gains even as investors reassess the prices they are willing to pay for future earnings.
The difficulty of diversification
Adding to the challenge is the declining effectiveness of traditional diversification. Correlations between asset classes have become less stable, making portfolio construction increasingly difficult. Government bonds have not always provided the protection investors once expected, while alternative assets face capacity constraints and accessibility challenges. The old playbook of simply allocating across shares and bonds may no longer offer the same level of resilience.
For Australian investors, the backdrop is even more complex. Domestic growth is slowing, household budgets remain under pressure and fiscal flexibility is increasingly constrained. Meanwhile, global divergence is widening. The United States continues to outperform, Europe remains sluggish and China faces deep structural challenges.
The lesson investors keep forgetting
Against this backdrop, investors should resist the temptation to choose between blind optimism and outright pessimism. The AI revolution is real. The productivity gains could be profound. But every great technological transformation attracts excessive capital at some stage of the cycle.
The lesson from history is not that innovation fails. It is that investors often overpay for it.
The winners over the next decade will not be those who ignore artificial intelligence, nor those who chase every AI-related investment. They will be those who recognise that extraordinary opportunities and extraordinary risks often emerge together. The future may be bright, but even the brightest future can become overpriced.
Michael Turner is a Director, Principal, and Investment Advisor at Hamilton Wealth Partners. This article contains general financial information only. It has been prepared without taking into account your personal objectives, financial situation or particular needs.