The Australian – Beyond the war: Why AI and U.S. debt are key threats for investors
While the Middle East commands headlines, investors have largely forgotten the factors determining their returns prior to the outbreak of hostilities. But when the conflict ends, investors will return to considering those factors, including artificial intelligence (AI), U.S. debt, and the possibility of stagflation.
Prior to the conflict, investors were debating AI’s immediate and long-term impact. While 2025 was about the rise of the AI “picks and shovels” – enablers like Nvidia – 2026 witnessed the emergence of agentic AI, and the narrative quickly became about the fall of the middlemen – the traditional software companies that built epochal and capital-light business on a per-seat revenue model.
This article was first published in The Australian on 25 March 2026.
It’s worth remembering that during the first 10 or so weeks in 2026, US$1 trillion ($1.4 trillion) in market cap was wiped from software stocks as investors feared that an AI agent replacing the work of 10 people would mean an enterprise customer didn’t need 10 software licences.
The emergence of agentic AI also prompted a debate about the wider economic implications of an unemployed generation. The pessimistic view leans heavily on data from late 2025 that revealed firms in AI-exposed sectors – finance, legal and media – had already recorded a 4.5 per cent reduction in total headcount. Their fear is AI will cannibalise junior roles, potentially creating a lost generation of professionals who fail to gain the entry-level experience needed to become senior leaders.
Atlassian cut 10 per cent of its workforce, or 1600 jobs, on March 12; Block cut 40 per cent, or 4000 jobs; Meta eliminated 16,000 roles, or 20 per cent of its workforce; Amazon 16,000; Oracle 25,000; and WiseTech 2000 jobs. These people, who represent the tip of a possible impending iceberg, may or may not find work elsewhere.
Optimists, however, point to productivity gains from automating mundane tasks that currently involve trillions of manual mouse clicks. Analysts at Goldman Sachs and the IMF argue that AI will boost global Gross Domestic Product (GDP) by 0.8 per cent to 1 per cent annually as AI shifts from data centres into operational workflows. They contend that while some jobs disappear, new ones – like “agentic workflow managers” – are already enjoying a 30 per cent wage premium.
History, however, shows that reskilling takes time, and many who lose their jobs never reskill, and are left behind.
History also shows the path to productivity gains from new tech is disruptive and destabilising.
The shift from an agrarian society to an industrial one was arguably the first great de-skilling event, which took 40 to 80 years and resulted in massive spikes in localised poverty as cottage industries collapsed. During Engels’ Pause – the period from roughly 1790 to 1840 in Britain – rapid technological advancements and industrial growth did not lead to increased wages for workers.
During the 1920s, when Henry Ford and Frederick Taylor introduced scientific management (Taylorism) to maximise industrial efficiency through worker specialisation and time studies, productivity soared by more than 40 per cent, but the era also witnessed the unemployment of skilled artisans and a transition so jarring it’s said to have contributed to the social instability of the Great Depression, where unemployment famously hit 25 per cent in the US.
More recently, in the 1960s and accelerating through the 1990s, computers triggered another great employment migration, this time from factory floors to office cubicles as physical labour shifted to knowledge work.
Even this transition took about 30 years and was characterised by an observation from Nobel Prize recipient and MIT economist Robert Solow: “You can see the computer age everywhere but in the productivity statistics.”
As manufacturing moved to automation (and overseas), rust belts formed in the U.S. and elsewhere. The people losing factory jobs weren’t the people being hired for new IT jobs, so the transition produced a permanent skills gap that lasted a generation. And between 1970 and the early 1980s, U.S. unemployment rose from 4.6 per cent to just shy of 11 per cent. Meanwhile, the Dow Jones did not rise for the decade between 1970 and 1980.
Beyond AI, another serious issue is the burgeoning U.S. debt.
It’s a little-known fact that the U.S. state of Wyoming reportedly bought 2312 ounces of gold in December after passing a law requiring the state’s investment portfolio to add precious metals as a hedge against economic turmoil.
Wyoming is worried about rising federal debt, inflation and a weak US dollar, prompting Republican state senator Bob Ide to sponsor the Wyoming Gold Act.
“I can’t put a timeline on it, but there’s gonna be a sovereign-debt crisis,” Ide said, adding, “There’s no will to rein in spending.”
In early 2026, the U.S. national debt held by the public had reached approximately 101 per cent of GDP, and annual interest payments had surged to nearly US$1 trillion, now rivalling or exceeding the entire U.S. defence budget.
A sovereign debt crisis occurs when investors lose confidence in a government’s ability to repay its debt, leading to a spike in interest rates and a potential default. For the U.S., there are arguably three catalysts.
The first is the interest rate on the debt consistently exceeding the economy’s growth rate. When interest rates exceed economic growth, the debt grows faster than the tax base used to pay it, and even if the government stops all new spending, the debt snowballs automatically.
The second is a structural increase in entitlement spending as Baby Boomers continue to retire. Without tax increases or benefit reforms, the generational avalanche creates a permanent, widening deficit.
Finally, in 2025, U.S. gold exports more than doubled to more than US$84 billion, accounting for nearly 90 per cent of the total increase in U.S. exports in some months.
Much of this gold is being routed through Switzerland before ending up in China.
It’s important to appreciate that China is buying gold as it aggressively sells US Treasuries. China’s holdings of US debt fell to about US$682 billion in early 2026, the lowest level since 2008. It appears China is moving toward a “value-for-value” settlement system, where, by holding gold instead of dollars, it insulates itself from U.S. sanctions and the weaponisation of the dollar-based (SWIFT) financial system. Could China be preparing for war? Why else would it be worried about sanctions?
For decades, China was a price-inelastic buyer of U.S. debt, effectively funding U.S. deficits. But if China (and others like India, Poland and Brazil) continues to swap Treasuries for gold, the U.S. must find new buyers for its debt. And with a growing debt pile, new buyers of scale are increasingly hard to find.
To attract new buyers to replace the Chinese central bank, the U.S. could be forced to offer higher interest rates. On a US$38 trillion debt load, however, even a 1 per cent increase in rates would add nearly US$400 billion to the annual deficit, accelerating the debt.
If the trend continues, the risk of a failed Treasury auction of U.S. bonds – where the U.S. government tries to borrow money and finds insufficient demand – increases. This would be the definitive starting point of a sovereign debt crisis.
While all eyes are currently on the Middle East, savvy investors will be thinking about what lies around the corner when the war ends. It appears a prolonged conflict is not ideal, but given the worries beyond the conflict, a short war may not be particularly appealing to investors either.
This article was first published in The Australian on 25 March 2026.