The body that manages tens of billions of dollars on behalf of the New South Wales government has a problem most investors would happily take on: its bets on artificial intelligence worked a little too well. TCorp, the state’s central financing and investment agency, has told the Australian Financial Review that it plans to push further into foreign markets after a strong run of returns, while conceding that those returns leaned heavily on an AI concentration it now wants to wind back.
It is an unusually candid admission for a public sector money manager, and it lands at a delicate moment for markets worldwide. For much of the past two years, the performance of the biggest global equity indices has been inseparable from the fortunes of a small cluster of American technology giants building the infrastructure of the AI era. TCorp, like almost every large institutional investor on the planet, was carried along by that wave. The question it is now asking itself is whether holding so much of its fate in one theme is prudent, even when the theme keeps paying off.
Why a state money manager is thinking about AI at all
TCorp is not a household name, but it sits at the centre of the NSW public finances. It manages investment portfolios for the state’s insurance and compensation schemes, its superannuation liabilities and a range of government agencies, and it raises money in the debt markets on the state’s behalf. When TCorp does well, the pressure on the NSW budget eases, because strong returns help cover the long-term cost of workers’ compensation claims, public sector pensions and other obligations that stretch out over decades. When it does poorly, taxpayers ultimately carry more of the load.
That is why the agency’s exposure to the US technology trade matters beyond the dry language of asset allocation. A fund with global equity holdings in recent years has almost by default become a fund with a large AI position, because the companies driving the market higher are the ones designing the chips, running the data centres and selling the cloud capacity that the AI build-out depends on. Getting out of some of that exposure, as TCorp now says it intends to, means deliberately diversifying away from the very holdings that have been doing the heavy lifting.
The case for easing off
The logic behind trimming is straightforward enough. Concentration risk is one of the oldest hazards in investing, and a portfolio that owes an outsized share of its gains to a single sector is vulnerable if sentiment turns. Valuations across the AI complex have climbed to levels that make even enthusiastic investors nervous, and the debate about whether the market is in a bubble has moved from the fringes to the mainstream. Spreading capital across a wider set of foreign markets, including regions that have lagged the American rally, is a classic way to reduce the chance that one correction erases years of progress.
There is also a governance dimension for a public body. A privately run hedge fund can afford to ride a concentrated position and wear the consequences. A manager of state money has a lower tolerance for the kind of drawdown that would show up in a budget update and invite questions in parliament. Taking some chips off the table after a good run is, in that sense, exactly the conservative instinct the public would expect.
The case for staying in
Against that sits an equally serious argument, and it is one that has humbled plenty of cautious investors over the past decade. Timing an exit from a rising market is notoriously difficult, and those who concluded the technology trade was overdone in 2023 or 2024 missed much of the gain that followed. If the AI build-out is a genuine multi-year transformation rather than a short-lived mania, then reducing exposure now could mean leaving significant returns on the table at precisely the wrong moment.
Sceptics of the bubble thesis point out that, unlike the dot-com era, the leading AI companies are enormously profitable and are generating the cash to fund their own spending. The counter from the bears is that the spending has become circular, with chipmakers, cloud providers and model developers increasingly funding one another, and that the revenue to justify the capital expenditure has yet to fully materialise. TCorp’s decision to lighten rather than exit suggests it is trying to split the difference: keep a foot in the trade while refusing to let it dominate the portfolio.
What it means for Australia
TCorp is the NSW piece of a much larger national picture. Australia’s superannuation system, now worth more than four trillion dollars, is one of the most globally exposed pools of retirement savings in the world, and much of that money sits in international equities for the simple reason that the local sharemarket is small and heavily weighted towards banks and miners. That structure means the typical Australian worker’s retirement balance has quietly become a bet on American technology, whether or not they realise it. When a sophisticated institutional manager like TCorp says its AI exposure had grown larger than it was comfortable with, it is a signal worth heeding for the big industry and retail super funds making the same calculations.
The tension is familiar to anyone following the local market. Australian investors have been pouring money into AI-linked names and listed data centre operators, and the appetite for exposure to the theme has driven a run of capital raisings and new listings. At the same time, warnings about froth in the sector have grown louder, with commentary about the risk of an AI-driven correction rattling financial markets becoming a regular feature of the business pages. TCorp’s move does not resolve that debate, but it is a concrete example of a serious player acting on the concern rather than simply voicing it.
What’s next
The practical question is how far and how fast TCorp actually rebalances, and which markets it rotates into as it diversifies away from the concentrated US positions. A gradual, measured reduction would be consistent with the agency’s conservative mandate and would avoid the risk of crystallising losses by selling into a downturn. The broader signal, though, is the one that will reverberate through boardrooms and investment committees around the country. If one of the state’s most important financial institutions has decided that its AI winnings are now a risk to be managed rather than a bet to be pressed, others managing Australian money will be asking themselves whether they should be doing the same.
For now, TCorp is keeping its exposure to the US technology boom while quietly reducing its dependence on it. In a market where the line between conviction and complacency has rarely been thinner, that may prove to be the more interesting position of the two.
Sources: Australian Financial Review
