Australia's AI exposure lives outside the ASX 20

An adviser who wants Australian exposure to artificial intelligence hits a wall at the top of the index. Banks, resources, property, supermarkets and utilities make up around 68 per cent of the ASX 200. Information technology accounts for roughly 3 per cent, against about a third of the S&P 500.

Jack Briggs and James Barker, portfolio managers on the Ellerston Australian Emerging Leaders Strategy, argue the exposure does exist on the ASX. It simply does not live where most portfolios go looking for it.

By Briggs’ reckoning, neither the infrastructure builders nor the AI adopters the strategy holds rank among the top 20 stocks by market capitalisation. Index exposure alone will not reach them.

That argument carries more weight after the June quarter. The strategy returned 17.1 per cent net for the three months, outpacing the S&P/ASX Small Ordinaries Accumulation Index by 13.8 percentage points and the Small Industrials Index by 8.7 percentage points, according to Ellerston.

Leadership moved, and the reason matters

The more interesting number belongs to the market rather than the manager. Small Industrials beat the Small Ordinaries by 5.1 percentage points over the quarter alone, reversing a year in which resources set the pace.

“That rotation matters because it suggests the market is turning back towards fundamentals rather than commodity price momentum,” Barker says.

Briggs points to one theme running through the investment universe: the build out of AI infrastructure and electrification. Contractors doing that work have visibility their sector rarely enjoys.

“Order books at electrical services and data centre contractors are now extending into 2028 and 2029,” he says. “That’s a level of forward visibility these businesses have rarely had.”

A productivity problem in search of a circuit breaker

Underneath the thesis lies a bleaker macro picture. Australia has recorded its first negative decade of productivity growth on record, averaging -0.2 per cent a year across FY21 to FY25, with FY25 alone down 0.7 per cent. Real income per person has barely moved in six years.

Barker frames the consequence in company terms.

“When output per hour worked is flat, a company can only grow revenue by employing more people,” he says. “Costs rise in step with sales, margins compress and growth becomes something a business has to buy rather than something it generates.”

His argument follows from there. Artificial intelligence offers the most credible circuit breaker available, and smaller companies are structurally better placed to capture the benefit than large incumbents.

Adoption remains early. Around 12 per cent of Australian businesses currently use AI, which Briggs treats as headroom rather than a shortcoming.

“The bulk of the productivity gain has yet to be captured. And because smaller companies don’t carry the legacy systems and restructuring drag that slow larger businesses down, the margin gain from AI adoption falls disproportionately to them,” he says.

Two legs Australia can own

The most original part of the Ellerston case concerns where Australia fits in the global cycle. Australia will not produce the frontier models or the hyperscale platforms. Barker accepts that, then argues it settles far less than investors assume.

“This is a rare case where Australia is not simply a price taker in a global technology cycle. We won’t own the platforms, but we do own the two legs that follow; the build out itself and the productivity gain from adoption.”

On the build out leg, the strategy holds Southern Cross Electrical Engineering and GenusPlus Group, with SKS Technologies and Mayfield Group in its wider coverage universe.

On the adoption leg, Briggs nominates holdings such as Vista Group, a software business built on proprietary data and embedded workflows that the market, in his view, “wrongly assumes generic AI models can replicate”.

“The companies that convert AI adoption into operating leverage will simply grow faster than the economy around them,” Briggs says.

The research argument, and its limits

Ellerston targets what it calls Australia’s emerging leaders, roughly 757 listed companies with market capitalisations between $50 million and $2.5 billion. Many carry little or no broker coverage, which the managers treat as the source of their edge.

Over the past 20 years, they note, top-quartile small cap managers have delivered approximately 4.7 per cent per annum of alpha and stayed positive at the one, three, five, 10 and 20-year horizons. Top-quartile large cap managers have not matched that record against the ASX 300 at any horizon.

“That reflects a structural inefficiency that’s best captured through deep fundamental research,” Barker says.

Advisers should weigh that carefully. Top-quartile figures describe the winners and say nothing about what the median manager in the segment returned, so manager selection carries the whole argument.

The valuation case has thinned too. Australian micro and small caps returned 28.6 per cent in FY26 on Ellerston’s figures, ahead of the S&P 500 and in line with the Nasdaq. The ASX 200 managed 6.1 per cent over the same year.

Large caps now trade on roughly 21 times forward earnings for around 11 per cent growth, which keeps the relative argument intact. Buyers arriving today still pay for a re-rating that has already run hard.

None of that settles the question of exposure. An adviser can decide the small cap risk premium is not worth paying and still face the structural point Ellerston raises: a portfolio built around the ASX 200 owns the industries of the last boom, and very little of whatever this one turns out to be.

The order books run to 2029. Whether earnings arrive on the same timetable is the part no index can answer.