Navigating the AI boom: Who we think is best placed to win

Artificial intelligence is highly likely to change the world as we know it, in ways that we probably will not anticipate. 

The same can be said for many developments that have occurred over the last few hundred years. The steam engine and the railways that followed it collapsed the cost of moving goods and people, turning sprawling nations into connected economies within a generation.

Electrification in the late 19th century freed factories from waterwheels and on-site steam plants and eventually enabled the modern assembly line. The telephone, the internal combustion engine and the motor car, radio and television, the mainframe and personal computer, and then the internet and the smartphone each did something similar in their own era. 

Each development changed how people worked, communicated, shopped and were entertained, and in doing so remade entire industries and reordered which companies sat at the top of them. AI looks set to take its place in that lineage. 

The question is, does being right about a technology's importance mean being right about whether to invest in it? 

We do not think there is much debate to be had about whether AI will be transformative. The evidence to date and the sheer scale of capital being committed to it both point in this direction. 

But it is important to remember, while the nature of the technological change is different, the behaviour of investors during such periods is often similar and can be studied. Significant change and disruption have accompanied many major technological developments through time. 

However, investing in new technological developments has not necessarily been a clear winning strategy. And the reason for this is fairly straightforward. Exciting developments attract attention and capital. 

When capital floods into a new space, absent any specific and enduring economic moat for an individual company or group of companies, the result is intense competition. And intense competition diminishes the economic rent that can be extracted by the companies competing. 

What we know about AI's impact so far 

The evidence of AI's impact to date is real but mixed. Large language models have demonstrably improved productivity in software development, customer service, content generation and parts of professional services, and adoption of these tools across businesses and consumers is happening rapidly.

At the same time, the capital being committed to build AI infrastructure has reached a scale with few historical parallels. Goldman Sachs estimates the capex of the major hyperscalers (Alphabet, Amazon, Meta, Microsoft and Oracle) is circa US$800 billion in 2026, with consensus expecting this to grow to north of US$1 trillion in 2027.

Monetisation is running well behind that spend. Goldman Sachs has recently estimated that for the large hyperscalers to break even on their 2026 and 2027 AI-related capex, they would need to achieve annual AI revenues of approximately $300 billion in the next few years. 

While it is unclear what the hyperscalers’ aggregate AI revenues are currently, Goldmans Sachs calculated that in Q2 2026 these businesses’ combined cloud revenues were annualising approximately $70 billion above the pre-AI trend in 2024. 

The current spending race appears to us to be akin to a prisoner’s dilemma of sorts for the largest technological companies. Each company can see the potentially transformational benefits of AI developments, and realises that if they fall behind it could lead to supernormal profits for a clear market leader, which would in time hurt their existing business. 

As a result the bigger risk may well appear to be not investing and potentially facing redundancy in time. 

Investors are naturally drawn to technologies that are obviously going to change the world. AI's long-run impact on productivity and corporate earnings is likely to be large, but the timing and distribution of that impact is highly uncertain, as it has been with new technologies previously.

 The excitement of untold possibilities, and the fear of being left behind if the technology turns out to be as important as it looks, is precisely what draws in such extraordinary amounts of capital.

But by definition, the intense competition that results often leads to disappointing returns for investors in aggregate. 

History suggests the scale and obviousness of the opportunity is what draws in far more capital than the eventual market can profitably support, and that oversupply of capital competes away the returns on it, regardless of how much value the technology goes on to create for the economy as a whole. 

There are a number of historical precedents. 

• Britain's Railway Mania of the 1840s saw approximately £220 million raised through railway share issues to fund hundreds of new railway companies as track mileage quadrupled from around 1,500 miles in 1837 to over 6,000 miles by 1847. By 1845, capital expenditure on railroads accounted for 7% of the United Kingdom’s Gross Domestic Product (GDP). More than 200 of those companies subsequently failed or were absorbed into competitors, and many investors lost substantial sums. The railways themselves went on to permanently transform the British economy. However, investors who financed that transformation, on average, did not do well out of it. 

• The same pattern repeated with the build-out of the internet in the late 1990s. More than US$500 billion was invested in telecommunications infrastructure in the five years following the US Telecommunications Act of 1996, much of it debt-funded, as companies raced to lay fibre and build networks ahead of demand. By the end of the subsequent downturn, telecom and internet-related stocks had lost around US$5 trillion in combined market value from their peak, and bondholders of failed telecommunications companies recovered only a little over 20 cents in the dollar. The Nasdaq, having risen roughly 600% between 1995 and its March 2000 peak, fell 78% over the following two and a half years. The internet, like the railways before it, went on to change the world. Yet most of the capital that built it was destroyed.

In these examples the technology succeeded. It is precisely when a technology's potential becomes obvious to everyone at once that capital floods toward anything associated with it.

The result is that new entrants proliferate, incumbents overinvest for fear of being left behind, and this is enabled because financing is cheap and abundant. The resulting oversupply of capital relative to the eventual size of the prize competes away the economic returns on that capital, even as the technology itself goes on to reshape the world exactly as advertised.

A crowded, extremely well-funded race  

If overcapitalisation has been the recurring reason past booms disappointed investors, the AI boom currently underway bears a lot of similarities. 

On one side sit the hyperscalers: Microsoft, Alphabet and Amazon are each guiding to US$175– 220 billion of capex in 2026 alone, the large majority of which relates to AI infrastructure, with Meta a little behind at US$130–145 billion. 

On the other side sit the dedicated frontier labs, including OpenAI, Anthropic, xAI and others, which are themselves raising and spending capital at a scale that would have been unthinkable for a private company a decade ago. Anthropic and OpenAI have reportedly been pursuing financing that would value both companies at over a trillion dollars, against revenue that is a small fraction of that figure. 

It appears at present that model capability across the leading labs continues to converge rather than diverge (termed the Platonic Representation Hypothesis), computing capacity is increasingly available to rent rather than being a genuinely scarce input controlled by one player, and for a large share of commercial AI applications, switching costs between providers remain low. 

In other words, several extremely well-capitalised competitors are racing for the same prize without any of them having yet secured a clearly durable structural advantage. Such a dynamic has historically been associated with poor returns on the capital invested. We would not be surprised if several of today's AI leaders turn out, in hindsight, to be extraordinary businesses. However, this might not be enough to make them good investments given the valuations being attributed to them. 

Limited, and mostly indirect, exposure on the ASX 

Investors have been eager to get exposure to the thematic. In a domestic setting, the Australian market offers surprisingly little in the way of genuine, direct exposure to AI. There is no exposure to frontier labs or hyperscalers. 

The real centre of gravity for AI, on both the model-development side and the infrastructure side, sits almost entirely offshore. Some frontier companies, such as OpenAI and Anthropic, are not yet publicly listed. There is scepticism about the economic moat surrounding the neocloud and data centre operators. 

Given the pace of technological development, it is unclear what the economics of these businesses will look like in only a few short years, making a long-term investment decision difficult. 

Many investors are looking towards the so called “picks and shovels” of the AI boom and the associated infrastructure spend. The most obvious beneficiaries are the chip designers and manufacturers powering the enormous compute requirements of the frontier labs. 

Irrespective of questions of the durability of the elevated margins these businesses are currently experiencing while demand for chips greatly exceeds supply, none of the major players are listed in Australia. 

The resulting list of indirect ASX-listed exposures to the AI boom are development and construction companies associated with data centre and neocloud development, the energy providers that will see incremental demand from these energy intensive sites, and the commodity producers that are feeding in critical elements for the electrification needs and production of the different components of the AI boom. 

In each case, the exposure is a step or two removed from the actual technology. 

We are wary of capitalising a short-term earnings boost, even one lasting several years, into a valuation into perpetuity. 

This is particularly so when the company is obliged to deliver on projects that often dwarf any past undertaking for the business. It is important to remember that significant contracts also carry significant risks.

The impact from the build out of AI associated infrastructure on commodity demand is real but requires context. Copper is a good example. BHP has once again become the largest listed company on the ASX. Our perception is that this is because of the hype around the growth in copper demand as a result of AI. BHP is trading far more like a copper company than an iron ore company, despite iron ore constituting more than half of group EBITDA and cash flow in FY26, after adjusting for minority interests. 

The performance of copper and iron ore have diverged over recent years.

 

BHP has been trading in line with the copper price, despite the continued significant contribution from iron ore. BHP has experienced multiple expansion representative of share price growth outstripping earnings growth.

 

Yet as shown in the chart below, AI and data centre demand is on track to account for only a little over 4% of global copper demand in 2026.

 

Further, while the copper price has been strong throughout 2026, and this has occurred at the same time that AI development has been rapid, the correlation may well be coincidental, rather than causal. 

There has been heavy speculation that the US will enact tariffs on imported copper from the start of 2027. This has seen elevated purchasing by US companies resulting in rapidly rising stockpiles. In fact, US stockpiles are up more than tenfold in the last 18 months to levels not seen previously. 

There is a reasonable chance that at some point the stockpiling will cease and there may be a draw on inventories. Historically, a decline in inventories has been associated with weakness in the copper price. There is a chance that in fact the recent copper trade is a tariff trade disguised as an AI trade. For this reason we are cautious on associated exposures.

 

Could the real winners be the fast adopters, rather than the inventors? 

If history suggests investors should be wary of the companies racing to build a new technology, it also offers a useful, if less intuitive, guide to where value frequently ends up instead: the businesses, often in entirely unrelated industries, that are quickest and most capable at putting the new technology to work. 

Electrification illustrates this point. Factories began installing electric motors from the 1890s, yet economy-wide productivity did not visibly accelerate until the 1920s and 1930s. 

The delay reflected how long it took for supporting infrastructure, safe wiring standards, trained electricians and insurance practices for electrical risk, to mature enough for factories to redesign their entire layout around electric power, rather than simply swapping a steam engine for a single large electric motor. 

Once that infrastructure existed, the factories that reorganised fastest around the new technology captured the resulting efficiency gains, not the original electrical equipment makers and generating companies that had financed the technology's early development. 

The internet tells a similar story. 

Many of the companies that physically built the internet's backbone including laying terrestrial and undersea fibre optic networks, including WorldCom, Global Crossing and NorthPoint Communications, went bankrupt, taking most of their investors' capital with them. The vast oversupply of fibre and bandwidth they left behind, however, made data transmission and computing extraordinarily cheap for everyone who came after. 

The businesses that captured the greatest and most durable share of the value over the following two decades were not, in the main, the infrastructure builders themselves, but companies that were fastest to use that now-cheap infrastructure and the data it generated to change how they reached customers and ran their operations. 

It is often difficult during periods of rapid change to identify, in advance, which companies will ultimately benefit most from a transformative new technology. But more often than not, it has been the fastest and most capable adopters of a technology, spread across many existing industries, rather than the original inventors or developers of the technology itself, that have captured the largest and most lasting share of the value created. 

Already putting AI to work across the portfolio 

We think the distinction between those building the technology and those adopting it is a useful lens for thinking about how to position for AI today. 

The Auscap Funds have significant exposure to companies that have demonstrated they are rapidly adopting AI to improve their operations, accelerate developments and leverage their existing workforce in a more effective manner, which we expect will deliver revenue growth and margin gains in the future. 

CAR Group's (ASX: CAR) online marketplace model depends on a network effect between buyers and sellers that has taken years to build and provides a strong economic moat around the business. 

They are rapidly adopting AI, with an innovation hub in Brazil focused on rolling out features to improve the customer experience and accelerate product development. AI-enabled features such as AI agent enabled search, one-touch vehicle listings and automatically generated walkarounds are examples they have provided already of features enabled by AI. 

AUB Group (ASX: AUB) is using AI to automate routine back-office insurance broking work and pre-check policies prior to finalisation, freeing its brokers' time to focus on winning and servicing business rather than paperwork. 

We continue to think there is an important role for a broker to play as an intermediary between the insurer and the client, a relationship that may become increasingly important in a more complicated insurance environment.

Resmed (ASX: RMD), already the dominant player in sleep apnea devices, is implementing AI-enabled solutions. 

Its Smart Comfort feature, trained on over 100 million nights of de-identified sleep data, was the first FDA-cleared AI-enabled medical device that recommends personalised comfort settings to help people with obstructive sleep apnea use CPAP treatment most effectively. It has also developed a virtual assistant called Dawn which offers round the clock support and integrates with products like Apple Health. 

None of these businesses are selling AI as a product. They are using AI as a tool to reinforce their advantages, which often lie in the data, networks, customer relationships and distribution they have that are difficult to replicate. 

It may be that the leaders in each field are best equipped, if they are responsive to change, to enhance their moat further by investing in the benefits that AI can provide to their business.

 We remain focused on ensuring that the businesses we are invested in are responsive to developments in AI and are looking to utilise it to improve their operations. 

The path ahead 

We think AI is likely to prove a hugely transformative technological development. However, history suggests that being right about a technology's importance has rarely been enough, on its own, to earn an adequate investment return because the scale of the opportunity itself attracts far more capital than the eventual prize can support. 

Today a number of extraordinarily well-funded competitors are racing for an outcome that remains genuinely uncertain, against a backdrop where the monetisation path is unclear. In Australia, the chase has been on to identify companies that will directly benefit from the extraordinary spend announced by the major players. 

By contrast, our approach is to own category-leading businesses that are already moving quickly to apply AI within their own operations. 

We think this gives the Auscap Funds a genuine, and considerably more durable, way to benefit from the AI boom. The continued flow of capital into passive equities strategies and AI development related companies has left many of these high quality businesses trading at attractive valuations. 

We think this presents a long term opportunity for the patient, prudent investor. 

To stay up to date with all the insights from the team at Auscap, click here.

Managed Fund

Auscap High Conviction Australian Equities Fund

Australian Shares

View

Managed Fund

Auscap Ex-20 Australian Equities Fund

Australian Shares

View