Capital IdeasTM

Investment insights from Capital Group

Global Equities
Episode 10 - Investing in AI, every day
Matt Reynolds
Investment Director

With the rapid rise in take-up of ChatGPT, many commentators are now forecasting AI to become as disruptive and epoch-making as the internet. With the abundance of new information and activity in this space, investors are right to be wary of the potential for fads in the development and application of AI that will fall short of long-term success.  Investment Director Matt Reynolds takes a closer look at why the noise about AI has reached this level, whether it’s justified and where the investment opportunities might lie, both now and in the future.

Matt Reynolds is an investment director at Capital Group. He has 25 years of industry experience and has been with Capital Group for four years. Prior to joining Capital, Matt worked as head of Australian equities at Colonial First State Global Asset Management. He holds a bachelor's degree in economics from The University of Sydney. He also holds the Chartered Financial Analyst® designation. Matt is based in Sydney.

Hello and welcome to today's podcast on AI. We've called this podcast ‘Investing in AI Every Day’, and clearly artificial intelligence, at the moment, is really the topic of all the media commentary and a lot of hype and a lot of activity around it. Investors have seen substantial stock movements, and fundamental changes as well with adoption rates for this new mass market technology that have never really been seen before.

It took just five days for ChatGPT to reach 1 million users. The next fastest was Instagram, which took a whole two and a half months when it first started, and Spotify, which took five months when it first started. So clearly, there's a lot going on here, and with these sorts of adoption rates, a lot to learn as well. Our analysts have put it succinctly in saying that the use of AI has reached a point of utility and ChatGPT was like the starting gun in a race where everyone was already jogging.

The bulk of today's podcast is to share some views as to how we might go about investing in companies that may benefit from the evolution of AI over time and share some thoughts on a framework that we can use to look at those opportunities as they evolve. Thinking about the history of AI helps us to appreciate that this is not an overnight phenomenon, and I can put into context some of the investment angles that we use to look at upcoming opportunities within the sector.

The term ‘artificial intelligence’ seems to have been first coined in the Dartmouth Conference in 1956, and in the 1960s, we saw the first computer vision, natural language processing, and expert systems come into existence. After a period of low funding and limited progress in the 1970s, the industry saw advancements in the 1980s in these areas. And by the 1990s, the focus of AI research had shifted to machine learning and neural networks. These efforts saw rapid acceleration in the 2010s due to the availability of large datasets and more powerful computing resources, and this continues today. I think this history is important because it can help shape a view that the high profile announcements we are seeing today have a longer and supportive history and they're not merely overnight flashes. AI's fast adoption is really the result, in my view, of the combination and contribution made by significant improvements in computing power, a collapse in the cost of solid-state storage over the last six years, and an exponential increase in the model sizes that can be used for training and then inference purposes.

There are industry estimates that suggest that the amount of data that will be created over the next three years will exceed the amount of data created thus far in human history. And of course, data is the lifeblood of AI.

So, two questions arise – one from the economics team being, could AI be a solution to the stagnating level of productivity we see worldwide, and in fact, outright declines in productivity, at the moment, in some countries? Secondly, from the perspective of all of the long-term and focused investors, is this really a compelling investment opportunity? Well, in my view, the answer could be yes to both. One of our most senior portfolio managers phrased it well, seeing AI as an open-ended opportunity for companies that can leverage the technology to genuinely differentiate their product offerings and deliver enhanced productivity.

The early days of mobile and cloud were similar in terms of creating tremendous and enduring opportunities for companies to improve their productivity and differentiate their products. Looking at industry research, this suggests that the AI market size in 2022 was around 120 billion US dollars a year. This is increasing to about US$165 billion this year, and by 2030, is expected to reach a somewhat staggering US$1.6 trillion a year. The foundations for this extraordinary growth rate are, in some respects, already here.

When we look across a range of industries, we can already see solid adoption rates in a number of areas. For example, a recent survey showing the percentage of firms using AI for specific business functions had reached around 38% for tech and telecom companies in their risk management functions; 31% of companies in the consumer goods sector were already using AI in their service operations and a similar number of financial service companies were using AI for product development. Around 22% of healthcare and pharmaceutical companies were using AI for their risk assessment programs.

So, what are the risks to the forecasts of the growth that I mentioned earlier. Well, as ever, it is likely to be commoditisation and price deflation for the intellectual property involved. Assuming that there is potential for growth, the other natural question to ask is, how can investors invest in this? So, to that end, I thought it would be useful to share with you today a framework that one may consider for investing in the AI opportunity more broadly. And I can summarise this in terms of four levels of investment.

I'd like to step through each of the four to give it a little bit more colour in terms of the areas of potential going forward. So, for me, the foundational level or level one, if you like, is the compute function, and this is probably the most basic level, the enabler, if you like, in terms of the physical building blocks needed for AI to be created. And this really does come down to the world's need for processing information and storing it for both the short and the long term. And the key here is the extensive use of semiconductors empowering AI. We can see semiconductors being applied to AI applications in quite a number of different areas, from the graphics processing units, through to the custom chips being used by the cloud computer providers to manage the analysis of data, the memory used to be able to store that data and access that data on demand.

Looking at some of the semiconductor industry growth rates through 2030, we see that the current industry growth rate of around 12% per annum really comes from the server data centre and storage industries part of the industry. Interestingly, and really as an aside for today's topic, one of the biggest competitors to AI in terms of demand for semiconductors is actually the automotive industry where the expectations are for 11% growth per annum in its part of the market through to 2030. In terms of the growth of the semiconductor industry overall, we can see that it took the industry 40 years to reach 200 billion in sales around the year 2000. The industry sales are currently around 500 billion a year, and this is expected to almost double to around 940 billion by 2030. Now that's a lot of revenue for an industry where the number of participants has been dramatically declining over the last 30 years or so.

Okay, so let's turn to the second level, which I'd call the infrastructure level. And this covers the areas where the semiconductors live and do their work in data centres and other networking facilities. We can see particularly that the cloud providers are the companies that provide the on-demand access that is needed to be able to undertake the training of the algorithms and these companies have grown tremendously as the use of cloud computing has increased over the last 10 to 15 years. As a reminder, cloud computing is really the shift away from having your own compute-and-storage capabilities on site, to one where you use the computing power of a data centre. The two largest suppliers of public cloud computing are Amazon Web Services and Microsoft Azure.

The public cloud access industry is also very concentrated, with these two companies supplying more than 50% of the public cloud market over time. So really, when we look at the cloud infrastructure industry, we can see at the moment that it's a reasonably large industry at around $69 billion a year in terms of revenues and this is shared amongst a fairly small number of players. When we look at some of the industry research that's out there, the expected growth rate for the cloud infrastructure industry, partly driven by the demands of AI application usage, is expected to be extremely strong, growing around 28% a year through to 2030 on some estimates. So this makes the cloud hyperscale industry an extremely large one from a revenue runway point of view.

The last two areas of the framework are probably some of the most interesting areas, but compared to the first two, compute and infrastructure, they are also some of the least formed.

The third level of the four level industry structure model is to look at solutions and these are the foundational models and other tools and applications that are used to generate output from artificial intelligence. It's really an area where the investment market is a lot broader. It's an investment opportunity set that is less concentrated in terms of the number of companies available to invest in. So, from an industry perspective, there's more questions than there are answers, at this stage, for the AI solutions stack, if you like, in this area of the market. As one of Capital Group's software analysts points out, getting the competitive dynamics of the companies in this part of the market is absolutely critical. A company's ability to monetise their applications will be key over time, so to assess this potential, some of the issues we are researching in this part of the market are questions like, how quickly are the foundational models of the AI in question likely to be commoditised?

Can the fine-tuned models actually be competitive and go mainstream? How sustainable and enduring is the value that's going to be captured by the company selling those models over time. In this third level of solutions, it is the data ownership which is likely to be one of the most important criteria for picking long-term beneficiaries Companies that are productising AI successfully are likely to benefit meaningfully and fast. Successful companies in this area will likely have a very direct monetisation level, most likely by upselling customers into newer AI-enabled products, or by raising their own prices on existing products.

Let's just touch on the final area in terms of beneficiaries, and really the beneficiaries of AI in my view are quite unlimited. In some respects, this is probably the most important area in terms of our investment over the longer term. I think the beneficiaries of AI will be seen across almost all industries where AI is used as part of a solution across different use cases, such as research, marketing, and legal functions within a company. The use of AI can utilise an organisation's own data, as well as third party and copyrighted data to make recommendations that create synergies over time. Initially, the most likely application of AI will be to target cost synergies, reducing costs through improvements, and then on a more medium-term basis by creating recommendations for new revenue opportunities.

Every indication so far is that the speed of delivery here could be much faster than expected. I think the cost of initially weaving AI into existing operations will also dramatically surprise on the low side. I don't think the AI products being sold to enterprises will be free, but they may not be far from it, particularly if they can plug into other currently available commercial enterprise software platforms. I think the return on investment from spending capex on implementing AI has the potential to be amongst the highest amongst all those competing for demands on a firm's cashflow over time. So, similar to the already broad base of company users of AI referenced earlier, we can focus on identifying beneficiaries of AI in many sectors. In this area, global research depth, an active investment approach, and a good understanding of AI's impact across numerous sectors may help identify long-term investment opportunities.

So, to pull this together and draw it to a close, to summarise in three key points:

1. Artificial intelligence has been a long time coming and the foundations of its potential growth from existing technology suppliers and other related industries appear very solid.

2. Artificial intelligence has an economic tailwind via its delivery of improved efficiencies that makes it attractive to almost every industry.

3. Finally, as a long-term investor, you need to have the scale and analytical resources to be able to identify those companies with the greatest long-term potential to benefit and stand aside from some of the current hype.

We're always trying to get better, so if you have any feedback, including topics you'd like to see addressed in future episodes, send us an email at

For Capital Ideas, this is Matt Reynolds reminding you that the most valuable asset is a long-term perspective.

Legal and regulatory information


The information included on this site is neither an offer nor a solicitation to buy or sell any securities or to provide any investment service. Past results is not a guarantee of future results. Statements attributed to an individual represent the opinions of that individual as of the date published and may not necessarily reflect the view of Capital Group or its affiliates.

While Capital Group uses reasonable efforts to obtain information from third-party sources which it believes to be reliable, Capital Group makes no representation or warranty as to the accuracy, reliability or completeness of the information. The information included on this site is of a general nature and does not take into account your objectives, financial situation or needs. Before acting on any of the information you should consider its appropriateness, having regard to your own objectives, financial situation and needs.

In Australia, this communication is issued by Capital Group Investment Management Limited (ACN 164 174 501 AFSL No. 443 118), a member of Capital Group, located at Suite 4201, Level 42, Gateway, 1 Macquarie Place, Sydney NSW 2000 Australia.

All Capital Group trademarks are owned by The Capital Group Companies, Inc. or an affiliated company in the US and other countries. All other company and product names mentioned are the trademarks or registered trademarks of their respective companies.

© 2023 Capital Group. All rights reserved.


Hear from our investment team.

Sign up now to get industry-leading insights and timely articles delivered to your inbox.

By providing your details you are agreeing to receive emails from Capital Group. All emails include an unsubscribe link and you may opt out at any time. For more information, please read the Capital Group Privacy Policy

Past results are not a guarantee of future results. The value of investments and income from them can go down as well as up and you may lose some or all of your initial investment. This information is not intended to provide investment, tax or other advice, or to be a solicitation to buy or sell any securities.

Statements attributed to an individual represent the opinions of that individual as of the date published and do not necessarily reflect the opinions of Capital Group or its affiliates. All information is as at the date indicated unless otherwise stated. Some information may have been obtained from third parties, and as such the reliability of that information is not guaranteed.

Capital Group manages equity assets through three investment groups. These groups make investment and proxy voting decisions independently. Fixed income investment professionals provide fixed income research and investment management across the Capital organization; however, for securities with equity characteristics, they act solely on behalf of one of the three equity investment groups.