Investing in AI: a 360° approach
Artificial intelligence (AI) is a theme which is attracting a lot of interest among investors. How would you describe what happened during the past year?
The recent boom in AI use can be attributed to a combination of technology breakthroughs, increasing data availability, and increased awareness of its potential.
However, from a technological perspective, there has not been a great revolution in the sector, but rather a change that vastly improved ease of use and accessibility for the general public. In particular, the AI company, Open AI, modified its user interface and implemented it into a chat service, called Chat GPT. Essentially, the company figured out how to make it easy for end users to retrieve information out of the Large Language Models (LLM), and introduced the world to Generative AI. Some have referred to it as AI’s iPhone moment.
The public discovered that ChatGPT could engage in endless conversations, answer questions, write poems, and write code. It has been trained on a huge set of information, and can generate responses which can be surprisingly accurate, but also can provide answers which are alarmingly wrong, highlighting the need for continued refinement. All at once, ChatGPT exploded in the news and was the main topic of countless articles and conversations. It is the fastest-growing app ever in history, reaching a whopping 100 million active users in only two months. Citing the growing enthusiasm for generative AI, chief executives of major tech companies began allocating significant resources to it due to the great exponential growth potential of these models. In particular, there has been an increase in investment in data centres and cloud computing systems to provide the enormous amount of processing power needed to run AI applications. This helped ignite an AI investment rally with companies using their quarterly earnings calls to promote their plans on how they were going to monetise AI in their businesses – whether real or imagined.
How would you invest in AI? Which companies would you focus on?
In the short and medium term, we want to follow where we think the spending is going. We believe one of the biggest opportunities is in the companies supplying the picks and shovels of AI – the companies related to the buildout of the infrastructure or plumbing needed for AI to reach its full potential. These businesses include cloud providers, data centres, semiconductors and network communication companies. Examples of companies in which we have invested and have often been cited as immediate beneficiaries of AI include Nvidia, Microsoft, Alphabet, Arista Network and Broadcom.
AI-leading companies have already reported astonishing performance. Are you looking for second-tier stocks?
We have invested in companies that are not known as direct beneficiaries of AI, but which we believe have unique products and services that will give them a competitive advantage as the AI infrastructure continues to be developed. These companies include names like Synopsis, a world leader in Electronic Design Automation (aka EDA, a term for chip design software), which recently launched Synopsys.ai, a suite of AI-based design solutions, the verification, testing and production of the most advanced chips.
Another example is Marvel, a leader in semiconductor solutions for data infrastructure, which sells chips for optical interconnects within data centres, designed to increase the bandwidth and performance of cloud data centres.
Another significant aspect of the infrastructure concerns the electricity needed to operate the data centres, which, in fact, require a large amount of energy. One of the main problems worldwide currently is the significant shortage of power transformers. One company that is capitalising on this element is Bloom Energy, which has natural gas-based fuel cells that run on hydrogen. In the event that a data center wanted to build an AI building with many servers and graphics processors, but encountered resistance from the local authority, Bloom could provide a solution more quickly.
Is investing in semiconductors a way to exploit the AI theme? Is there a cyclical risk in the sector?
Many semiconductor companies that benefit from AI are also well-positioned when it comes to the rest of their businesses. So, even if AI wasn’t such a big trend, companies like Lam Research, Broadcom or Nvidia would still be performing well. AI represents, in this case, a sort of icing on the cake. The chips needed for AI consume a huge amount of 12-inch wafers: the dimensions are becoming increasingly large, so much so that designers increasingly have to divide them into multiple chips and then package them together.
Furthermore, I would like to point out that the semiconductor industry is much less cyclical than one might assume. The latter, in fact, in addition to the final markets of PCs and portable devices, is now fundamental in sectors such as cars (electric vehicles require a greater quantity of semiconductors to power infotainment, battery management and security), data centres, Internet of Things, machine learning, artificial intelligence, robotics, household appliances and consumer electronics, to name a few. Furthermore, this industry has gone through a phase of significant consolidation in recent years.
At Columbia Threadneedle Investments, we believe the semiconductor industry is the common denominator and central building block of many of the ongoing secular trends. In this sense, we invest in profitable, growing companies that often trade at below-market multiples.
Would you invest in companies that leverage AI to run their business?
We believe there are many businesses outside of the technology sector that will benefit from the adoption of AI. To date, it is possible to identify opportunities in various sectors, including:
Advertising, which could use generative AI for ad creation, allowing companies to reduce costs and reinvest the savings to create more personalised commercials, then verify their effectiveness through data analysis, leading to higher sales and improved margins. margins.
To continue feeding the world population, agriculture needs to have more and better-yielding farmland outside the United States and the European Union. Farmers currently use smart farming to understand the optimal quantity and ideal placement of seeds and pesticides. Currently, fertilisers and labour account for 50% of expenses – a significant reduction in these could lead to a decrease in the costs of corn and feed corn and, ultimately, also of beef.
There are also various opportunities to increase efficiency in healthcare; in the US alone there is an annual expenditure of $4 trillion, of which 20-30% is estimated as waste attributable to administrative complexity, pricing failure, excessive treatments, fraud and abuse within the system and lack of coordination of care. The use of AI in medical care companies could increase more efficient use of resources, reducing the administrative complexity that has burdened providers and insurers with excessive costs related to billing and coding. Additionally, AI could help improve standards of patient care through doctor-assisted diagnostic tools, increasing efficiency in obtaining pre-authorisation from insurance companies.
Finally, we see interesting application opportunities in customer service and help desks. Increasing the use of competent virtual agents should help keep human headcount significantly below the amount that would have been needed between now and 10 years without AI, thus also reducing the related costs. Help desk software companies will charge higher rates to make up for lost revenue from their previous business model of “per seat” or “per agent” pricing. Virtual agents will handle less complicated requests through natural language, freeing up operators to handle more complex requests. Initial studies on AI tools conducted by the National Bureau of Economic Research have shown that they can boost the productivity of call centre workers by almost 14%.
During the dot.com bubble, we saw a lot of interest in companies that later either disappeared or recorded a poor performance on the stock market. It took a while before some of them were able to regain their performance. Is there a risk that this could occur for AI companies?
The internet has developed in overlapping waves of innovation that have included many successes and failures. However, the adoption of the secular structural trend has been to increase its capabilities and opportunities. We believe that this modus operandi will also apply to AI: today we remain in constant observation and analysis of the evolution, proliferation and impacts that it can produce in the economy.
Every firm is focused on cost control, and adopting AI creates greater efficiency with fewer resources. In the long term, AI will be a deflationary factor and we believe we are at the early cusp of this secular trend today. However, it is advisable to remain vigilant, because we could find ourselves in a phase of excessive enthusiasm. Indeed, although there are many companies currently benefiting from AI, we are convinced that there are others that have not yet emerged.
Conversely, there are some companies that are perhaps overexaggerating their potential to benefit from AI. A similar trend appeared in the late 1990s, where a company could add “dot com” to its name and see its stock rise, even as its business model remained the same. Over the years, we have experienced many of these technology cycles, including e-commerce, cryptocurrencies, fintech, and many others. At Columbia Threadneedle Investments, we are committed to deeply scrutinising each company’s business models, the validity of their technological innovations and their market valuations before making an investment decision.
