cash.ch: Mr. Lin, the beginnings of artificial intelligence (AI) go back far into the 20th century. But why has AI become so popular in the recent past?
Jeffrey Lin: In fact, computer scientists started to think about AI in the 1950s and 1960s. But for many decades there was not enough computing power - we measure the performance of an AI-processor in terms of tera operations per second. And the numbers have become quite astronomical in recent years. In 2017, some of the Nvidia processors had 8 to 12 tera operations per second. In the new AI-enabled computers there are 35 tera operations per second.
So all comes down to how much processing power you actually have - really?
Yes - and today, we have more computing power than ever before. That is why the technology is currently expanding rapidly. However, there are previous technological developments, which enable AI - cloud computing and high-speed data communication for instance.
What suggests that AI will further develop at this rate?
Let us look at Nvidia. The company has moved to a new product introduction cadence of about every twelve months - six months faster than we were previously used to. Hence, Nivida is actually accelerating. We feel therefore confident that the performance will continue to improve. In addition to that, the company has experts in different vertical markets for their technology - it is broad-based. It has people who understand – for instance – life science, but also advanced communications and the oil and gas exploration industry. So they have a lot of people who understand the computing demands of their customers. That is a lasting business model.
What could slow down or even stop the development of artificial intelligence?
There are a few things, for instance: Do consumers want it at all? Do consumers want the technology? Do they find it useful? That is crucial. There will be a market demand, when things become easier and less expensive. From a technical point of view, the size of semiconductors is challenging.
In what way?
In the 1990s we were talking about micrometers, today semiconductors are about a few nanometers. So there are some physical limits in how small you can make these things. The question is whether one can pack more and more performance on smaller and smaller devices. Maybe quantum computing can improve performance going forward. However, this technology is not commercial yet.
What are the differences between the situation today and the dotcom bubble that burst at the beginning of the millennium?
The revenue growth. There has been a strong pattern of revenue growth and profitability from AI in many companies - Nvidia or IBM for example. In 2025 revenues from data centers will be up over 100 times since 2017 and approach 100 billion dollars. So there is real revenue there, which comes from companies with strong cash flows. This contrasts with the situation at the beginning of the millennium. At that time many companies wanted to make money on the Internet. But they were not as strong in sales and not as profitable as the markets had previously expected. So the bubble burst.
You estimate the economic potential of artificial intelligence at several trillion US dollars per year. How do you justify this estimate?
This estimate comes out of market research from McKinsey and Accenture. And we think this number is plausible. However, we also think a bit more bottom up about the opportunities. Let us take the example of ride sharing versus autonomous taxis. Changing to autonomous driving would lower costs for end consumers - which increases demand and makes the market grow. The same logic applies to many other markets, from finance to healthcare industry. And that amounts to an economic impact of several trillion dollars.
How can investors benefit from the emergence of artificial intelligence?
Basically, when a business or technology gets bigger at strong profit margins, the value goes up. More precisely, there are AI-enabling companies like Arista, Broadcom and Nividia. They have been benefiting from the early stage of AI quite well and show top line growth and profit margins. Apart from that, we think of the providers, Adobe for instance. They deliver enterprise software with AI features. These companies are already very profitable and grow nicely. They can charge more and have well working relationships with their customers. So the customer lifetime value increases. Therefore, we think there is a tailwind to their valuations. Then we think about companies outside of core technology that are strategically using AI to improve their operations and services to clients and customers - think of healthcare companies or car manufacturers. These firms are likely to have a better top line growth and better profit margin, again, creating value for investors.
Nvidia is currently on a roll. Are there other stocks that are particularly lucrative due to artificial intelligence?
Let me illustrate that: Companies can use the technology to create large language models for their own software or customer relationships. These firms become kind of like countries. They have their own dialect of how they do things and how they speak and interact. As a result, they are supposed to deliver better performance. So, we think this is a good spot to be as an investor.
Can you give an example?
A good example is John Deere, an industrial company. It already has an autonomous tractor that can drive through a farm field with a camera. It interacts remotely with the farmer and can determine whether a plant is wanted on the field or not. Accordingly, it sprays it with pesticide or not. That automated process increases productivity.
Which European and especially Swiss companies benefit from artificial intelligence and are therefore worth considering for investors?
There are actually some companies. SAP is one of them, a dominant software company with very strong revenues. It is incorporating AI into their software. So they are a provider of technology. Certainly ASML - it has played a key role in the semiconductor manufacturing by providing advanced semiconductors. When we look to Asia, there are others. There are others in Japan. There are some in China as well. So there are investment opportunities around the globe.
What exactly is in it for shareholders in terms of returns?
We can not talk about specific return or expected return. But whenever the market and computing power increase the market for computing increases, too. This leads to higher growth rates for the companies that are involved with it. And if things play out, the companies experience high growth rates for quite a few years, which should lead to good investment returns.
What relapse risks do you see for these stocks?
Many companies are having discussions at the board level about how they are going to deploy AI. They ultimately may have to transform the entire company, which takes time and costs money. Apart from that: Regulatory is discussed quite a bit. There should be a balance here. Regulations should protect people, but also maintain the long term growth opportunity as well. This process of balancing is still going on.
What do you think about the latest regulation of the European Union?
We do not see it hindering technological development. It sets the ground rules for how to do things. And this is important, because the greater population needs to have trust in technology. And having regulation in place and having companies being responsible is extremely important.
What about cyber attacts as a risk for companies, which engage in AI?
Cyber criminals can use AI, too. For example, they could use it to replicate your voice. However, the counter to it is also AI. Many of the leading security vendors are already using AI to counter a lot of those issues. Cyber security has been a topic for a long, long time. This has not slowed down technology development. But it is a cost that one has to absorb to counter the occuring threats.
What alternatives are there, if technology companies' profit growth slows down soon?
Then, investors can move to non-technology sectors that benefit from artificial intelligence. Think of banks. They can use AI in several different areas. On the front end they can improve customer services; on the back end they can boost operations. For example: As a bank knows that a family has a small child, it can easily generate college saving plans. Or it can offer a tailor-made mortgage product, as the family is willing to buy a home. In general, companies that embrace AI will have competitive advantage. And companies that do not embrace the technology will fall behind over time.
Jeffrey Lin has been Head of Thematic Equities at M&G Investments, an international investment company, since January 2023. Before joining M&G, he gained around thirty years of experience in the financial industry.
