Valerio Baselli: Hello and welcome to Morningstar. Everyone wants to invest in artificial intelligence, but there are several ways to do so. Which one is the most suitable at the moment? Today, I'm joined by Rahul Bhushan, Global Head of Index at ARK Invest Europe.
Rahul, Nvidia (NVDA) recently became the most valuable company in the world. Its impressive rally has pushed many investors towards the theme of AI. But now, you at ARK Invest, think it's time to start showing some caution. Can you explain why?
Rahul Bhushan: Yeah. First of all, thanks Valerio for having us on. The first thing is, we've gone through, and we might still be going through a massive, unprecedented AI CapEx cycle. And so, here's some context. The combined CapEx of Microsoft (MSFT), Google (GOOGL), and Meta (GOOGL) is set to skyrocket this year by around 70% versus last year. So, we're going to hit anticipated $152 billion from $87 billion last year. So that's nearly a doubling and as a percentage of sales, that's going from about 13% to almost 20% of sales as combined CapEx. So, if you look at the trajectory there, these figures are – they're pretty mind blowing. We had $52 billion in 2020; we had $65 billion in 2021; we had $86 billion in 2022; $87 billion in 2023; and then $152 billion. So nearly double anticipated for the end of this year.
And so, all of this investment has benefited and continues to benefit Nvidia. Especially if you go back just a year ago, we actually had a GPU shortage, and that GPU supply shortage, as you know, has eased. But what it's meant is that companies like Microsoft and others spent massively on hoarding and over-capacity and stockpiling, which has contributed to Nvidia’s revenue. But GPU computing, we know, is rapidly becoming a commodity. Many people have been saying that. And this is creating a price drop and depreciation in current generation chips. So, we're now eagerly awaiting next generation chips like Nvidia’s Blackwell, which is going to succeed the Hopper series, and which is going to have significant advancements, particularly in things like ray tracing capabilities. So, while Nvidia’s advancements may continue, the broader AI and AI chips ecosystem, you can think of memory chips, for example, which are more standardised, may not deliver the same revenue expectations. And this, we believe, warrants a more cautious approach, let alone all of the things that have happened over the last seven days.
VB: Right. And so, in the light of this, where do you see the best opportunities within the AI universe right now?
RB: So, we believe that investors should be shifting their attention to software, AI software. And AI software, I think the best way to characterise it is we believe it represents wave two, so to speak, of the AI opportunity, if you assume that wave one was chips, aka, Nvidia and SMCI (SMCI). And wave two is exciting because it's less obvious, at least not fully obvious to everybody yet, which means that we've been able to be highly opportunistic in our buying of companies. And so, the way to think about wave two is to visualise the full AI stack.
And Frank Downing, who is our director of research for AI, he laid this out recently quite brilliantly. There's basically three layers, right? So, you've got infrastructure as a service, which is the foundational hardware and compute for AI development and deployment. So, these are your companies like AWS, Google Cloud, Microsoft Azure. And so, these are the infrastructure providers that make hardware widely available and accessible, using service contracts. Then you've got number two, platform and infrastructure software companies. So, these are the tools and systems that developers use to build, deploy, and maintain AI applications. And finally, you've got SaaS, so software-as-a-service. So, these are the companies that deliver applications over the internet. So, an example would be Salesforce (CRM) or HubSpot (HUBS).
So, for clarity, SaaS applications are built and maintained by developers who use tools from the platform and infrastructure software category, so the category number two. The key distinction between platform and infrastructure software companies and SaaS is whether the tool is used to build or secure an application, or if it is the application itself in the case of SaaS. So, our research suggests that the segment that is growing the fastest is the second segment. So, the platform and infrastructure software category, and as the cost of AI development falls, there's a greater incentive to build more customised AI applications. And this is often happening at the expense of companies in the third category, the SaaS category. And signs of this are already apparent. Majority of stocks in our AI and robotics ETF really fall in this second category.
VB: That's very interesting. So finally, to be even more concrete, can you name three companies, three stocks, that could replicate Nvidia’s success in the coming years and briefly explain why?
RB: Sure. So, the first stock we'd highlight is Palantir (PLTR). It's a much talked about name today, certainly, given how strongly it's performed over the course of this year already. But essentially, what Palantir does is they help organisations make sense of large amounts of data. They provide the tools that analyse the corporate data to uncover patterns, to uncover trends, to deliver new insights that inform better decision-making using that, leveraging that data. So, we're pretty confident that AI is going to be a game changer for Palantir.
Palantir Technologies (PLTR)
• Morningstar Rating: ★
• Fair Value Estimate: $6.00
• Economic Moat: Narrow
The second company I'd highlight is Teradyne (TER). So, Teradyne makes machines that test electronic equipment and robots. These are robots that typically are robots that are helping automate manufacturing in some ways. It's a well-known name, but we believe it's an underappreciated name, certainly in the last 18 months where all of the attention has been on seven magnificent companies. And Teradyne, essentially – the way you can think about Teradyne is they ensure that robots are working correctly. And so, if we're right, what we're going to see with this company is a massive improvement in the precision, efficacy, and efficiency of their testing, so leading to fewer errors, downtime, and ultimately, these robots can become more productive. So, it's going to create greater productivity for Teradyne customers, which is going to accrue a lot of value back to Teradyne, we believe.
Teradyne (TER)
• Morningstar Rating: ★★★
• Fair Value Estimate: $135.00
• Economic Moat: Wide
And number three – and this one is perhaps a little more obvious and certainly been heavily talked about this year – is Meta. AI is already enhancing Meta's ability to deliver personalised content and improve their ad targeting. We use Meta ad targeting as a company. You can already see that the quality of their targeting is improving, and you already see it in the numbers for Meta. They gave up that big ambition focused on the Metaverse, and they are going full steam ahead down this AI opportunity. And it's already unlocked new revenue, and it's continuing to unlock new revenue. And we believe there's still much more upside. So those are my three picks.
Meta Platforms (META)
• Morningstar Rating: ★★
• Fair Value Estimate: $400.00
• Economic Moat: Narrow
Baselli: Very interesting. Thank you so much for your time, Rahul. For Morningstar, I'm Valerio Baselli. Thanks for watching.
