Which Stocks Really Count as AI?

Everyone wants exposure to artificial intelligence, but what does that actually mean in practice? In this video, Morningstar's Kennth Lamont dives into this question: how do you determine whether a company is truly an AI investment? Drawing on both analyst research and the collective holdings of AI-focused fund managers worldwide, Lamont reveals how the team built a framework for identifying the companies most closely tied to the AI theme.

The discussion uncovers some surprising findings, including why some of the world's largest technology companies may not rank as highly on AI exposure as many investors assume. Lamont also explore the growing overlap between AI-focused investments and traditional technology portfolios, raising important questions about diversification, thematic investing, and whether investors are paying a premium for exposure they may already own.

The full research, Opportunities in Artificial Intelligence, dives into:

• A clear investment case for AI, including bull and bear scenarios, valuation signals, and bubble risk analysis.
• A combined equity and manager research framework ranking US AI-themed funds by true AI exposure.
• Identification of AI “hidden gems” and “false friends”, where manager positioning and fundamentals diverge.

Robby Greengold: We should talk about, and maybe Kenneth you can help us out here, to talk about your way of defining what qualifies as an AI company. If investors are looking for exposure to this theme, they need to understand who the beneficiaries are, who is going to benefit over the next 5-10 years. So maybe you can just introduce this concept and methodology for establishing what is an AI company.

Kennth Lamont: Yes. So Equity Research have their own definitions that the analysts look at a company and decide how much revenue in the future will come from AI, which is a very direct and very personal way of defining what AI is. But on the Manager Research side of the fence, completely independently, we really leaned into the strengths that we have, which are fund holdings. 

So what we did, we went out and identified all of the thematic fund managers globally that target AI and then looked at their fund holdings and did a frequency analysis, essentially crowdsourcing what an AI stock is from the holdings that are out there. And by scoring stocks based on that crowdsourced approach, you can get something like a beta of a theme. And actually the index that you saw in the first graph today was built up using stocks that are most frequently held by managers globally. So I don't need to even know exactly in what way a stock is related to AI. There are many ways in which a company can be considered AI.

I was up in Sweden for work a few weeks ago and they're very proud of their data center ventilation company, which is AI, right? So, how central is that to the AI story? Or a company that takes AI like Delta Airlines and uses it better than its peers to improve its margins? Is that an AI company? So, what we did was we used this score to essentially build a beta or a core basket of stocks which are exposed to the global AI theme. 

Greengold: What about a stock like Apple? How does that rank under this methodology?

Lamont: So, Apple's an interesting one. So what we did was we took the two lists: We took the Equity Research scores that they've come up with independently and we took the scores that we developed independently, entirely quantitatively on the fund research side. And they mostly aligned, but there were some key places where they didn't align. More than a third of artificial intelligence-focused thematic managers globally hold Apple, whereas our research analysts and the research team gave Apple a score of zero. So it didn't meet our minimum threshold of exposure to AI, which I thought was really interesting.

And our equity research analyst came back and said look, Apple, it's an iPhone company and it will be for the foreseeable future when we're talking about revenue. So AI in that context is perhaps secondary. It's a secondary consideration and it's a big company and of course they'll have their fingers in many pies, but again when you really boil it down to where is the revenue coming from and is that connected with with AI, which I think in some ways highlights one of the flaws with a crowdsourcing approach at a global level as well. You get this tendency to select mega cap stocks, which might not on closer inspection have quite the revenue exposure you'd hope for. 

Greengold: Can you talk about the overlap in the AI stocks and just technology more broadly? How do those relate? What is the correlation between them? And is AI simply technology these days?

Lamont: Well, that's a fantastic question. It's one we've answered in the paper. So what we did was we took this index which we consider to be, let's say the beta of the theme, and we essentially did an active share calculation versus just a technology benchmark. The active share is the bet that one, in this case index, is making versus another. So the overlap between the two has just grown and grown and grown, but the active share has just fallen away through time, and that's essentially the theme becoming almost a victim of its own success.

From a fund investment perspective, that brings up all sorts of questions. You know, why are you paying extra, and you almost always are, with thematic funds for something that might just be giving you broad technology. So it's a question investors have to ask themselves. And we also went a little bit further, and we actually took a different index and showed that the tracking error between the two fell dramatically through time as well. So every sort of indicator that we have suggests that we're sort of overlapping.