REDUCING THE NOISE OF AI INVESTING

“Reducing the Noise of AI Investing”: In this Wealth Actually episode, Frazer Rice speaks with KEVIN SHEA, Senior Equity Analyst at BNY Wealth, about AI Investing and how investors should think about artificial intelligence as an investment theme rather than just a headline-driven trend. They discuss the difference between hype and durable fundamentals, how to segment AI opportunities across infrastructure, software, and end-user adoption, and why free cash flow still matters when evaluating companies tied to AI.

The conversation also explores circular financing risk, the role of management vision in fast-moving markets, which industries may be disrupted or strengthened by AI, and how large institutions are using AI internally to improve productivity, analysis, and client service.

Chapters

• 00:00 – Intro and episode setup
  Frazer Rice introduces the episode, frames AI as a dominant investment theme, and welcomes Kevin Shea to help unpack AI Investing for the audience.
• 01:00 – Hype versus disciplined investing
  Kevin explains that disciplined investing is what allows investors to separate hype from durable opportunity, and argues that AI adoption, spending, and earnings revisions point to real underlying fundamentals.
• 03:00 – How to bucket AI investment themes
  The discussion turns to how investors can organize AI exposure, including beneficiaries versus disrupted companies, technology bottlenecks such as GPUs and networking, and industry adoption themes across sectors.
• 05:30 – Valuation, momentum, and free cash flow
  Kevin discusses why free cash flow per share growth remains one of the most important drivers of stock performance and why parts of the semiconductor ecosystem may deserve a valuation re-rating.
• 08:15 – Circular financing and risk in the AI ecosystem
  Fraser asks about the growing concern that AI companies are financing one another, and Kevin outlines both the bullish “escape velocity” case and the downside risk if business models do not become independently profitable fast enough.
• 11:45 – Infrastructure buildout and competitive uncertainty
  Using analogies like railroads and golf courses, the conversation highlights the risk that early builders may not be the ultimate winners, especially in a market with heavy spending and rapid leapfrogging among competitors.
• 13:00 – AI Investing: Public versus private market exposure
  They examine whether owning public companies such as Alphabet offers meaningful AI exposure, versus gaining more direct but harder-to-access exposure through private investment vehicles.
• 15:45 – What strong AI management teams look like
  Kevin emphasizes that in an environment with no clear historical playbook, vision, execution, and the ability to identify durable differentiation are critical traits in management teams.
• 19:15 – Adaptability and strategic pivots
  Fraser adds that thoughtful adaptation matters, and Kevin notes that sometimes acquisition activity can signal whether a company is innovating ahead of the curve or scrambling to catch up.
• 20:45 – Which industries are most exposed to disruption
  The conversation shifts to sectors under pressure, especially parts of software and IT services, while stressing that disruption does not necessarily mean extinction.
• 24:45 – Why law and accounting may evolve, not disappear
  Fraser offers a contrarian view that AI may make strong legal and accounting professionals more valuable, and Kevin compares that to earlier fears that Excel would eliminate accountants.
• 26:15 – How Kevin uses AI in practice
  Kevin describes how AI has made his team materially more productive, especially in data aggregation, scenario analysis, industry research, and portfolio risk work, while also helping BNY operationally across onboarding, security, and client communication.
• 29:10 – Where to find Kevin and closing remarks
  The episode closes with Kevin sharing where listeners can connect with him and Fraser noting how quickly the AI landscape continues to change.

Links

KEVIN SHEA on Linkedin

RICK FERRI on BRING SIMPLICITY BACK TO INVESTING

Transcript of AI INVESTING

Frazer (00:01)
Welcome aboard, Kevin.

Kevin Shea (00:03)
Yeah, thanks for having me. Appreciate it, Frazer.

Frazer (00:06)
We’re going to tackle two words that have basically taken over the investment world for the last six months: artificial intelligence.

Before we do that, whether it’s AI or crypto or tulips or anything with a lot of hype or buzz around it, how do you think about delineating between investing based on hype and doing it within the confines of a disciplined approach?

Kevin Shea (00:32)
They really do go hand in hand. You need a disciplined approach in order to recognize whether it’s hype or not.

The reality is that it’s pretty impressive, the adoption we’re seeing with AI: the amount of spend, the companies that are participating in and benefiting from AI. There was some concern with the stock movements that many of these companies have seen about whether the market was getting ahead of itself.

Yet we have seen significant estimate increases throughout the year. If you take a look at some of the networking companies, their earnings expectations for 2027 are up almost 50% versus where they were just six months ago. The same is true with memory, GPUs, and CPUs.

Fundamentally, we’re seeing a lot of these companies have expansion in revenue growth and earnings growth, which is quite supportive of a durable trend.

What’s also very important is that adoption of AI is increasing. You can look at enterprise adoption: nearly two‑thirds of enterprises pay for an AI service. You can look at token usage — that’s how much companies are using AI — and that has been parabolic as well.

Look at the revenue generation of these AI models. Right now, they are some of the largest, fastest‑growing companies that have ever existed. So we don’t really see this as a tulip scenario, or even comparable to the internet bubble. We find it very different. We think there are fundamental drivers to this trade, and we’re seeing that through earnings growth.

Frazer (02:37)
Cool.

AI to me is a term that encompasses a lot of different things, and in some ways it’s become like real estate or water — it’s starting to touch a lot of different industries. It’s not just a thing unto itself, but something that’s becoming integrated into a lot of other types of things.

How do you define and bucket the investment themes so that it’s digestible for the investor, and it’s not just, “I’m investing in Anthropic or Google,” but people can parse out where it fits within a portfolio?

Kevin Shea (03:14)
It’s a great question and probably one of the most important ones.

Part of our overarching thesis is that for AI to fulfill its promise, it has to be in every geography, in every industry, at every company, and at almost every employee layer. We’re seeing that when you look at the business units that are adopting AI: customer service, product development, marketing — basically divisions that almost every single company in every geography has.

You phrased it as water, how it touches everything, and we’re seeing that.

So how do you segment it? There are a number of different ways:

• First, you can break it into: who are the AI beneficiaries, and who are those that will be disrupted by AI?
• Second, you can break it down into different bottlenecks. That’s a way I frequently use within the technology landscape: GPUs, CPUs, memory, networking, storage, data centers. Then you look at that framework and see which companies are most exposed to those bottlenecks.
• Third, you can ask: which industries will benefit from adoption? Is that biotech, transportation, warehousing? Which companies could be more negatively influenced — maybe that’s software?

That’s how we try to create an AI Investing framework for where we should focus our investment efforts and determine the allocation that our clients can benefit from.

Frazer (05:17)
As we dive a little bit into how you’ve bucketed these themes across different areas, there’s the concept of benefiting from momentum or valuation versus maybe the cash flow and fundamentals of these different investments.

I could imagine that, with the hype and mania around the space, there’s a lot of interest. How do you temper that valuation play versus analyzing what the cash flows look like?

Kevin Shea (05:49)
One of the most highly correlated metrics to stock outperformance is free cash flow per share growth. That’s often the most important metric, and we watch that heavily.

What’s incredible — and we talked about this earlier with estimate revisions — is that many within the AI ecosystem are generating extremely healthy free cash flow growth and margins. A lot of that is in AI infrastructure. They’re being paid to supply all the equipment and semiconductors.

There’s also this concept that valuation multiples shift to where there’s value creation. I’ll give an example:

The SOX, the semiconductor index, used to trade at parity with the S&P. But there’s been a paradigm shift. A lot of the intelligence that’s being created through these models is powered by semiconductors, networking, packaging, and hardware.

You’ve seen semiconductors go from trading at parity to trading at almost a 50% premium. At the same time, the market is intelligent; it’s shifted its view of software. Software used to trade at a 70% premium, and we think the intelligence layer has moved just one layer above where software applications normally sit.

As a result, you’ve seen valuation compression for the IGV, the software index, from that 70% premium down to about 20%.

Some people might look at the semiconductor index and say it’s more expensive than where it historically trades — maybe that’s hype. But we actually view it as a shift in where the value creation is occurring.

So we think it’s a healthy, understandable move within the market.

Frazer (08:16)
One of the questions that pops up is that there’s a lot of news around the circular flow of cash, where a lot of these companies are all investing in each other. You hear “five hundred billion is going from Google into Anthropic,” or different flavors of that, where it seems like the money is rotating.

And there’s a question as to whether it’s rotating and expanding, given sales and so on. How do you think about that and make sure that we aren’t wandering into more of the sort of things that are happening off balance sheet that we don’t see, while still recognizing the investment that’s taking place?

Kevin Shea (08:57)
At minimum, it raises the risk profile. There are many circumstances and scenarios where this has occurred in the past — the internet being the most commonly referenced — and that obviously did not work out.

There are multiple scenarios that could happen, but for simplicity we’ll break it down into two.

The first scenario is that this is such a capital‑intensive expansion that companies are doing an “all‑hands‑on‑deck” effort. The faster you can get capital from well‑capitalized firms, the faster you can build your infrastructure and reach scale so that these large language models are profitable.

If you can expand and take capital from everywhere, then you can provide enough compute for all enterprises and consumers to utilize your product and your model. You reach “escape velocity” in the sense that your scale allows you to lower costs and become more profitable faster. That’s the glass‑half‑full environment.

Glass‑half‑empty is that they do not reach escape velocity. The business models needed more time to bring the cost of delivering AI down enough to be profitable on their own; they didn’t need this extra capital to reach an enormous amount of scale, and they’re moving too fast.

If that scenario plays out, and these companies are not able to be profitable on their own, and the financial markets become tighter, that creates more downside risk for everybody in the ecosystem.

We don’t see that right now because, at the moment compute is available, it’s being taken right away. We still feel comfortable with the financing occurring right now, but it is one of the top risks that we monitor. It’s not that it’s systemic, but it provides less clarity and disclosure, and it creates a riskier profile as we go through this expansion.

Frazer (11:43)
In the back of your mind, you’re probably saying, “We want to make sure, if there are winners and losers in AI Investing, that we avoid the railroad scenario,” where you build this whole infrastructure and companies have to go bankrupt twice before they actually reach profitability.

Or the bromide that golf courses only become profitable, if they ever do, because the person who built it — a passion project — didn’t make it work, then it goes bankrupt, then the bank is stuck with it and doesn’t know how to run it, then they get rid of it, and then the third person has learned the lessons from the first two and is able to push forward.

Kevin Shea (12:23)
That’s a good point. When we look at all these different models being created, right now you have an environment where everyone is spending and keeps leapfrogging each other at different times.

It’s still a very unknown outcome for all of these players. There’s a lot of competitive intensity in the large language model space and the broader AI ecosystem. It’s certainly a very dynamic environment right now.

Frazer (12:59)
As investors are trying to access this, there are the public companies. You can go on your Fidelity account or talk to your advisor at BNY Mellon or anybody else and say, “I’ve heard about Anthropic or Google or all of these things.”

As far as a good proxy for exposure, how do you think about that?

For example, if I looked at Google and understand that they have underlying investments in their portfolio — in addition to their regular businesses — into these different scenarios, is that a way to get shorthand exposure? As opposed to trying to access a venture fund where the entry points are difficult, the hurdles are high, you need to write big checks, and access is gated?

Kevin Shea (13:53)
It’s a very astute point when you mention circular financing. That doesn’t just happen with public companies; a lot of these vendors and companies in this ecosystem are investing in private companies as well.

When those private companies go public, you find out that Company XYZ is a top owner, and one of their suppliers.

There has been a growing awareness that, with certain public companies, you have exposure to a handful of private companies.

For BNY, our Fujio funds do a lot of our private investments. That’s usually the best way to gain direct exposure.

Frazer (15:37)
Sure.

Not to be flippant, but you’re getting paid to own it at that point via their dividend, as opposed to you paying — at the SPV or LP level — to gain access to it. But yes, it’s definitely not a pure play. I wouldn’t buy Google just to be in a venture fund.

And just to reiterate for listeners, this is not investment advice. We’re trying to learn and talk through different types of scenarios.

As you’re thinking about this and looking at these different companies, what does a good management team look like?

You’d think: a bunch of PhDs, great at coding, lots of experience in the venture community, maybe hung out in Silicon Valley. But everything is so new and dynamic. When you’re evaluating these businesses, what does a good management team look like as they’re trying to scale at warp speed, while profitability may or may not be a thing?

Frazer (17:49)
I’d add that I think there’s an interesting component to AI Investing: a track record of what I would call thoughtful adaptation.

When your business plan gets punched in the face and you’re able to pivot — meaningfully pivot — I’m not talking about a dog food company suddenly putting “.ai” at the end of its name, but someone who can shift and take advantage of opportunities as they come up, as you say, without being so rigid in their vision that they end up getting lapped.

I think that’s an interesting facet to focus on.

Frazer (20:50)
When I try to get my arms around this, I bucket things in terms of:

• Disruption: blowing up something traditional
• Optimization: taking something that’s already good and turning it into great
• World‑building: taking a vision, starting from zero, and building something that didn’t exist before

On that first point, what industries do you think are under attack, and how do you invest around that so you’re not left holding the bag — you’re not a buggy‑whip company as Tesla releases their next issue?

Frazer (24:48)
As an example, I run into all sorts of law firms and accounting firms, and I hear the comment that law firms are going away. I have a contrarian view.

First, I think law has a wonderful ability to metastasize, to find issues, and I think AI is going to be great at finding those and keeping lawyers busy.

Second, for lawyers who are good, I think the ability for AI to make them more efficient and help them graduate to even more detailed and “higher‑value” discussions will only increase.

So when people say, “Law is going to be dead,” I don’t really agree. I think that ties into your point that AI will help some companies that can adapt and use it well to drive further value, probably even charge more. For others, they’ll be left behind or become cottage industries.

Frazer (26:11)
And there will be more and more issues to solve. I don’t underestimate that.

I think AI is going to start poking holes in different things we didn’t think about. Then it will take good brainpower, made more efficient by AI, to deal with these new issues as they pop up.

In your day‑to‑day job, what are you using AI for? Maybe through Bank of New York, and maybe informally, when you’re doing other research — to be smart not only about the company areas, but what you’re doing personally to be more efficient, take advantage of AI, and learn about cool stuff.

Frazer (29:11)
Cool stuff. How do people find Kevin Shea, and any final thoughts?

Kevin (29:20)

KEVIN SHEA on AI Investing

Frazer (29:29)
Terrific. Thanks for being on, and we’ll be sure to stay in touch, as I’m sure everything will be completely different in not just six months — probably six weeks.

Keywords: AI Investing