Jenine Garrelick:
Hello, everybody. Thank you so much for joining me. We have a topic that everybody wants to talk about, AI, and I am so excited not to be talking about it with just AI, but I have my guests here. Sheridan, you have come back to us.
Sheridan Culhane:
I have. I've returned.
Jenine Garrelick:
That's good. With big demand, Sheridan Culhane from our business development team, but then we also have James Wilson, our senior investment strategist, who's going to give the investment side of the talk of AI. So thank you for joining us. I know you've been with MFS only for a couple of months.
James Wilson:
Yeah. Coming up on four months.
Jenine Garrelick:
So we threw you right into this.
James Wilson:
Yes, you did.
Jenine Garrelick:
But you've been in the industry for how long?
James Wilson:
About 12 years.
Jenine Garrelick:
12 years. So let's kick it off with the investment side of it. You shared with us earlier that it's almost like a three-year anniversary, right?
James Wilson:
Yeah. It was November 30th, 2022 that OpenAI launched ChatGPT. So it's almost been about three years. And I think it's sort of important to point out that milestone just to point out how fast we've gotten here. If I think about a history of what's been happening with AI, it's been around for a long time. And even the models as we currently see them really started in 2017 when a researcher at Google launched a paper on what they call a transformer model.
But there's been some breakthroughs in technology that really put that from theory into practice. One of them is there's been just a ton of internet data. As the world has become more digitized over recent years, there's just so much information online. And these models are only as good as the information you train them on, so now there's just so much more information to train them on. And then the other technological advancement has just been in computing power. And some of that's been advancements in semiconductors, most notably the GPU. So now that you have this very strong computing capacity and a lot of good data, you sort of marry those together and you have the models of today.
Now, the other thing I would point to is we've now gone through this period of the democratization of AI, and that largely started when ChatGPT was launched. And what that means is you used to have to be a very advanced computer scientist to really benefit from these models, but now it's available to anyone. And ChatGPT was really the start of that. And now over the last three years, you've sort of had this dizzying amount of product innovation as people try to bring different applications to consumers and enterprises, and that's been just the typical chatbot, like the ChatGPT or Gemini or Perplexity, to more application-specific type offerings, like Copilot's a popular one that's rolled out across the Office Suite by Microsoft. And then there's image generation and short video generation. And even within some specific industries like healthcare, there are now models that help with patient diagnostics or drug discovery.
So a lot has happened in three years, and it's almost overwhelming all the choice that consumers and companies now have in terms of tools to use, which is why I think it's an interesting topic to learn from Sheridan how advisors can start thinking about this.
Jenine Garrelick:
So the big question everybody has, because it's all of a sudden just been this tidal wave, right, are we in a bubble?
James Wilson:
Yeah. Well, I think it's a question I get almost daily. The honest answer is we don't know. We can't tell the future-
Jenine Garrelick:
I mean, some of these stocks have grown to incredible valuations.
James Wilson:
Yes, they have. And I think maybe we could bring up the slides to sort of talk about this. So when I think about where we are with AI, there's the supply side and there's the demand side. Now, what this slide is showing is the supply side, and what I mean by supply is all the money being spent to stand up leading-edge AI data center capacity. And this is just showing the top four hyperscaler spenders. You can see that just from 2023 to this year, CapEx spending has doubled, and now expectations are that it's going to continue to grow rapidly. Nvidia has been anchoring to this number of a trillion dollars of AI data center spending by 2028. More recently, we've now had estimates of trillions of dollars by 2030.
So very much the spending in this area is significant, and it continues to rapidly scale. And as you had mentioned, this is creating a lot of exciting opportunities in really any company that sells into this supply chain. And it's not just leading semiconductors; it's really anything that goes into a data center. It could be networking infrastructure, it could be cooling technology, it could be the electrical components to source the new electricity. It could be even heavy machinery to excavate these sites to build these data centers. So a lot of stocks that have been tethered to this trend have done very well, and we still see a long road of growing data center spending ahead of us.
But if we move to the next slide, I think this is the thing that you need to be focused on if this is a bubble or not. If we're moving to trillions of dollars of spending in the next five years, we better see meaningful revenue and earnings generation by these AI companies really justify that spend. Now, if you look at the graph on the left here, one of the ways you can really monitor how is the growth in AI usage growing, you can look at a metric called token generation. So if you ask ChatGPT a question, depending on how complex it is, it'll generate a certain amount of tokens. Now, Google just released their token generation numbers on their last earnings call, and just in the 60 days from May to June, token generation doubled, so a sign to us that things currently are ramping pretty fast.
And if you look on the right, we are seeing a lot of early signs of companies deploying AI technology to drive efficiency and profitability. I think the one I would call out here is targeted advertising. We've seen in a number of media companies that they're using AI to create hyper-personalized advertisement targeting, where you're seeing in the financial results where these companies are getting more clicks per ad, which means they can charge the advertisers more. So there's certainly near-term evidence that companies are leveraging AI to generate strong financial results. But again, we've built all this supply, and as we look forward, it's going to be sort of wait and see how fast this technology evolves, and are consumers and enterprises going to view this as so mission-critical that they're willing to pay up for it? And-
Jenine Garrelick:
Well, you said the word hyper twice, hyperscale spending and then hyper-personalization. So the spending will follow the personalization in a very hyper way? Is that what you are-
James Wilson:
Yeah. I suppose.
Jenine Garrelick:
Okay. Which could be a bubble.
James Wilson:
Yes. Now I think there are a couple of important considerations to make. I think the parallel people are always looking at is the dot-com bubble, and there are some-
Jenine Garrelick:
Right. Because that's how it started. They weren't making money.
James Wilson:
Right. It's the same sort of situation where you built out a lot of supply in hopes the demand's going to come. Now in the dot-com bubble, it was laying down fiber-optic cable. The internet eventually did soak up that supply. It just took a lot longer than people anticipated. If you think about society today, we have much greater ability to embrace new technology. We all have computers in our hands. We've all been going along this increased the digitization economy over the last decade or so. So we seem like we're in a much better position to really rapidly deploy a technological change like this than we were 25 years ago.
If I make a couple other comparisons, if you look at the dot-com bubble, another common theme was the indices were concentrated, and those concentrated positions did exceptionally well. If you look at a company like Cisco back then when they performed two or 300% in a pretty short period of time, it was mostly multiple expansion. If you compare that to the great performance we've seen of the mega-cap stocks here in the US, a lot more of that's been earnings-driven than multiple expansion. So when we compare the valuations today versus back then, we think we're not nearly at those levels.
The other point I would make is these mega-cap tech stocks all have very successful entrenched businesses that make a lot of money. So a lot of these investments they've been making in building data center technology has been paid for with free cash flow rather than debt, which is a big differentiator from the tech bubble when a lot of that was debt-driven. So we do think it's certainly we're in a supply-demand mismatch, but we think from a valuation standpoint, at least amongst the US public mega-cap stocks, valuations aren't nearly as stretched as they were previously, and we think as a society, this is a breakthrough technology, we think it will shape and shift how we do things, and we have a society that's going to be able to lean in more to adopt something like that rapidly.
Jenine Garrelick:
And, Sheridan, you're seeing this with advisors; they are leaning in. So to your point of people embracing this quicker, it's happening.
Sheridan Culhane:
Yeah, the catalyst for the conversations with me with advisors was that list that Microsoft put out over the summer. Did you see? It was the 40 jobs most at risk from AI, and a personal financial advisor was on it. And that's because right now-
Jenine Garrelick:
Podcast host? Was podcast host on there? Probably.
Sheridan Culhane:
There is some tech out there that exists.
Jenine Garrelick:
Yep. I'm sure.
Sheridan Culhane:
There's enough audio on you where I'm sure they could get it. But really, if we're being honest, it's like there are tools out there that do a really good job right now answering questions and performing tasks having to do with things like portfolio construction, asset allocation, tax minimization, financial planning, estate planning. So while these used to be ways for advisors and teams to differentiate themselves, in the future this is going to be table stakes, something that people baseline expect to get. But the quote that I've said to you a couple times-
Jenine Garrelick:
You have. Yeah.
Sheridan Culhane:
... is that AI isn't coming for your job, people that know how to use AI are. So the best teams out there, they're actually not shying away from it; they're embracing AI and finding ways to use it to their advantage. So most teams right now are using it to save time, to create capacity, mainly by offloading some of their administrative work. So four common use cases. Number one is to use it for transcribing and summarizing meeting notes. Jump is the popular one that I hear of teams using for that. Number two is for creating standard operating procedures.
Jenine Garrelick:
How is that?
Sheridan Culhane:
So some teams will just create generic ones using ChatGPT, create an SOP for client reviews. That can give it to you. Or I know two different teams that purposely did their training and onboarding of new hires virtually. That way they could have the meeting automatically transcribed, and then they actually use Copilot to create their custom SOP. So it's kind of bringing those two together.
Jenine Garrelick:
Yeah. That's wild.
Sheridan Culhane:
Yeah. Number three is for drafting and editing things, so this is emails, scripts for videos, educational materials, marketing content. For podcasts and visual stuff, Descript is the one that advisors rave about to me. And then the final use case that's common is to use it for generating ideas and for providing feedback, so almost to think of it as like a creative partner. You and I were in a meeting a couple of weeks ago; an advisor said he was going into a meeting in Florida with a business owner, so he quickly popped in, "What are five things I should mention to a client that's a business owner in Florida," and he ended up using two of them in the presentation. So it's all about how you can use it to try to get better at what you're currently doing.
So if you pull up the slide... So it's nothing too drastic. All that stuff is kind of everyday things. But what it's doing is it's saving advisors a couple minutes here and a few minutes there, which over the long term is going to give them hours, days, and eventually weeks back in their year to focus on the things that they should be doing, which if they want to achieve growth, client meetings, client acquisition, and business development, those are the three things that the fastest-growing, highest-performing teams focus on: client meetings, business development, and client acquisition.
Jenine Garrelick:
So with that business owner, it is literally just scraping all the information out there and just serving up what they think the need is going to be.
Sheridan Culhane:
Exactly.
Jenine Garrelick:
So let me throw this. I'm the old lady. I have a paper calendar, okay? And I will embrace this. I will. I will. Hallucination, when they talk about this hallucinating, it is because too much information is being sent in and it's still processing? Because to your point, everyone says you still need a person behind it. So is this a challenge longer-term, or is it going to get...
Sheridan Culhane:
Yeah, the way that teams are using it... If you notice, most of those things have to do internally with what they're doing within their team. There are still many advisors and firms that are reluctant to use these materials directly with their end clients because there is that risk factor of hallucinations happening. So you absolutely need eyes on things, especially with what advisors are doing for clients, which is managing people's money. It's all about building trust, and you don't want to have some sort of AI get in the way of something that's that important.
Jenine Garrelick:
Yeah.
James Wilson:
I think something important to point out is the models we're using today will be the dumbest ones we use. The next ones will be smarter. The next ones will be smarter. But in the near term, you probably do want to keep an eye on what's coming out. That's why if you think about an advisor's practice and some of the things that could be automated, the role of a more entry-level person probably isn't going away, but their skill set might be more of a prompt engineer or someone being able to really leverage these models and make sure they're operating the way that you want.
Jenine Garrelick:
It's interesting because there's a lot of articles about that entry-level position and the certain skills that they're just not going to need anymore, which though can affect longer term, and so you're seeing it on these teams. We do a lot now more with financial advisors teams that are growing that will definitely affect the senior individual.
Sheridan Culhane:
Mm-hmm. And that's why I personally think long-term AI, what it's going to do is it's actually going to put a premium on the human experience that advisors deliver to clients, because AI can't replace that stuff. They can't sit down face-to-face in a meeting with a client. They can't tell them not to sell when markets are going down. They can't keep them invested over the long term, support them through challenging times. I heard one advisor refer to it as the financial hospitality business. So I think it's going to kind of shift more to being about providing a customized experience, delivering a top-tier service model, and on anticipating future needs.
Jenine Garrelick:
Yeah. AI is not going to give a hug.
Sheridan Culhane:
No. So in terms of an action that advisors can do right now, ChatGPT yourself, right? See what it says. Look up yourself on Gemini, on Grok-
Jenine Garrelick:
I used to say Google yourself. No longer.
Sheridan Culhane:
So that's what's going to happen is that that's what's going to be replaced in the future. People are going to ChatGPT you. So figure out where you show up the best. Where do you show up the worst? How are they the same? How are they different? Right now, it's all about just content, getting to the top of somebody's feed. In the future, it's really going to be about the quality of the content that you're putting out there. What's all that data and information going to get crawled, and what's that going to say about you? So advisors that right now have a real dialed-in approach to their brand, their ideal clients, their value proposition, their services, they're going to carve out a nice niche market for themselves and be one of the winners of this AI bubble, if it is it, long term.
Jenine Garrelick:
Yeah. Well, we always say even if it's a bubble, dot-com happened, we still have technology. Housing bubble happened, we still live in homes. So there will be winners or losers, but it's definitely not going away. How is MFS using it? You talked about with the advisor's practice. Any insights?
James Wilson:
Yeah, I think an interesting use case when I think of AI and where it's most beneficial, what are things that would take humans a ton of time to do that's something they can do instantaneously? And even though MFS is largely known for their fundamental business, we do have a large in-house quantitative team that's doing a lot of cool things with data and AI. And I think one of them that's really cool is using natural language processing to scour notes and regulatory filings, so one of which is we have this army of analysts that are always posting meeting notes or reaction to earnings notes, and there's now a program that will read those all for changes in sentiment, to grade stocks on sentiment based on how analysts are writing about them.
And there's another one where there's been research done where when a company meaningfully changes their regulatory documents, they add a lot more meaningful risks in that part of the disclosure document that tends to lead to challenges in the business. And now with a model that we created, it can instantaneously read all these regulatory filings and highlight where are companies where the language they've used has changed? So if you think about how much time it would take for someone to read regulatory filings, we have a model now that can do thousands of stocks instantaneously.
Jenine Garrelick:
That's wild. That is wild. So how is AI affecting advisors' clients?
Sheridan Culhane:
Honestly, the way that I see it most directly impacting advisors' clients on a day-to-day basis is scams. They've shot up through the roof in terms of the frequency, how many times people are getting them, and also how many people are becoming victim. So if we're translating this for advisors, what may happen in the future is that your clients may get a telephone call from what looks like your office and what sounds like your voice telling them about an investment opportunity and providing them with wiring instructions. So that's sci-fi movie script a couple years ago. Now it's-
Jenine Garrelick:
It's real.
Sheridan Culhane:
... not that far-fetched. So what many teams are doing in anticipation of something like this happening is they're developing a game plan, so documenting who within their firm they need to contact, what specific action steps they need to take and when, and finally how they're going to communicate what happened to their clients. Thinking through this stuff ahead of time allows people to act faster in the moment, so that's critical.
Jenine Garrelick:
Oh, it's awful. It was already an issue with scams, with... Can't tell you how many times the IRS has been after me, but... Not really, people, but-
Sheridan Culhane:
But ahead of the holiday season, because that's right around the corner, quishing, it's QR phishing, so that's when a scammer sends a package to your house-
Jenine Garrelick:
Wait, it's called what?
Sheridan Culhane:
QR phishing, or quishing.
Jenine Garrelick:
Okay.
Sheridan Culhane:
So it's when a scammer sends a package to your house, and you, not expecting to receive anything, trying to figure out where it should have been delivered, you scan the code, you download malicious software or you provide personal information. So that's a great one for advisors to make clients aware of ahead of the holiday season. And then the other big one, I got hit with this over the long weekend a couple weekends ago, the text from E-ZPass saying that you have an unpaid toll. Yeah, I would expect a lot more of those throughout the holiday season.
Jenine Garrelick:
Okay. So go back to an advisor best practice for their clients, because that I would say would be an immediate takeaway.
Sheridan Culhane:
Yeah. Educating clients, it's extremely important to do so, whether that's advisors sharing news articles about different scams that have happened, hosting events. I've seen some advisors position themselves as that trusted resource that clients can turn to to confirm whether or not something's real. And then the other thing that's important to do is to re-examine your verification protocols. Many teams, they're not apologizing to clients anymore for making things difficult. They're actually using this as a way to drive business. They're really putting it as part of that risk management and protection that they're committed to providing to their clients. So done in the right way, this is one of those risks that advisors can turn into an opportunity, particularly with clients whose parents, so if you're trying to reach up that family tree, that's a great way to add value. "Hey, if this is something your parents are struggling with, I can be that trusted person that they can turn to here."
Jenine Garrelick:
Wow. That's a great idea. Where are the risks that you're seeing?
James Wilson:
I wouldn't say so much a risk, but maybe something that should be on advisors' radar, there's a lot of talk about, is this a bubble? And there's certainly portions of your portfolio that should be investing in these trends. There's a lot of exciting growth happening. But at MFS, we always preach diversification and to make sure you know what you own in those buckets of your portfolio that you want to have as diversifiers. And the reason why I bring that up is these Magnificent Seven stocks, some of them actually just recently got reconstituted into the Russell 1000 Value Index. And we've done the data and looked at the top, large-cap value active strategies, both from trailing performance and the ones that brought in the most money, and a lot of them own multiple of these mega-cap stocks. Not saying that's a terrible thing, but just make sure you know that if there's a part of your portfolio you want to be a diversifier from this exposure, look under the hood and really see what's going on there.
Jenine Garrelick:
Yeah. I think a lot of people would be surprised when they hear that of the reclassification of some of those stocks into the value index, or even in their own value sleeve. I find that a problem. Personally, I do. But I know we're getting close to time. I have a couple of takeaways, but I want to hear your last thoughts, your last thoughts.
James Wilson:
I would say I think AI is here. I think the world is going to look different three years from now, and I think that factors into, is this a bubble or not? The bar now to generate value is higher now that we're spending more on building this supply, but I think that there are a lot of green shoots that show that this is pretty transformational.
Then my other takeaway would be what I just mentioned. From an asset allocation standpoint, there's a lot of different strategies that are venturing into these names, and just make sure you know what you own and where.
Jenine Garrelick:
What about you?
Sheridan Culhane:
From a business standpoint, start to use it. Use AI to your advantage to create capacity. Ashley, who was on Straight Talk last time, she said that they did a study in 2024 and only 30% of advisors and teams were using AI tools. You're not far behind the curve if you haven't started to use it at this point, so let's say on a daily basis, instead of Googling something, toggle over into that AI mode and start to use it. Just by breaking that daily habit, you'll see all the advantages of it.
And then the other thing, especially for clients, we're closing in on 2025, we're starting 2026, so make sure that educating clients on cybersecurity and identity theft is a part of your plan for 2026.
Jenine Garrelick:
Yeah, I'm going to tag on to those words of wisdom. So we talk previously about IRIS, which I will tell you what it stands for. Do you know what it stands for?
James Wilson:
No.
Jenine Garrelick:
Yeah. I didn't either, so I always have to have it written down. Thank you, Michelle, for writing it down again, Investment Research Insights Suite. The beauty of IRIS, just what James is saying, make sure you understand what you own, because whether it's a bubble or not, you really want to make sure that you understand what's under the hood, especially in that value strategy. You might be overweighted in this area and you might not even know it. So if it is a bubble, you want to be aware.
Second, protecting your data. You'll see in the console there is a checklist for clients to be aware of, but we also have a educational seminar on it. If you want to call your MFS representative, they could do a seminar.
And lastly, knowing that we talked about teams, and we've been talking to a lot of high-performing teams, we do have top tips of high-performing teams if you want to talk to your MFS representative about that also. So tons, tons to talk about.
And I will leave you with this because I did use ChatGPT. Because the one thing that it can also not do, it can't give hugs, it can't... But I asked it to give me a close for Straight Talk. Here's what it came up with. "That's all for me. Unlike your clients, I won't be charging a management fee for this advice." Second close was, "Remember, past performance of this presentation is not indicative," we can't speak, can we, "of future results." And lastly, "If you liked this talk, great. If not, let's call it a tax-lost harvesting opportunity."
Sheridan Culhane:
How appropriate with this time of year.
Jenine Garrelick:
These were terrible. Terrible. So I do think there is some room to grow with ChatGPT, but I thank you all for joining us. I know this is a topic that we will continue to talk about, and thank you for coming.
James Wilson:
Thank you.
Jenine Garrelick:
And, Sheridan, thank you again. And I can guarantee you, you will see their faces again because you guys were great. So say goodbye to our audience.
James Wilson:
Bye.
Jenine Garrelick:
Bye. Thanks for joining us.
Disclosure:
The views expressed in this presentation are those of the presenter. These views should not be relied upon as investment advice, as securities recommendations, or as an indication of trading intent on behalf of any other MFS investment product. MFS does not provide legal, tax, or accounting advice. Clients of MFS should obtain their own independent tax and legal advice based on their particular circumstances.
Past performance is no guarantee of future results. No forecasts can be guaranteed.
Unless otherwise indicated, logos and product and service names are trademarks of MFS® and its affiliates and may be registered in certain countries.
Distributed by: U.S. - MFS Investment Management and MFS Fund Distributors, Inc., Member SIPC; Latin America - MFS International Ltd.
66812.1