Speaker 0:
Welcome to the winter flood research, uh, panel session on generative artificial intelligence. Clearly A I has experienced rapid growth and attracted global attention

Speaker 0:
to explore the impact. Today, I'm joined by a panel of investment trust fund managers from Silicon Valley. Um, Mike Seidenberg, uh, lead manager of all technology Trust. Mark Shepherd, the lead manager of Manchester and London Investment Trust, as well as Nick Williams, analyst at on the Polar Capital Technology Trust as well as the Polar capital Artificial Intelligence Fund.

Speaker 0:
Mike, perhaps your first impression. Uh uh, for the audience over the last few months, this rapid development, uh, in generative a I what has surprised you most.

Speaker 0:
I mean, I think the thing that surprised me most is just the overall just excitement and willingness of companies to really think about this technology and start actually playing around with it with respect to kind of what it means to their businesses. Um, I had, you know, I had I was at a basketball game the other night with the head of innovation from

Speaker 0:
Permanente, which is equivalent to, you know, think of the NHS right, and he was just talking about what it's going to mean to health care delivery with respect to better outcomes, cheaper costs, et cetera, et cetera. You know you don't hear stuff like that very often. I've been in technology for 20 plus years, so I think that there's a real will

Speaker 0:
this and a desire by companies to to to take a look at this and really embrace it. Um, you know, look, it's early days, and I'd like to remind everyone on the call that you know, we are early, but I I This is, you know, this is a This could be a seminal moment when somehow with with respect to this technology and what it means to companies,

Speaker 0:
thank you, Uh, and Nick having, uh, worked on this topic for many years now, uh, is there anything in the in the in the last few months that you've seen that that has still surprised you?

Speaker 1:
Yeah. I think the, um the fact that we've reached this inflexion point, um, and the pace of development since that inflexion point has has moved so quickly, you know, development in a I and and research and progress in a I has has been, you know, very strong for for quite a long time.

Speaker 1:
But then, with the development of GP t three that triggered kind of an inflexion point in the underlying technology and then with the release of, uh, chat GP t uh, in late last year,

Speaker 1:
we all of a sudden reached this point where so many more people could access this technology and play with it and start experimenting with the benefits that previously had been, um, you know, less successful and only, uh, usable by a few people in in in the past. And since the the release and the fact that,

Speaker 1:
um, you know, more people can

Speaker 1:
envisage what they might do with it and start experimenting with the technology. We've seen potential use cases, you know, coming as as quickly as kind of people's imagination. Really?

Speaker 1:
Yeah, I think that's

Speaker 0:
right. Um,

Speaker 0:
and Mark, uh, given the concentration of your portfolio in Microsoft, I'm going to presume that this will be a Microsoft related, uh uh uh, answer. But is there anything in particular you you've seen that that that has really struck you, uh, to lead you to think this is an inflexion point?

Speaker 0:
Yes. Um, I think to me, Bill Gates is the most important genius of the last 100 years. I think if you read, uh, ancient history or religious text, the word ubiquity comes into the sort of definition of a God quite regularly. And software is everywhere in the world. Now it is ubiquitous. Everywhere you go, it's even in you,

Speaker 0:
you or on you. You know you cannot do anything in the world without software. So for a man like Bill Gates to turn around and I I've got two quotes that he said recently here about a I he said a I will go further and faster than software. He then turned around and said, We are now entering a stunning technological age. So

Speaker 0:
you know, whether you want to listen to three fund managers who have all got a portfolio stuff with this stuff is questionable. But I do strongly suggest you listen to Bill Gates, Uh and then ask a question to yourself, You know, can you be anything but anything but overweight to this, uh, this epoch defining revolution?

Speaker 0:
So across this emerging space of generative artificial intelligence, uh, potentially across text video image generation elsewhere, is there a particular development that you would like to highlight, Uh, may Maybe we start, uh, with Mike, Uh, that you that you would like to highlight That you see, uh, the clearest application for and that figures into your, uh, investment process, Uh, ever so slightly.

Speaker 0:
Look, I think that, as you see, these multi multi modal models start to emerge. You know, I really think in terms of infrastructure and what it takes in in order to to kind of generate the answer. So, you know, I don't have a clear, uh, you know, killer or use case. But what I do think is,

Speaker 0:
you know, Harken back to, you know, the gold rush. You know, I'm I've always preferred to be in the pick, picks and shovels business versus, you know, trying to stake my claim on a mine. Um, so that that's how I think about it. Just because I don't think we know yet, um, kind of what the killer application is. I do think that, you know, as as time goes on, if you think about

Speaker 0:
some of the, you know, societal pains that we have that this you know, this this movement will allow us the ability to solve some of those problems. But you know it's going to be multiple iterations, or it's going to be fits and starts, et cetera, et cetera. So I don't see I don't see a clear, you know, winner or kind of killer winner out today. But that doesn't mean that, you know that I won't, you know, a year from now.

Speaker 0:
No, that makes sense. And I think that that brings us to one of the first key key topics that we're going to cover today. Uh uh. Being the picks and shovels business of semiconductors. Uh uh, perhaps Nick, uh, to you, the question is NVIDIA the only game in town

Speaker 1:
A very interesting question because NVIDIA has picked up a huge amount of of interest. Um, given it has been, you know, at the forefront of a I development for a long time.

Speaker 1:
NVIDIA's GP US, uh, you know, have an overwhelming share in in terms of, uh, their use in training A I models and training these these large language models which are so compute intensive because of the, um the parallel computing nature of GP US

Speaker 1:
Is NVIDIA the only game in town? I think the the honest answer is we We don't know, uh, at the minute. Um, certainly they are continuing to innovate massively. Uh, as we see from the g t c conference. You know, every year, um, and they we would expect them to to retain, you know, healthy advantage in in the GP u market. Um, certainly, with respect to training, um, given, you know how embedded they are and the fact that they're at the leading edge.

Speaker 1:
I think if we look at where, um, computing power, uh, is going to be focusing over the next few years. We are moving from the point of training these models to the inference stage, and that's starting to, you know, uh, apply the models and and actually run, you know, get outcome from them rather than kind of preparing them for widespread usage. The the inference market is going to grow massively, Um,

Speaker 1:
and the training market will continue to grow. But I think you know, the the longer term future growth is definitely an inference. There you have more of a mix of technologies being used. So GP US, uh, are used in inference as well. Um, of which NVIDIA you know, obviously still have AAA very good chair. Um, but you also have x 86 more traditional C p US. Um, and there are different partnerships going on there.

Speaker 1:
Um, so if we look within the GP u space first, uh, NVIDIA, Uh um, you know, have a very big share in GP US. Um, in inference. But we've seen that a MD, uh, they've released the M I 300 which is a, um, a GP u that they are hoping to start gaining some market share with. We think that they are probably more likely to gain share in, um, inference. Uh, and I think that's where the the market hopes that they might make some make some gains.

Speaker 1:
Um, And then you've got the X 86 architecture. How that pans out. You know, we we we don't know. Um, it's Microsoft have been using, um f GB f GPA s, which are, uh, programmable chips. Um, they've been working with intel for a long time.

Speaker 1:
That requires kind of expensive, um, hardware engineers. So it has its own drawbacks, but people are coming at this from a different angle. So training has been very much a GP u game. Inference is, uh, a slightly wider set. And

Speaker 1:
in video,

Speaker 1:
we will see, we expect it to, you know, maintain a dominant position. But,

Speaker 1:
uh, there are other players attacking this market, and there is more than one way to come at the inference market. So we will see how this shakes out. I I don't think we can say conclusively at the minute. Um, but the market is very big, and established players are are chasing it. And the pace of development, you know, continues to be very strong. And I think semiconductors have been picked up as one of the most obvious beneficiaries from the growth in a I because everything you know, whether it's inference, whether it's training more models, all of these things require more compute power.

Speaker 0:
That is very interesting. Thank you. Um, on that note, perhaps, Mark, uh, do you see this same shift from training to inference, Uh, in terms of AAA I model processing, Uh uh, driving a wider utility across the semiconductor space. And are there any other, uh, uh, names, particularly that are attracting your interest?

Speaker 0:
Well, I think it might come down, um, to what the hyper scalar want. Because, um, normally in particular, if you look at the sort of architecture network of the people that the hyper scale use in their data centres, they don't like someone getting too much power within their sort of ecosystem. So I think

Speaker 0:
in there, um, it's a great opportunity for a company like a MD because, um, the hyper scalar are almost going to want to include other players apart from just NVIDIA, uh, within the whole architecture. And of course, you have to remember, as we all know, um, that the hyper scalar themselves are now undertaking their own specialist chips as well. So

Speaker 0:
getting those designs, so so you know, and I don't think it's going to end up just with the video being the only game in town on on that on that aspect. But, um, I think the really interesting way to look at this maybe is what we are going to see is a huge amount of design of chips, and they are going to have to be much more optimised, much more specific for the task. They have to take the data

Speaker 0:
to the algorithm, bring it together, and make this work in the most optimised way. That is effectively. What undertaking an A I project is about and you know, Then you start looking to, I think, sort of companies like synopsis and Cadence, which also can feed into this on the design angle and and maybe also on the manufacturing straight manufacturing angle of

Speaker 0:
s MC as well. And so there are other players as well. So I I think NVIDIA is clearly as someone described on C NBC NVIDIA will be the grand horseman of the A. I mean, stocks and it and it will. I I I'd be very surprised if that changes, but that, you know, this sort of, uh, position they have will feed through into other layers.

Speaker 0:
All right, Thank you. Um, Mike, maybe, uh, whether you, uh, concur with the description of NVIDIA as the fourth horseman of the A I apocalypse. Uh, but also, uh, beyond that, um, uh, potentially given that across the semiconductor space, we've seen great share price returns. Uh, over the year to date, uh, obviously linked to this anticipation and and actual, uh um factual, uh, increase in demand.

Speaker 0:
Um, will the semiconductor industry be able to cope with that because it is not too long ago that we were facing global chip shortages.

Speaker 0:
Yeah. I mean, I think we're in a different different position today than we were during the pandemic. Uh, I just want to add on to Mark's point. I think that I think, you know, it is wise to be thinking about, uh, chip and system design. Um, as you know, a a 1st and 2nd derivative of this particular this particular theme. Um, and, you know, uh, that's something that I've spent a decent amount of time on and that you're gonna have You know, that

Speaker 0:
the ability to design the chips is really predicated on the on the software needed to actually design them. Right? So, you know, therefore, that that becomes a real important space. Um, that, you know, you have You have very few players who can facilitate that. So I think that's a really interesting space. Um, you know, into Nick's point, which I thought was really well, well articulated. Look, you know, this is somewhat of this is so

Speaker 0:
open field, right? I mean, when we get into production, I think in terms of learning and then production inference. You know, however you want to say it, I think that, you know, we're still TBD with respect to, you know, who wins when these models actually go into, you know, production, you know, go from learning to inference. Especially if you think about multi model multi mode models, right? I mean, people just don't know. I do think that it just it just

Speaker 0:
it's just an exciting time to be in the space. Uh, just because of just the slope of the curve is so steep right now. Um, And you just in my career, you just don't see that that often where you have, you know, kind of, You know, you know, as somebody alluded to earlier, Bill Gates is a man of few words. So when he says things, you know, people's ears should perk up. He he we he doesn't shoot from the hip as anyone who's ever followed his career. He's very demonstrative when he says things.

Speaker 0:
Mark, uh, if I could turn to you and post you to question uh, the, uh uh, semiconductors. Uh, maybe if we could, uh, briefly shift into search

Speaker 0:
specifically the idea that the integration of, uh, chat GP t dolly. Any other generative technologies into bing, which is being trialled as we speak, uh, will allow Bing to take search share, but also the idea that potentially chat could be either a replacement or a compliment of traditional keyword search. Uh, how do you see this? Sort of new, uh, emerging paradigm?

Speaker 0:
Well, you know, I think I really this is a key thing for me. Um, the real excitement about a I lies actually much more in the enterprise than the consumer. And, um, it's very interesting. And and these guys may understand this as well. Um, it is so much easier for people. I read a sort of psychological paper about people conceptualising um uh, products or services. So, you know, every fund manager

Speaker 0:
has to own Apple because the guy on the street who buys your fund knows what apple is. But if you try to explain to him why you own cadence, I mean, they're just not interested so people don't like components. People don't like invisibles. They like to see a a consumer product that they can get and that engages them. So it's actually quite interesting in the development of various technological themes that the actual sort of instigator to really get it going was a consumer applicant

Speaker 0:
or a consumer side of it. But actually, then the back end of the history of it then becomes more enterprise. And personally, I think it will go that way. So thinking about search, what is search? Well, it sort of a consumer product and and, you know, the the vast majority of the market share of search. They're kind of habit formed. So I I I don't actually believe that Google will see much, um, much erosion of their market share

Speaker 0:
there in that market. But what I would say is you've got to remember searches like the sort of portal, the entrance, the doorway, the sort of like the chat G BT that you're asking and and basically behind that, what we are seeing is you've got to remember a I, as I said, is just about algorithms and data. So what you're doing is, uh, or algorithms and models. So what you're seeing is you're seeing AAA huge plethora now, uh, of of companies that are becoming specialists in searching, say images

Speaker 0:
or specialists in searching scientific the questions, et cetera. So they so they're basically segregating and deepening the database. So the way I see this is Google will probably sit on the front, and it will recognise that. You know, the search back end has changed dramatically, and it will act as a sort of portal that will reconnect into a number of these very specialist data sets. Because effectively, if you keep a more

Speaker 0:
streamline data set rather than sort of Google trying to search through everything, it's going to be quicker. It's going to be better, and it's gonna be a better service. So I I really see it sort of being a sort of step process going forward in this. So to answer your question, I don't really see Google losing much share, but I do maybe see them losing part of the economics. But But, of course, because they're at the front end only the consumer, I would have thought they'll retain a good proportion of that economics.

Speaker 0:
That makes sense. Um, Mike, uh, your take on this. Yeah, I I you know, habits are hard to break, right? I mean, a one that uses I often remind myself I still use a credit card, Uh, which, you know, which is You know, I just find very easy. I I I have apple pay. I just don't use it. I know lots of millions of people do. So I think that there are There are a lot of you know, there are a lot of pre established habits

Speaker 0:
that people have on search. And in fairness, Google does a great job of giving really good answers. You know, I you know, I can't tell you the last time I used Bing, you know, and I was on a call this morning where I actually asked my coworkers if anyone had switched to Bing. So I thought no one, which, which I thought was a little bit interesting, given we all work in technology. Um, you know, having said that what I really will be concerned about longer term is,

Speaker 0:
you know, there's a whole generation of, uh of of, you know, productivity, uh, users that don't use office, right? If you're if you have Children in school, they primarily use G suite. So as that generation comes into the market, what does that mean to Microsoft? And, you know, I guess the similar. A similar analogy would be as a generation of folks that may never use Google as.

Speaker 0:
So what does that mean, longer term? I'm not worried about, uh, about the Google about Google Lo losing share to, uh, you know, some type of new interface, you know, anytime soon. Longer term. You know, it's you know, it's TBD. Uh, but I you know, I just always remind myself breaking people's habits, um are are really hard when you have a great product. And I would make the argument that Google search is a great product. Um, you know, But that's just my my opinion on it.

Speaker 0:
Thank you. Um,

Speaker 1:
Nick, I I think it's worth, um, considering the thought that

Speaker 1:
with the release of of chat GP t um,

Speaker 1:
Microsoft won the the PR War, you know, by having the first integration with being, um with open a I obviously the the depth of their kind of relationship. And then Google came out with Bard and slightly bungled the launch, you know? And that was reflected in the share price movements. You know, on the day. Um PR. Sorry. Microsoft definitely won the PR war. Um, but have they won the technology war?

Speaker 1:
Probably probably not. Um, and we definitely can't say that at at the minute. You know, the the reality is Google was the first to introduce transformer models. You know, the type of model that drives these, um, large language models and all of the the kind of the the feedback that we've been hearing from specialists in the space is that Look, even if you're not seeing it from, you know, the other big tech companies, everybody is working on this. So

Speaker 1:
we fully expect that everybody will have capabilities in the space. Um, Google, you know, are definitely up there. From what we've been hearing in terms of the technological capabilities

Speaker 1:
they've lost, the the the initial, um I guess the round one. But we fully expect that the the you know what the model and the product looks like in the future, um, you know will improve. And also Google is the incumbent in search as as, um, you know, Mike and Mark have said,

Speaker 1:
And when you are the incumbent, there is a greater risk to introducing new products. You know, Microsoft had a tiny, tiny share in search with with Bing, and so they could afford to take the risk of integrating this product and basically seeing what happens and generating a huge amount of hype. Google Search is slightly caught in the innovator's dilemma because they are the incumbent with so much share, they have to make sure that they, you know, they're not kind of destroying their own market or destroying their own economics as they, you know, increase the functionality and introduce these capabilities.

Speaker 1:
So it'll be really interesting to to see how it shakes out. But I I I agree. I think the the demise of Google in search is, uh is maybe a bit premature.

Speaker 0:
That is fair enough. And And it's not long ago that Google had employees quitting because their internal A I was too much alive. Um uh, the the the I think the expectation was that it had a soul. Um, so they must have. They must. They must have

Speaker 0:
a reasonable, uh uh, quality of of, uh, of technology underlying that. Um, so we'll see how that develops over the next couple of months. If we, um uh maybe, uh, switch over to, uh, to another area, which is a key enabler in the space being ploughed.

Speaker 0:
Mike, maybe I know this is an area you spent, uh, quite a lot of time on. Uh, how do you see the dynamics there does the the advent of of widespread use of generative A I have a significant impact on cloud. Uh, and on the competition, uh, competitive dynamics. Uh uh, Between between the key players there,

Speaker 0:
Well, I mean, I think here again, I think you want I want to delineate here because cloud to me is a delivery modality, right? I mean, it's it's it's it's the way software should be delivered. Um, what I really am focused on is what companies are doing the best job of integrating a I into their products so their customers have the best best results. Um, and here again, you know, uh, if I think about it, it's very difficult for me to find

Speaker 0:
pure play Cloud A. I play. They're just, you know, everything is kind of encumbered. What I really want to focus on is what specific companies are using that that technology to basically give their customers better results better, better, more efficient, better sales. You see, 33 reasons why people buy software, right? Either are going to sell more stuff, have happy customers or take costs out. Right? Those are kind of the three reasons why people buy software

Speaker 0:
works. Um, and you know, the companies that come up to come to mind is you think about a sales force dot com. You think about a service Now, um, you think about, you know, Adobe Just the usual suspects, Um, you know, personally or just professionally, I I feel like service now has done a nice job of, uh, it really fits with kind of the problem they solve. Um, in my opinion, So integrating a I into their their product is really kind of in the

Speaker 0:
house, right? If you think about what a I might do, which is just, you know, making things just kind of some of the less the more mundane aspects of our job that we don't like service now kind of helps that it helps with that solving that problem to begin with, and therefore they've done a nice job of integrating it thus far into into their into their into their technology stack. Um, but, you know, I I think it's going to be iterative, and I think about I wouldn't count out, you know,

Speaker 0:
wouldn't count out any of the big, um uh, enterprise software. I'm not sure which on my screen, Uh uh, any of the big, uh, software, Uh uh, players, You know, whether it's service now, whether it's adobe, whether it's, uh, you know, hub spot. I mean, you can run the game, but I really would encourage, uh, people on this call Think about companies that you that you innovative with respect to their products and then think about how they might apply artificial intelligence. So

Speaker 0:
that is fair enough. Uh uh, Mark, uh, any view on this? And I guess, um, you can include any view on on sort of enterprise, uh, as well. There.

Speaker 0:
Yeah. I mean, I you know, I do think that the the the players with scale will, will, Will will get more power from this situation. So I do feel that the, uh, AWS and, uh, and GCP and, uh and, uh, will will, it will consolidate their position. One thing I do think

Speaker 0:
it's worth thinking about is, um, on the enterprising, which is slightly different. Answer is a lot of this enterprise application software is designed currently for a human being to use the software to undertake a process which enhances productivity. Um, and then along comes a I. And then basically they a lot of

Speaker 0:
companies I I I'm slightly concerned about. So he's just viewing it as a sort of a a bit of a tack on, uh, and they stick it there and they talk a lot about a I I about how it's an extra sort of side pocket within the software that does this, and it takes the data and it suggests ideas to the human, and it's very collaborative and all this sort of thing. But I just wonder,

Speaker 0:
and what I'm worried about is one could see a situation in the future where there is this sort of conflict between human and a I and and you know it. It may be very promoted in the media press, but it it in an economic element. It it is. There it is there it's machine and man,

Speaker 0:
uh, and and therefore one could envisage in the future that actually, it may not be the human controlling the software using the A I, but actually, the whole software has to be redesigned from the beginning process so that actually a lot of the processes undertaken by the A. I using

Speaker 0:
the software to undertake the process. And I do feel that, you know, of course, when you think about that, um, you probably perceive that's a long way away. And I probably agree it is a long way away, but it could happen a lot quicker than you think. And I do feel that one has to be careful that,

Speaker 0:
you know, your companies in enterprise software and enterprise application software have got a I front, central and core to their future designs of the software rather than some sort of side add-on. Otherwise, I can see them being disrupted, not just in 10 years, but later, when people just say, Look, we're not gonna write software for a human to use, we're gonna write software for an A I to use to undertake this process almost end to end. And the human is the add on just to control how it goes.

Speaker 0:
That's really interesting. Thank you. Um, and we'll get to, uh, our outlook, I think shortly, Um but but maybe just still on the on the topic of of cloud and enterprise. What are you seeing in the competitive dynamics? Uh, and and how does, uh, a I impact

Speaker 1:
that? Yeah, I think, as my said cloud is is kind of the delivery method. Um,

Speaker 1:
and we are, you know, when we're in an environment, particularly with, you know, running inference on these large language models, which you know, if you're not running it on the edge, you're running it in the cloud, the potential to transform. Um, you know, companies how they use data, their data ingestion. How much? Uh, you know, analytics. They're going to run the fact that everything is now moving or the vast majority of workloads are moving to the cloud. You know, that bodes well for for cloud volumes and and spend. Um, you know, over the next few years,

Speaker 1:
with respect to enterprise software,

Speaker 1:
it's It's super super interesting. And it's probably too early to tell because,

Speaker 1:
you know, if you think about what we've Well, first of all, the the

Speaker 1:
chat GP t changed. Um, I guess the paradigm of how we interact with technology and I think that's why it was such a revelation. And it was on, you know, mainstream news all over the world because it was so easy to interact with.

Speaker 1:
And

Speaker 1:
that's triggered a huge kind of, um,

Speaker 1:
you know, wave of potential applications. But we are still so early in what this will look like embedded into into software. Um, in the future, you know, we have one scaled application, which is Github's copilot, which has been a a, you know, a fantastic success. Um, and the the early indications are that, you know, in some cases, this is writing up to 50% of of the code that a coding engineer would put out, and that's, you know, amazing in a in a few months of of, um

Speaker 1:
I guess being rolled out what this means longer term for for software is something that we debate a lot. Because in some cases, we're seeing, um, you know, an environment where the introduction of a I

Speaker 1:
makes the best better. Um, you know, if you think about taking the coding, you know, for example,

Speaker 1:
the best engineers, the best coding engineers might be able to to understand the the A I generated code, you know, more easily and integrate that into their you know their output. And that makes the you know, their productivity, even even better than, you know, maybe the median worker.

Speaker 1:
But if you flip it on its head and maybe look at the creative process where something like Adobe's Firefly allows creatives to generate more ideas or generate ideas more quickly or start populating the early stages of of their process more quickly, then what it does is it actually brings, um, you know it it it narrows the the spectrum of of workers and it and it increases productivity for maybe the median worker. And we are still too early in

Speaker 1:
having these tools embedded into processes to be able to definitively say what the impact is going to be on productivity and or how it's going to shake out. But there's no doubt that a I is going to be transformative for, uh, productivity. I think Microsoft themselves said yesterday they expect 85% of companies to adopt this by 2026. I believe it was, and we could see productivity increases of 25%. You know, the numbers are are huge and in terms of

Speaker 1:
what it means for economic output. Um, how exactly that shakes out, I think is a bit early to say.

Speaker 0:
That is very interesting. Um, so maybe on that point, Mark, uh, in terms of the integration of a i into productivity software And indeed, the, uh, the the co pilot, uh, that it that has existed for github, but it is now being rolled out across the, uh, office 365 properties. Um,

Speaker 0:
how do you see the prospects of that? Um uh, and and And how how do you weigh this, Uh, in terms of your outlook on on Microsoft in particular?

Speaker 0:
Uh uh, the productivity shift. I mean, my my first degree was, uh, in the economics department. I just think the productivity shift for human beings or for the economy, uh, is is material, uh, dramatic. Whatever word you want. I mean, I think, um, I think it the you know, the the the point is what I what I kind of like about the human process of running a business is, you know, the the the A I

Speaker 0:
a. Just to put it aside, I I was I was talking to someone the other day. Who said that? A. I just doesn't understand a lot of human things. Yeah, if you can imagine. People want to imagine it's a human being. So a I can't understand how you would buy something because you're influenced by advertising. And they they don't really they can. They can model advertising. But if they had to buy something, they're just gonna look at what's the best? What is the optimal

Speaker 0:
cost benefit? And they're going to optimise that and whatever advertising is irrelevant. So, you know, if you if you sort of imagine, if you if you put as an A I as a person because people can think better like that and you explain to this a I that the human process of this business has. You know, this person stamps this form and then it goes off to that person. No, you know, no one can do,

Speaker 0:
I think, until that person stamped the form and then it goes there and this whole, you know, endless bureaucratic, crazy enterprise system that's built up over years of history of humankind. It it it can just be removed. I mean, it's just it's super. It's just no requirement even sensible human beings can see that. Never mind a I. So the economic effect

Speaker 0:
to a business that adopts this as I keep saying it's core principle, you can't just attack someone on the side. If you put this as your core principle, you are going to eradicate so much cost at that point, then you can get the product better, and you can

Speaker 0:
start destroying your competitors in the market. So the point is, you don't have an option here. If you're in a competitive market, you have to adopt a I, and you have to get your enterprise streamlined throughout this sort of productivity process. There is no option specifically, uh, if there's anything, uh that you think is worth highlighting in the cyber segment,

Speaker 0:
I had a risk for opportunities and any investment implications on the back of, uh uh generative A I

Speaker 0:
Well, well, I wish cyber was as brief as you just kind of, uh uh, the question out there. Look, I think that, you know, the cyber aspect of artificial intelligence is a is a massive opportunity and a massive risk. Right, because the last thing we want are these deep fake fake models, right that you know, the data is corrupt. To begin with, you put an output. Next. You know, it's out on the Internet.

Speaker 0:
People are reading it, etcetera, etcetera. So I don't think we can understate how important cyber security is to this opportunity. You know, from both a protection perspective, because you want to have a layered approach to cyber security if you're using, if you're creating a I products. But I just think that it really brings in a whole other aspect of it, which is, like, how do you make sure that, you know,

Speaker 0:
I guess you know, the evil is probably too strong. But how do you make sure that that data that you're that you're looking at is really kind of the right the right data that you're putting out to the world and that has a societal impact. So I don't you know, it's not my job to, you know, But I do think about these things. Um, and I do think that the cybersecurity companies have a big role in protecting and making sure,

Speaker 0:
uh, that the output is is secure and is the right output. Um, so I think you'll see you'll see spin there, right? I mean, every, uh, as everyone on this call probably knows every dollar that's spent on the cloud, uh, for cloud work that should be going to cybersecurity. That's just the nature of the beat. So I mean, here again, it's, you know, if I use if I use a neighbourhood analogy, the neighbourhood is a really good neighbourhood to be in long term.

Speaker 0:
It's not going away. Very sophisticated adversaries that want, you know, that want, uh, disinformation out there. Indeed. Um,

Speaker 0:
Mark, Nick, anything to add on that point? Specifically, I I'd just say that for people to sort of have a framework of thinking about how this is going to affect the economy going forward, you have to remember that a I is built on data and a I also tends to work very well with what um is

Speaker 0:
is deemed sort of uncertain or unfathomable by the human being that so it particularly has a strength there. So you know, when you get to an area like cyber, where it's all a bit opaque for a human being, that if you can give it enough data, it it's really it makes a material difference to existing systems.

Speaker 1:
OK, and I I think it's really, um

Speaker 1:
it's really important to keep in mind that, you know, with new technologies there, you know there are There are risks and uncertainties. And I think Mike made a number of very good points. You know, we've seen Italy, for example, has has banned the use of chat GP t for. For the time being, Germany has has mooted it. Um, you know, the eurozone and the EU have very strict data privacy laws, Um, and on how data is processed and where it's processed

Speaker 1:
and there will be speed bumps and hiccups as as these things are, um, are worked out and one of the major attractions of a I, as I think we've all spoken to, is how broadly this technology can be applied. You know, if you think about industrial data, you think about health care data, um, financial data, all of these things that you know, historically, there hasn't been a huge amount of economic value in

Speaker 1:
analysing, you know, it's been very expensive or very difficult to get any value, um or or analysis out of unstructured data. And that's what a I is really, really good at,

Speaker 1:
you know, with the breadth of the applications that we're, you know, looking at in across sectors, you know, no longer just in the tech sector, a lot of the data that we're gonna be using is sensitive data. And that means that, you know, data security is going to be a major concern for all of the companies. Um, you know that that want to generate value out of this. So the whole chain the you know, the whole chain from, you know, the the model to begin with. Um, And we've seen, you know, GP t fours come out with more guard rails, more content, philtres,

Speaker 1:
you know, spanning all the way through the cloud and and the cloud vendors and the infrastructure right down to the users. All of them are going to have to think about data security and data privacy and how data is is ingested and analysed so that we can create the the the productivity and the economic benefits which are undoubtedly there. I think we're all in huge agreement about the potential, but do so in a way that is acceptable for consumers, for businesses to, um, you know, to adopt them, and and that not be,

Speaker 1:
uh, a gating factor. And it's just a hurdle that we have to get over.

Speaker 0:
If I could just, uh, for the brief time that I still have you ask, we've discussed, uh, AAA lot of our expectations. Uh, sort of near term to medium term, uh, across a range of areas and industries in terms of investment implications. Uh, obviously, the majority of people and indeed the markets

Speaker 0:
seems to be positioning towards these, uh, large cap, uh, big tech, uh, alongside semiconductors and a few other places. Uh, for each of you, please, What is an area? Um uh, that you feel is underappreciated at the moment. That will be AAA driver or beneficiary, uh, of this field. Uh uh. Generative A I

Speaker 0:
Yeah. I mean, a couple of things come to mind. I think about, you know, they they have to move data around, uh, data centres. And there are a lot of companies, you know, that really help facilitate that Because, uh, this requires this massive amount of computational, uh, horsepower. So, you know, I think about some of the common equipment companies.

Speaker 0:
Um And you know how they may be interesting, uh, ways to play it. You know, I I really like the idea of, um, of of the companies that really focus on chip design and systems design. And just where the relative importance in this particular, uh, field it just increases, in my opinion.

Speaker 0:
So I think you know, I we already talked about it earlier, but, you know, some of the e d. A companies I think are interesting companies. Um, they're not. Maybe they're not as obvious, um, as the mega caps or as the chip companies. But, you know, uh, you know, leading edge manufacturers and semiconductors are gonna be really important companies. Uh, just to name a few, that's a useful, uh, point of view, Nick. Uh uh. For you. Is there any particular area that you think, uh, is underappreciated? Uh, as it stands.

Speaker 1:
Yeah, I think, um Well, firstly, I think Mike made a good point around the infrastructure, um, there and and how there is more to come. Um, obviously, we're in a a kind of a cyclical uptick for semiconductors as well. At the minute which, you know, as you look further right obviously, um, a I is going to add, you know, volumes and, um, and value to to the sector there longer term. But we're also shorter term. We're we're dealing with, you know, a rebound in expectations of the data centre and so on. So we've got the the cyclical tailwind for for the sector. And then

Speaker 1:
longer term, we have new, um, new architectures coming. So I think that is that's very interesting. Um, I also think that when you consider where the value could potentially accrue, you know, as you look a little bit further out, um, you know, the models themselves might not be, you know, where the ultimate value accrues. You know, we've heard the opinion that

Speaker 1:
actually these, you know, the models themselves will become more commoditized, and the value is going to be in the integration of of these models. And, um, you know, the ingestion of data, making sure that they are, uh, you know, effectively customised and aligned with the the tasks that you, you know you want to run and the value you want to get out of them.

Speaker 1:
So what that means is when everybody's maybe got, you know, access to these models, the value is going to be in who can derive the most from a more commoditized tool. And how do you tweak that tool to get the most out of it? And I think one area that's really interesting is

Speaker 1:
the potential to to turn like, you know, a commoditized tool with fine tuning, which is how you can tweak these models and then putting your your own data, you know, into the model and and layering that on top in your own proprietary data that nobody else has access to. And that is a method of differentiating the model. So we're you know, we're looking a little bit further out, and maybe the market doesn't want duration quite so much at the minute. But

Speaker 1:
the, um, the ability to to layer your own proprietary data in and and create differentiation from the market and your peers that way, I think, is going to be really, really interesting longer term. And and that is possibly where there's a great opportunity outside of the tech sector, because if you think about where companies are sitting on on, you know, on proprietary data, there is obviously a huge amount in in tech, but also within industrial companies within healthcare companies. And that is where

Speaker 1:
we could also see a massive tick up. You know, touching on the the economic side, uh, in productivity, you know, across the workforce, and not just kind of concentrated in one vertical or one silo.

Speaker 0:
Let me, please. Thank you all for, uh, for, uh, for joining us. Uh, and and contributing this was incredibly useful. Uh, that brings to close our panel session on generative A. I, uh our clients, uh, will have access to the, uh, the research report that we published on this topic recently. Uh, and there will naturally be more to come. Um, thank you so much, uh, for joining us, uh, and have a wonderful day.