In a thought-provoking episode of CNBC's tech podcast "Beyond the Valley," analysts Arjun Karpal and Tom Chitty delve into the future of personalized artificial intelligence with special guest Ben Wood, the Chief Analyst at CCS Insight. The episode forecasts the technological leaps expected by 2025 and beyond, highlighting an innovative prediction: the arrival of personal large language models (LLMs) trained specifically on individual user data. These models promise to revolutionize personal digital assistance by becoming intimately tailored to users' preferences and habits while balancing privacy concerns.

The discussion navigates through the technicalities of the predicted personalized LLM, its integration into devices, and its potential impacts on personal and professional lives. Arjun, Tom, and Ben assess the possibility of shifting everyday tasks onto these AIs, touching upon privacy, health, and even how these models could guide lifestyle decisions. Apple is earmarked as a potential frontrunner in individualized AI models due to its substantial data ecosystem and commitment to privacy. Other players like Samsung and Google are expected to compete in this exciting field.

Main takeaways from the podcast episode:

Please remember to turn on the CC button to view the subtitles.

Key Vocabularies and Common Phrases:

1. personalized [ˈpɜːrsənəlaɪzd] - (adjective) - Tailored specifically to suit an individual's preferences or requirements. - Synonyms: (customized, individualized, bespoke)

But this would be truly personalized

2. prolific [prəˈlɪfɪk] - (adjective) - Producing much fruit or foliage or many offspring; present in large numbers. - Synonyms: (productive, abundant, fertile)

The thesis behind this prediction is the fact that we've seen AI becoming more prominent.

3. provocative [prəˈvɒkətɪv] - (adjective) - Causing annoyance, anger, or another strong reaction, especially deliberately. - Synonyms: (challenging, stimulating, controversial)

We don't do stuff that's deliberately provocative.

4. revolutionize [ˌrɛvəˈluːʃəˌnaɪz] - (verb) - To change (something) radically or fundamentally. - Synonyms: (transform, innovate, overhaul)

These models promise to revolutionize personal digital assistance by becoming intimately tailored.

5. caveat [ˈkæviˌæt] - (noun) - A warning or proviso of specific stipulations, conditions, or limitations. - Synonyms: (warning, caution, condition)

With all of the user caveats around privacy and using that to create a model.

6. augmented reality [ɔːɡˈmɛntɪd riˈælɪti] - (noun phrase) - A technology that superimposes a computer-generated image on a user's view of the real world. - Synonyms: (AR, enhanced reality, virtual imaging)

We say that Apple will launch augmented reality glasses.

7. symbiotic [ˌsɪmbiˈɒtɪk] - (adjective) - Involving interaction between two different organisms living in close physical association. - Synonyms: (mutualistic, cooperative, interdependent)

It's a symbiotic relationship because, of course, all of these AI models get better through scale.

8. dichotomy [daɪˈkɒtəmi] - (noun) - A division or contrast between two things that are presented as opposites or entirely different. - Synonyms: (contrast, difference, division)

It's a dichotomy as kids already make friends with inanimate objects.

9. formidable [ˈfɔːrmɪdəbl] - (adjective) - Inspiring fear or respect through being impressively large, powerful, intense, or capable. - Synonyms: (intimidating, daunting, impressive)

Apple is a brand that has a certain amount of kudos in terms of being a product that people are proud to have.

10. frivolous [ˈfrɪvələs] - (adjective) - Not having any serious purpose or value. - Synonyms: (trivial, petty, foolish)

You know, things like pet rocks. What a nonsense.

Ever wanted a personal assistant? An AI one could be around the corner

Welcome to beyond the Valley, a CNBC tech podcast with Arjun Karpal and Tom Chitty. Have you ever wanted a personal assistant? Someone that can manage your diary, arrange travel, make notes in meetings, but also knows you personally? What your interests are, your habits, your likes and dislikes? Now, what if that assistant was an AI large language model trained on your individual data? This is just one of the many ways that future technology could impact our lives in 2025 and beyond, according to a list of predictions that have often become a reality. Beyond the valley.

Our guest this week is Ben Wood. He is chief analyst at tech research firm CCS Insight. In 2021, he was ranked as one of the top five tech analysts in the US and Europe by Apollo Research. Ben is also the founder and co curator of the mobile phone museum. But this week we're going to discuss his work leading CCS insights annual predictions event. Ben, thanks for joining us on beyond the Valley. Again, you are the first repeat guest we've had on the podcast. Well, I'm flattered to enter that hall of Fame. Thank you very much for having me back.

We're going to get a plaque of some sort, I imagine. Archie. Yeah, we'll start. We did promise maybe to give out mugs, didn't we, to some of our guests? That hasn't materialized yet, but we'll work on that. We only got three or four, so maybe third time supply chain issues. Yeah, exactly. Exactly.

So, Ben, let's get into it. Talk us through this event and how you come up with these predictions. So I've been in the analyst game for quite a while and I don't know, about 15 years ago or so, there were a few companies who were doing the odd tech prediction, but I felt there was an opportunity to do it in a more structured manner. We had a team of analysts and quite frankly, our job is to provide insight into where technology is going. And if a team of analysts can't come up with a few interesting predictions, then we're not doing our job properly.

So we started putting together a short list of predictions about how we saw things happening in the next year in the kind of five years out. So this year, for example, it'd be 2025 and beyond. But the charter for the predictions was they have to be definitive. It can't be kind of like might, could, would I like people to actually put it on the line and actually commit to something? Also, they have to be believable. We don't do stuff that's deliberately provocative. I think that's a bit of a fool's errand, but we will put things in there which might raise people's eyebrows, because if you just go for low hanging fruit and say, the sun will come up and grass will be green if it rains and those sorts of things, it's a waste of time. You have to have stuff which is potentially a little bit more edgy, and that means you will get things wrong.

And then the final thing that we put into it was the fact that we, we're always happy to, after we'd done our predictions the year later or a couple of years later, we'd go back and reflect on them and say, here's some we got right, but more importantly, here's the ones we got wrong. And why do we think that was the case? Well, we're going to get into those later in the episode, but first, let's talk about the prediction that we're most interested and explore it a little deeper before we do that, though. Stat of the week. Ben, you played last time. I can't actually remember the result I got in the ballpark, but I didn't get it exactly right. It was a Samsung stat. Yeah, I think you were really far off as well, Tom, as. Yeah, as usual, but that's fine. And remember, a few episodes ago, I just want to make this very clear. You were disqualified. Well, for looking at Google during the salad. I think it was reviewed and inconclusive. Okay, fine. There's still an inquiry going on as we speak, but the start of the week this week, $990 billion. $990 billion. It's a big one. It's a big one. Fantastic.

All right, so the prediction that we're going to discuss in the main for this episode is this. So the first personal large language model for AI arrives in 2025. A personal large language model is trained using only data specific to an individual user. In contrast to a standard model, it is tuned specifically to a person's communication style, needs, and preferences. It is trained using the individual's data based on previous interactions such as emails, messages, documents, and social media interactions. The model is continuously updated by drawing on the person's evolving data sources. What's the thesis behind it?

So the thesis behind this prediction is the fact that we've seen AI becoming more prominent. We're certainly seeing it becoming very much an on device phenomenon centered around the smartphone, and we're seeing the addition of things like contextual awareness. So if there's something on the screen, the AI model can look at it and it can help you provide information on that, something that we're intrigued about and excited about is the idea of taking all of the personal data that people have, and of course, with all of the user caveats around privacy and using that to create a model which is more tailored to that specific individual.

And if there's one company that's well positioned to do this, it's Apple. Because of the unbelievable repository of personal data that people tend to have on an iPhone or in icloud. Because Apple have been fantastically good at taking people on a journey of keeping all of their data from one generation of a device to another. And what this means is you would then be able to get some amazing results out of this kind of model.

So just take for example, some of these scenarios we see about asking AI to create a holiday for me or those sorts of things. It's still relying somewhat on some sort of general algorithms around what people like you like. But this would take it right down to that specific view in terms of where you've been in the past, what you like, perhaps doing analysis of pictures you've taken so you can tell who you like, going with things excite you. It could be a remarkable way of delivering context and for other things, just prioritizing important things in your life. If things are appearing in your email that maybe haven't been there for a while, but they all of a sudden are super hyper relevant and you need to know about them, it could surface those better than a general model.

So this would be something you see working on devices at this point, whether it's a smartphone or a tablet, laptop, etcetera, using all that. So in the holiday example, that could be something like scraping sort of your photo day and saying, well, you got a lot of pictures of food when you're on holiday. Specifically, you're really into food when you go abroad. Maybe we should throw up some good restaurant recommendations for you. Or when you go on holidays, a lot of beaches you like, a beach holiday. Is it that kind of way that it will work?

Absolutely. And then down to all of the individual things, you know, looking in your email to see what hotel reservations you've made and making sure that it's the kind of profile of hotel that you like, looking at the kind of experiences that you like. And, you know, make no mistake, we're seeing this with AI already. You can get some general information in this context and it will draw on some of your personal information already.

But this would be truly personalized. And the reason it needs to be on device is because it means that it's ring fence from a privacy perspective. You don't want that deepest personal data all being relinquished up into the cloud and then to goodness knows who. So can we just talk about that from a technical perspective on how it will work? Because we did an episode recently on large language models, and we're talking about all of these large language models being trained via these huge GPU chipsets, data servers, et cetera.

And then obviously in the cloud, you can access the services that are a result of them, whether it be chat, GPT, or something else. So that we understand. And when we're accessing some of the current smartphone digital assistants, whether it be Google Gemini, Samsung's Bixby, or the Apple Siri and Apple intelligence, when it comes again, it's not fully on device yet, a lot of it will still be coming from the cloud. So when it comes to a sort of personalized LLM, how would one train, or how would a company train that and then sort of deploy that to the user?

Well, I think it's interesting. We obviously continue to see the processing power on devices increasing. We're seeing the size of the models growing. So the latest Gemini nano model, 3.8 trillion parameters. Unbelievable. I mean, it's staggering, staggering numbers. So that will help. There will undoubtedly have to be some external assistance, but the idea will be trying to anonymize some aspects of that and keep the really most personal, private data on your device. But then go and get the context with regards to some additional elements which could be hotel selection, food types, those sorts of things which can be anonymized. It doesn't have to be so specific, but you've done that original legwork on the device itself.

You mentioned Apple being the sort of company that you would earmark as the that would lead on this. But surely Samsung people put all of those things that you mentioning, photos, personal details, things like that onto those devices. Why Apple? I think they're in a better position to do it because they've always anchored everything they do around a kind of privacy promise. So although Samsung and Google would be eminently capable of doing this, and I think they will go down this model, and I think other people will as well, meta is the perfect example. If you think about a stream of personal information, it doesn't get richer than that in many cases for particular users, but there'd always be that suspicion if it starts being positioned as a personal model that there could be some kind of leakage.

So it's not to say that this would be exclusive to Apple, but if you look architecturally at how they've looked at delivering Apple intelligence with an on device model enhanced with a kind of Apple secure cloud based on their own silicon and the ability of them to control everything from end to end, they're probably the best position to really push the beachhead out in this area. Is it also the reputation as well that is really crucial in this area? That reputationally, maybe they are seen as more open or transparent with what they're doing with their data than other competitors, and that's why they would be trusted with your personal data. Is that something?

I think reputation helps, but I'd go back to the point I made as well about the sheer amount of personal data that they have and the fact that they control all of those elements. So whether it's their photos app, or it's their email app, or it's the context app, all of those assets reside within that Apple ecosystem, end to end, even the silicon designs, all of those sorts of things. If you're Samsung, you're partnering with Google for some stuff, you're using Qualcomm for other things. There are all these different aspects that you have to take into consideration.

And let's not forget, the success of many of Apple's products has been that personal dimension to them. The fact that it kind of just feels like it just works. It's very intuitive. And of course, something which is really optimized for you as an individual and is drawing on your own personal context could help deliver on that experience that Apple have always prided themselves on with Apple intelligence. About what you mentioned, that sort of private compute cloud where they say that any of the data being processed in our cloud won't be stored, and so therefore it's a lot more private.

And we as Apple won't be able to look at any of that data. If these sort of personalized LLMs do come to the fore, will the likes of Samsung and others need to do something similar in order to compete in many senses, with the privacy aspect, as we do become more concerned about the privacy, particularly in this age of AI and data? Well, some of this comes down to business model. But let's not be under any illusion that Samsung, Google, meta are all making significant promises around privacy and respecting your individual information.

So they're doing that equally. We've got the regulatory angle, which is quite interesting. And certainly here in Europe, it's a lightning conductor for kind of controversy at the moment. Looking at capabilities not being rolled out, we've seen Apple intelligence not coming to european markets recently. We saw meta at their connect event, saying that their AI services won't come to Europe. So I think there's going to be a number of factors which could kind of gate the way in which these sorts of services could be delivered.

So whether it's the companies themselves who want to commit to this, what framework they have to deliver that in, but then this desire to deliver this truly personal experiences, which is when you get to the point almost, I want a phone to almost know what I want to do before I'm going to do it. And there are little elements of that that we see today. For example, on my Samsung phone, it often can have the intelligence to know if I'm looking for a particular app, it will surface it for me at a particular time of day. Google Maps knows on a Tuesday evening, I always go somewhere to drop my son off for something that he's doing. And magically it always puts that particular destination towards the top of my search library. So there are personal elements to many of the things that we do on smartphones today, but the idea of having a personal model is a very exciting concept.

Can I just ask, do you think people will want that in the sense of, you know, is it too far? You know, people always talk about, Arjun and I were discussing before, before we started this recording, what are we going to do with our free time? My point being is that when we leave work, a lot of the stuff that we're busy with is the day to day admin of life booking, organizing, you know, scheduling, whatever it might be. And this, what you're saying is sort of taking that all away from you and doing it for you, which in hindsight sounds great, but then what are you going to do with your free time?

I think there's better things you can be doing than worrying about scheduling stuff. I'm not convinced this is going to be something that people are going to be queuing up to get. I think it's an evolution that we're going to see with AI as models become more personal, and it could be a hybrid model to start with, insofar as there's an element of your personal information being ingested, which we're seeing on all models today. But this takes it to an even deeper personal experience. And I think it won't be that people go, I really need one of those personal LLMs. It will actually be that they'll go, wow, look at this. My phone's done this for me. And it's just so deeply personal that it's just a very, very, you know, pleasant and joyful experience. That you get because you all of a sudden you think, gosh, that's fantastic. That's actually surfaced something that I wouldn't have thought of or done, something that's made my life a little bit easier for me.

Yeah, I guess the way a lot of these software updates do roll out, incremental, slowly roll down. So, I mean, to some extent, we already use some of the email apps, and they're kind of suggesting sentences or trying to replicate your style of typing and sending an email. So those features are slowly creeping in. We just want to go back to the leading players situation as well. Because you alluded to the sort of success of Apple being around its end to end control of everything from the hardware right down to the silicon through to the software, is Samsung's problem as one of the biggest, if not the biggest, rival to Apple around the fragmentation in its software side of the business for so long. Right. Apple software is so tightly integrated to what it does with the hardware.

The key apps on the iPhone are all Apple apps, whether it's the mail app, the notes app, phone messaging, etcetera. With Samsung, you have the choice. Right. When I use a sort of Android phone, it's often Google services I might use, and I might use one or two of the phone makers own apps. But then my whole experience is to some extent fragmented, and you end up sort of siding or trying to side with one of those. So at least you've got a sort of coherent stretch of software across multiple devices, et cetera. In an age where the LLMs are coming in more personalized, does that software fragmentation pose somewhat of a problem for Samsung in being able to deliver an experience that may match up to the Apple experience?

There's no question that's a challenge for anyone in the Android ecosystem. But Samsung's the biggest, and they're the ones that have that biggest challenge. The great thing for Samsung, however, is their scale and their reach. And let's not forget, there'll be 200 billion Samsung devices that will be able to use Galaxy AI by the end of this year. They are in pole position with regards to scale and reach, and it also gives them tremendous leverage over Google. So you can be sure that Samsung are in Mountain View banging the table, saying, we have got to work as closely and deeply together in order to deliver the best experience we can in the face of a competitor who's deeply integrated. So, yes, it is a challenge, but if there's one company that's in a good position to take up that challenge, it's Samsung, because of their scale and reach. And it's a symbiotic relationship because, of course, all of these AI models get better through scale. And so there's a huge incentive for Google to work closely with Samsung, setting Pixel to one side. But you could argue that's somewhat a kind of out of the possible in the way that Microsoft did surface to really kind of kick the PC manufacturers and say, come on, you've got lazy. You really need to push this forward.

From a technical perspective, one aspect to think about is if we're sort of looking at the privacy portion of it, and we're saying, well, a lot of this needs to happen on device. We often these days, or many people often use multiple devices, right? Whether it's from your phone to a tablet to tv to something else. How then do you see sort of these personalized LLMs migrating across device right now, you do something on a phone, in an app, and instantly you open up your tablet, and that app is on there with whatever you've done, whether it's notes or a doc or whatever. How does that experience? Do you see that happening out, playing out in the personalized LLM age?

Well, it's a much easier thing to do if people kind of hook themselves to a specific ecosystem. And it's little surprise that companies like Samsung, Apple, obviously have a very strong ecosystem of products, but also other companies are chinese manufacturers. Look at what Xiaomi are doing in terms of trying to extend their reach across multiple parts of your life, whether it's smart home devices, your vehicle, your car, your devices. So in that kind of context, it works really well. Interestingly, that could be an opportunity where Google could become the glue, because it's a kind of omnipresent player meta the same. But then you do start to run into some of these issues around the fact that they're going to have to give you very, very clear commitments around how they're going to handle the data that resides in that personal model. Because if you've got to transfer it to one device to another, you've got to find a way to do that with a certain amount of data integrity.

So we don't know yet. But the point is, from my perspective, I have seen, and you will have seen, because you talk to many of the same people that I do, all of these companies are pushing this idea of an ecosystem of devices and services. And you know what? If you're a company that's delivering that, if you can get people locked into that world, my God, they're loyal to you. I suppose the next question is, how far does this go? Because I'm imagining a world where you have this large language model AI, sort of managing your life, but also maybe starting to give you advice on how to live your life. An example being that you go out for dinner with some friends and it's aware of the last time you did that, you had a massive hangover. And it says, you know, just be careful how much you drink tonight. They become overzealous. They become a bit controlling or telling you kind of how to live your life.

Do you think that's a world that we could end up living in? Well, I think with a personal model, you'll be able to tune it, so there'll be an opportunity for you to say things you like and dislike. So as you see on many of the services, you can tell you say, you know, why am I seeing this ad? Do you want more of this? Do you want less of that? I think that's a way in which you'll be able to fine tune it yourself. However, I do think the dimension that you're touching on there, which is the intersection of health and this kind of personal information, is deeply important. And we are so excited about the intersection of personal health and AI. And if you look at what's going on in the world, you look at governments looking at how on earth they're going to manage this really challenging problem around unhealthy populations, aging populations.

You need to get more into predictive health, personally incentivized to take better note of that, and a very personalized model that looks at your health aligned with and intersecting with a strong privacy promise, because you're going to get quite nervous if you start to get a lot of your health data being shared with your insurance companies and those sorts of things. If you want to do that and you think it's going to get you a better policy, that's great. But if you're super healthy, that might be a great thing to do. But I think that this idea of having something that can nudge you along from a health perspective could be fascinating. And we're seeing that. A great example of that is what aura are doing with their smart ring. Their app is becoming more and more personalized.

Things like symptom trackers that they're looking at and they can actually say to you, I had it recently, I've been out traveling. If you've been on a plane for too long, the chances are you might pick something up. The symptom tracker is saying to me, well, you're getting sick 12 hours later. Guess what? I've got a sore throat. I'm not feeling great. So those sorts of things which are very, very personal to you, intersecting with some of those other things could be very, very interesting. And again, the personal large language model is aspirational. Some of that very deeply personalized experience will start to surface itself very prominently in some of the services we're seeing today, too.

And you're seeing companies set up for that, right? You mentioned Ura there, the Samsung ring. I promised I wouldn't talk about this on this episode, but one of your predictions. I was going to save this for another day, but one of your predictions is Apple will introduce its own smart ring by 2026. Might as well, since we're on the topic, give us a brief of that. Well, is it a no brainer? It feels like a no brainer for Apple, given their sort of imperative on health. Well, just to be clear that there is a previous episode, which we recorded with Ben, which was on this, on Samsung's ring. But. And you did mention. Yeah, you did mention the apple.

So there is no question in my mind that health has become a fundamental pillar for apple. In fact, I would go as far as to say at the point that Tim Cook decides to relinquish control and he retires, and someday I'm sure he will, I would like to think that one of his major legacies from Apple would be around personal health. And if you go all the way back to the inception of the Apple Watch and simple things like the sedentary lifestyles that we have and that little tap you get on your watch, and now you get it on other wearables as well. Fitbit were doing it previously. It wasn't a new thing from them, but they popularized this concept of get up, walk around. They looked at things like looking at noisy environments and alerting you to that. Then we saw some of the stuff around heart health, and, you know, atrial fibrillation was a good one. We've seen sleep apnea introduced now. We've seen hearing introduced.

Now. Given how deeply and personally invested Tim Cook is in health, I think that a ring is a very complementary extension to Apple. I also think that their retail footprint is ideally suited to it as well, because rings are complex products to get to market. We talked about this on the podcast. You know, three different skews, nine different sizes, yada, yada, yada, all in there. So on that basis, yeah, we think Apple could do it. And also, Apple is a brand that has a certain amount of kudos in terms of being a product that people are proud to have. And I think a beautifully designed ring from Apple could be one of those things that it's almost a bit of a status symbol and opens up potentially more revenue opportunities through services and health.

Right, absolutely. And again, another part of an ecosystem of devices and services that locks people into to that world. And that is exactly the same philosophy that Samsung are taking and why they've gone into this space as well. There's a lot more predictions that we could get into, but we don't have time. But I'll touch on a couple of others that we read, which was by 2030, half of UK school children identify a generative AI avatar as one of their best friends. And another one which we were interested by was again, by 2030, a country appoints the first AI agent into an administrative role.

I think, again, it touches upon this sort of bleed between the sort of technology and your personal life. And essentially they've become more humanized in their performance and what they're able to do. And that relationship, right, between the humans and the AI. That's what it's touching. Yeah. I mean, the kids one's interesting, right. Let's be under no illusion of the fact that kids already make friends with inanimate objects. We've got roadblocks out there. We've got pieces of plastic and rocks. When I was growing up, that's a long time ago, there were things like pet rocks. What a nonsense.

But kids will do that. I think that as social platforms like Roblox, like Minecraft, gain AI co pilots that they can work with, you can see that the relationships that kids who are often very much at the bleeding edge of this kind of capability, could become engaged in that. Now we can go into the whole question in terms of, are parents going to like that, the societal issues of it, all of the concerns around it. But I don't think it's out of the realms of possibility that something like that could happen. I think the other one was the AI agent in an administrative role. And we cited the example of El Salvador back in 2021, adopting bitcoin as a currency. And with the ever increasing capabilities of AI, it could be that a country somewhere rolls the dice and decides to use AI to drive some of the kind of legislative agenda. It doesn't mean to say that it will have the final word on it, but it will definitely, in our view, become an input into broader governmental direction.

Let's look at then, some of the past predictions that you've made in previous years that you've got right. Yeah, it's always nice to talk about the ones you get right. We'll start with those. We get some wrong, too, as I said, right at the top of the show. So, I mean, there are some famous ones. One we did was we predicted that Motorola would be acquired by Google, the Motorola devices business. And I will always remember that prediction because of the fact that as a result of that prediction being published, we got some extremely angry phone calls from senior management in the Motorola devices business saying that what an outrageous thing to do and what a ridiculous suggestion. And of course, a couple of years later, that one came to fruition.

So they're the ones that you kind of remember and you like. We had another one, it's quite UK specific, but we had a prediction that three or Hutcheson would make a bid for zero two. And that one, we even had the CEO of zero two in the audience at our predictions event and we kind of announced this. He was great. And he actually put his hand up and he said, you know what, guys? I'm going to sleep well tonight. I'm not worried about this at all. Two weeks later, boom, it happened. Now, you don't get many like that, but then there are other ones. Some of the ones a bit like the apple ring one we've talked about. Apple's a good one. It's a hot topic conversation.

We predicted last year that they'd extend their satellite capability to offer imessage over satellite. That came to fruition. This year we've done some broader brush ones. So back in October 2020, we talked about the fact that regulation on AI and big tech more generally would be driven by the EU. And I know that seems blatantly obvious now, but five years ago it wasn't as clear. I think other things, trends going back another 120 21, talking about how green credentials would become an anchor part of the way in which consumer electronics would be positioned. There was some early, you know, real leading edge companies doing that, but now every single player watch the Samsung keynote, the Apple keynote, whatever, they're all anchoring stories around that.

So they're the sorts of things. Some we get partially right. We say that Apple will launch augmented reality glasses. We said that back in 2019. It took a lot longer than we anticipated. And do we want to talk about some of the ones that you may have got wrong, or did you? Oh, absolutely. No. We're always happy to talk about the ones that we got wrong just for balance. And I think it's hugely important to do that. So if we think about some of the ones where we've got it wrong, a nice one that I liked and one I was really bullish about, was the concept that back in 2022, we thought an indian smartphone maker was going to break through and become a global force in the smartphone space. There was lots going on in India. You were seeing all of the initiatives from the indian government in terms of really pushing manufacturing. And you know what? It just burnt out. Nothing happened. Because what actually happened was there was so much imperative for the big companies like Samsung, the chinese manufacturers, even Apple now have huge amounts of production capability in India to move into the market that they completely obliterated the in country manufacturers. And now, you know, Apple is exporting loads of product and Samsung exporting product out of India. So that was one that we absolutely got wrong.

Another 1 October 2021, we thought that Apple would launch their own social network. We thought, well, what a logical thing for them to do. There was all the hysteria around what meta were doing and there was this. We felt there was an opportunity for someone to kind of rise above it all and go, we know you want a social network, but with all of those kind of guardrails around it, that you're not going to have all of your information sold and everything else. Of course, it didn't happen. They did have a social network a long, long time ago. You might remember, Arjun, they had a social network around music, which disappeared very, very quickly. I think it was back in 2010, you can see that. And then you almost feel at that point in time, with the EU sort of getting more and more strict around Internet companies and social media in particular, look what they have now in terms of regulation. Part of the thinking was probably like, we're already being looked at so carefully, we don't need another thing out there that regulators can latch onto and sort of pull us up for.

That may have been part of the thinking, if they were thinking about any kind of social media problem as well. I just think we got it wrong. All right, thank you for trying to support me. I mean, I honestly, you're right. I think that was the case. But at that moment back in, you know, the time that we made the prediction, it did feel like there was a gap in the market for someone who'd nailed that. And then there's the ones that haven't come true yet, but we thought they would have happened. And one I really like is we'd predicted that by last year, so 2023.

So this was back a few years ago. We predicted that a network operator would have this fully operational swarm of 5g enabled drones. I had this idea of. Well, actually, it wasn't me, it was one of our guys in our networks team. But this concept of being able to throw up a network if there's been a humanitarian disaster, for example. So these drones just kind of like, whoosh, take off and just land anywhere they can on half destroyed buildings or whatever, and set up this kind of amazing 5g network which extends off the network that's still working. That hasn't happened, but it still could.

But at the moment, it's one that we've got wrong. But I live in hope. Is that still a possibility? Even as we increasingly see things like Starlink and satellite services? Is there still room for that kind of product? Well, that's a very good counter. And you could argue, you know, is there? Maybe there isn't. Maybe Starlink has usurped that. And, you know, as a Starlink user, I mean, it blows me away. It's an incredible service, but it just felt like it was such a kind of brilliant way in a crisis situation to rapidly deploy coverage.

But you're right, it could be that we have. That's one of the commercial reasons why you wouldn't need to do that anymore. Fantastic stuff. Thank you. Ben. Before we let you go, stat of the week, which Arjun's going to remind us of, $990 billion, who'd like to guess whether you want to go first or second? Well, I have seen some absolutely enormous numbers around the investment that's going into AI infrastructure. I'm not sure it's the right answer, but if you looked at the collective investment in AI infrastructure from the hyperscalers, I could believe that's a stat around someone estimating how much they're prepared to invest over a period of time.

But I don't think it's the right answer. The amount of money invested in large language model models last year globally. Both wrong, but both correct in referencing sort of AI. It was a very, very recent report, actually, from Bain and company valuing the market for AI products and services by 2027. Right. There you go. There we go. Fantastic. All right. Well, at least I didn't. This, of course, includes. Includes, you know, the LLMs and chat bots and potential agents. I'm just relieved I didn't lose. I just, you know, lost with you. Oh, thank you. Rather than on my own.

Lucky it wasn't another week of disqualification, to be honest. Yeah, I just. I still can test it, you know, we'll talk about it again. I'm sure it's going to keep coming up. I don't think you're going to live. Okay. That's, that's all we've got time for. If you do want to contact Arjun and I or have a question about this or a past episode, us@beyondthevalleymc.com please follow and subscribe to the show, and you can even leave us a review. Thank you, Ben. Pleasure. Thanks very much for having me again. Thank you, Arjun. Thanks, Dylan. We'll be back for another episode of beyond the Valley. Goodbye. Beyond the balance.

Technology, Ai, Innovation, Apple, Personalization, Data Privacy, Cnbc International