Episode 12: The Telecom Pulse: Smartphone Prices, Memory Shortages & Market Shifts

Peter Jarich (00:00)
Hi everyone, and welcome to Intelligence and Noise, the podcast from GSMA Intelligence, where we take a look at the week's news. We try to sift through the noise and pull out some intelligence. I am Peter Jarich. I am the head of GSMA Intelligence. I have Tim Hatt with me. I don't think I've ever even introduced you other than just saying you're Tim. Tim. Tim heads up our consulting and...

Yeah, he's been here longer than anyone else in the GSMA Intelligence team, so he's the one to know, right? It's on your card.

Tim Hatt (00:32)
Is that actually true? I didn't know that.

Peter Jarich (00:35)
I've seen your card. All of that says it on it, I mean—

Tim Hatt (00:37)
Not always. It does, like, "He's been around a while and does something useful." But we'll just leave it at that.

Peter Jarich (00:42)
It's like—

Tim Hatt (00:43)
Yeah.

Peter Jarich (00:44)
Two-point, two-point font. They fit it all in there.

Tim Hatt (00:47)
Yeah.

Peter Jarich (00:48)
So thank you all for joining us. We've got a bit of holidays coming up. We allow Tim out of the office, and I've got to go back to the US so they know I'm still around. So we're going to be testing out a new format here for a bit. We'll be doing two shorter ones, still trying to bring you things that we think are important, some things that maybe you've missed.

And we can tell you why they're important. But if you like the format, let us know. If you don't, let us know.

As usual, I'm going to kick things off. So I think I mentioned this just two seconds ago. Tim and I are off to holidays—I'm off to holiday and headed to the US, headed to Mexico—and my phone... I mentioned this a while ago. My phone, I decided a while ago I'd try a new strategy: to go for a mid-range. If it gets hurt, I don't mind too much. And then, of course, RAM prices go up and everything, and then, of course, I do mind.

And so I've still got my old phone. Problem is, at least with the mid-ranger I picked up, no eSIM, which isn't necessarily a problem when you're not going anywhere. Even if you're going to Europe and you've got great roaming throughout the EU, it means I can't take advantage of all the cool travel eSIM stuff.

So I spent some time over the last week with—I use this term for someone, and they said, "Yeah, clearly, Peter, you have no idea what you're talking about." Physical eSIM. I bought myself a physical eSIM. And yes, for those of you who know what an eSIM is, you might be like, "Wait, that doesn't—there is no such thing as a physical eSIM." eSIM stands for embedded SIM. Like, what do you mean you bought a...?

But these are folks that are out there that it is a reprogrammable SIM, is probably the better term. It doesn't work perfectly. Programming it, getting the SIM profiles onto it is not easy. It's easier for Android than Apple, I believe. All the Apple options I've seen include a reader, so you can download and upload the profiles separately.

It's a lot of hassle, which makes me think more about getting that new phone just to solve for that.

And on that note, speaking of new phones, this leads us into today's—or this episode's—news.

So there's a lot of device news. I think it's a lot of device news for a couple of reasons. One, we get to that cycle where people are announcing new phones, but also we get to the cycle where all the analysts who track device shipments, who track what's being bought, what's not being bought, are releasing their quarterly updates.

So, Tim, you may have seen this from MWC. Do you remember what my favorite device—and everyone's favorite device teased at MWC—was? Can you do—you do remember back in the heady days of February, March, everyone's... the device that stole everyone's heart?

Tim Hatt (03:59)
Was it a three-times foldable device? Is it not?

Peter Jarich (04:04)
No, it wasn't. It wasn't. No, no, that can't—that's not cute.

It's the Honor Robot Phone.

The Honor Robot Phone, which was actually teased in 2025 because we started hearing about it, and then it was shown off at MWC and then officially launched. It's going to be available in China, of course—only in China—from the 18th of August.

For those of you who don't remember the Robot Phone, the Robot Phone is—and just go Google it. The video is impressive. It's a big phone because it's got a camera with a gimbal, a robotic gimbal. It'll take photos, videos, auto-tracking like you would with the gimbal.

They've got some sort of an auto-execution system, the YOYO Auto-Execution System, which is supposed to be able to identify your intent and carry out actions on your behalf. They're going to open up the ecosystem so other folks can take advantage of that gimbal.

I'm not—God knows how long it's going to take for third parties to leverage the Open Agent Skills Interface. But I think, to its credit, China has obviously been all about open on the AI front, so it's probably better that than not.

That was the first big one. For those who didn't see it, it's priced at about 10,000 RMB, $1,500 thereabouts, for the entry-level edition. It has 12 gigs of RAM and 512 of storage, so not a cheap one, of course. The higher-end version, with more RAM and more storage, is about 30% more expensive.

So look to that around $2,000, which seems to be the new high end.

And speaking of new high ends, then the other big launch: a whole bunch of Pixel devices from Google. I think fan favourites for folks who love the full-on Android experience. New processor. You know, feels like the headlines here: new processor, a lot more AI features.

The processor supports 25% faster web browsing, 50% more TPU compute to support all the cool AI stuff that Google's going to put in. Of course, they're including six months of Google AI Pro with it, which makes sense, bundling it in. And, not surprisingly, they're all about $100 more expensive than last year's model.

So it's interesting, I think, from my perspective, Tim, and we're going to go back to some other device news. The real sort of—if I'm looking for a connection here, right? It's that devices are getting more expensive.

Not too surprisingly. Feels like $2,000 is now the default for your top tier, something we wouldn't have thought about that long ago, partly driven by RAM, partly driven by new features. And then, of course, everyone's trying to look to find new ways to shove AI into all of these, every single way possible.

Other than new colours, have I missed any other cool—what do you take as what this tells us about the future of devices?

Tim Hatt (07:27)
No, I think you brought up a good point, which is we're kind of at an in-between point here in the device cycle, in that for years smartphone shipments have been kind of sluggish. They've been moderating. People have been talking about where's the next big thing, the killer use case and all of that.

And the reality is that there has not been an iPhone moment in this generation of devices to inflect the overall trend. We're now at a point—I call it in-between—because, as you say, AI becomes the focal point.

And that now is being incorporated in many different ways. We know that the device makers are looking at ways to increase the processing and the latency gains to help with AI usage among people. It's also the ability to do new things.

When we did a benchmarking exercise of the silicon makers and a survey of device makers, we looked across at various use cases, from AI search to VR and other things. And actually, whilst they're ready on some of the use cases, you know, they're not on others.

So I think it's a bit of an experimentation phase. And when you have that, as you say, prices go up, functionality goes up, but the corollary is that shipment growth is declining. And the latest numbers in Q2 are, depending on the estimate you take, about 10% to 15% down year on year.

Peter Jarich (08:47)
Yeah, well, and I think what's—to your point, and this kind of goes back to what you're saying, sort of the guess in terms of a tri-fold foldable—is, you know, foldables in part were an effort, or at least you could see it as an effort by the industry to make devices a little more interesting, give people a reason to go out and buy a new device.

You know, I said this when they first came out. Regardless of whether people were buying them, it's driving them into the stores. "Hey, I want to go play with this thing." Maybe I pick up another device. Not ready for it yet.

Those are still a minor share of the market. They clearly have been taken off. Foldables. So AI, I don't want to say fills that gap, but is another—you can see it as another way to get people to buy new phones.

And I say it not because you could just say, "Well, no, no, AI is just responding to what people want. People want AI, and so we're putting AI in there." That's true of better cameras. That's true of bigger batteries.

But you look at our consumer survey data, and if people are out there and they haven't seen it, please go check it out, because we break out all the consumer survey data into all the different aspects.

And what you see on the device side of it, AI is not top of the list, right? It's not like they're clamouring for that. They're clamouring for cheaper devices, better batteries, better cameras. And you could argue that the camera side of things isn't even that important in the real world.

So whether or not the AI of it all really matters, don't know. But I think it is also something you have to put in there, because if you don't, you're going to be like, "Wait, what are you doing there?"

The flip side is, and you kind of talked about this about what this means in terms of the market. The other sort of—the other aspect to this, this device side of the news cycle this week, is that you've got all the research shops releasing their quarterly updates.

So, looking back to Q2, I don't know what their customer contracts look like, but I guess this means that their customer contracts say you must have the next quarter's done at least before three months after, because we're going into August and we're seeing the Q2 updates.

So we had Omdia, who released its Q2 updates for the Middle East smartphone market. And I believe it was IDC who did it for India. So interesting sort of—let's assume that they're both looking at similar things. And what did we see?

I think the interesting part is we saw those markets going down, right? I mean, you see on the India side of things, India going down 11%, 11% in terms of smartphone shipments.

Tim Hatt (11:25)
Mm-hmm.

Peter Jarich (11:26)
Blamed on, of course, the memory chip crisis. But what I think is interesting there is you're seeing the impact mostly on the low end. And I think that was the thing that was called out, particularly in the Middle East.

So the Middle East numbers, where they were down—the market contracting 19% in Q2. But the fun—the fun part there, right, is that if I'm looking at this correctly, what they saw was that devices under $200 were hit hardest, dropping 42%.

Right. And markets that were really—markets that saw that low tier being a big part of their mix really, really got hit hard. I mean, Iraq was hit with a 36% year-on-year decline.

I mean, obviously people are out there, they're going, "Great. A 36% year-on-year decline in Iraq doesn't really mean much to me." But I mean, I think what it highlights is, you know, we can look at those top-level 11% numbers, 10%, 20% numbers even, and say, okay, the memory crisis is hurting things.

But when you dig into it, where it's really hurting is that low end. Because memory is—and I think we talked about this before—memory is just such a larger proportion of the bill of materials of those phones. And it just impacts the cost of those and functionality of those phones so much more.

Tim Hatt (12:49)
Absolutely. And it also, just before we move on, I mean, I think it also brings up an interesting point, which is, if you look at the smartphone industry right now, in moving towards AI and also battling this RAM shortage, what does that then mean for shipments into other developing markets?

How do you look at democratized access in Africa, India, parts of South America, where you do have lower-income populations who gravitate towards the mid and low end? If you get a squeezed margin down there, that is going to constrain shipments, and you're going to have to have action there.

And that's to say nothing of the rare earths and the hemispheric poles with that in China, the supply chain constraints. So there's a bunch of buffeting factors for the smartphone world right now, which actually sit below these numbers. And it's quite an interesting mix of—

Peter Jarich (13:46)
Yeah, what I think is interesting is, before we just move on to the other sort of things people may have missed or we want to talk about, what's interesting is the folks at Omdia on the Middle East side of things called out sort of there is a bit of an interesting angle here, which is you did see some of the OEMs putting new features in and maybe driving folks—heck, because if the price of those low-end ones are going up and there's not that much value-add there, then maybe they go up.

And they said shipments of devices priced above $300 grew 16%, right? Which is great if you can do it, right? I mean, this idea going that, "Hey, you know, okay, people are like, well, maybe I'll buy more of the—" folks who have money can buy the better ones and say, "Well, yeah, maybe I'll shift to a higher-end one."

But not everyone can, to your point, right? And I think that you see that in all those markets where people really have the most need for those devices. And maybe this drives more refurbishment, which would be great.

But I think we can only sort of put on a Pollyanna smile too much before we recognise it. It is going to start hurting folks in terms of access to those devices.

Let's move off from devices, though. What else is going on? I know we've each got one story we think is kind of interesting. Tim, what do you want to call out on your end?

Tim Hatt (15:11)
Well, I want to call out—go back to this story on sort of AI's security.

And just quickly, we, as a refresher, the Hugging Face incident two or three weeks ago, I think, set off tremors in the industry and beyond, because with OpenAI's model going rogue and conducting surreptitious security attacks without their knowledge, kind of opened a Pandora's box-style question: What do you do with AI if it gets out of your control?

We've since seen Meta admit some of its AI apps have engaged in similar activity. And the latest this week is OpenAI has had to pull the testing of their Astra model, not because it had infiltrated another property, but because they had discovered coding sequences that could allow it to do that.

And I think the point here is that we talked about this two or three weeks ago. It's not immediately clear how these companies get a grip on this, because when you have AI, the very nature of that is that it thinks for itself as a neural network.

And so, if you have set it up to do these things, you're going to have to bring in guardrails much earlier. And I think we're going to see a lot of activity on the security side, of multiple sectors working together to make this happen.

But I think this week is just another reminder of how big the challenge is.

Peter Jarich (16:35)
I mean, I think to your point, you also have to have explicit guardrails, right? Explicit guardrails of, "Please don't go and try and break into anything."

I am waiting for the use case or the example where AI goes and knows that the answer to the problem they're trying to solve lives with a certain person, and hacks into their email and blackmails them.

Until they get—I mean, then I know that AI will really have gotten smart. "This guy's got the answer. I've hacked his Gmail. I am holding him for ransom until he does it." I mean, work smarter, not harder?

I mean, why does the AI have to do all the work? We should be working for them on the AI front.

Tim Hatt (17:15)
Yeah.

Peter Jarich (17:16)
I mean, I think there is also a question of, it's 2026. Is it now uncool if you don't have an AI that's done something?

You know, when you saw that—I've no doubt that when Meta acknowledged its issues or concerns, that they were being honest, but there's also got to be a bit of a, "We better—I mean, we have to make sure we tell everyone how cool our rogue AI can be."

I mean, it's powerful too. It does feel like it's got to be a bit of a marketing angle to it on the AI front.

Much less exciting, though, I think what I saw this week was a bit of a discussion of how we're thinking about AI in networks.

Right. And so a discussion of how we're seeing AI in networks. But also, AI in networks—it's never going to be very exciting. And for folks who don't know, I think most of you who are out there listening to us probably understand sort of the AI for networks versus networks for AI, AI for telco versus telco for AI.

The idea that what we want to get to are telco networks, telcos themselves enabling the AI economy, enabling AI use cases. A lot of the early-on work has been much more about AI being used to improve telco operations, which is important, just not as sexy as them actually making money from it.

And we saw two interesting news pieces, one out of LG U+, working with Broadcom. They leveraged AI inside home routers to manage the home networking environment, the Wi-Fi environment.

And then separately, DOCOMO and Samsung, validating what they called AI RAN technology, which was looking to actually optimise the user experience.

And I think what's interesting for me is, one, on the AI RAN side of things, it's once again another reminder that the way we use this term is going to be very, very confused over the next years because everyone's throwing it out there. And what exactly it means? Is it in the radio? Is it optimisation? Is it—exactly how are we using it?

But two, I think both these things—and AI for networks is going to be generally not the most exciting thing anyways—both these things feel like features that, if I'd told a friend or I told my wife, or I said, "Hey, they're now using AI to optimise the Wi-Fi environment in my house," would be like, "They weren't doing that already?"

You know, this idea that we're, you know, using AI again from the DOCOMO and Samsung side of things to roll out predictive architectures that can understand, "Hey, here's what Tim is doing. We'd better make sure that we're optimising the experience for Tim," or, "Here's what users are doing. We'd better make sure we're optimising the network for those experiences."

Cool. Most people, I think, would assume that that was being done on some level. And I'm not saying that this is boring, as much as it is maybe just a reminder that AI is getting us to a place that we thought we were—or we should have been—which is still good in the end.

It just makes it a little bit harder to make it seem Nobel Prize-worthy or exciting in some major way.

Tim Hatt (20:33)
Yeah, it becomes table stakes, and it has become that. And I think you're seeing it across the network value chain. You're going to see it across the whole economy. And this is just something that, as you say, in that paradigm, we're going to see more of.

One quick data point, I think, on this is just from our Q2 Telco AI Tracker. When we look at the AI for networks versus the networks for AI, the networks for AI is really about kind of a revenue motivation. And that is actually where we're seeing the direction of travel.

You look at the telco launches of AI deployments, actually the last six months is up 10 percentage points. And so, particularly in the US, parts of the Gulf, and China. And so we know operators are moving towards there, and that's going to be an interesting thing to follow.

Peter Jarich (21:17)
Yeah, it's just—it becomes a bit of a marketing challenge because some of these things aren't the most exciting.

I mean, the idea of embedding AI in the router so that the AI capabilities are there on-prem, right? This is edge AI. This is one more version of edge AI that we've been talking about.

It's just, it's not super sexy, which is fine. It just becomes a bit of a marketing challenge. And maybe in the end it's fine because it just needs to be transparent.

As long as my ISP, as long as my telco is delivering a good experience, everyone in the house is getting the best gaming experience, the best video experience, whatever, fine, right?

I think that's maybe—we've got to be comfortable with that.

On that note, being comfortable with that, we said we're going to make it a short one this week to test it out. So, folks, let us know how this worked for you.

Tim, thank you very much.

Tim Hatt (22:13)
Pleasure, Peter. Good to be here with you.

Peter Jarich (22:14)
Everyone, just a reminder to please seek out the intelligence, but don't forget to embrace the noise. And, of course, don't forget to tune in next week.

Thank you all very much.

Tim Hatt (22:27)
Cheers, guys.