Episode 11: Telecom’s next frontier: SpaceX, AI data centres & 6G

Peter Jarich (00:00)
Hi everyone, and welcome to Intelligence & Noise, the podcast from GSMA Intelligence, where we take a look at the week's news. We try and search for some intelligence by filtering through the noise.

I’m Peter Jarich. I’m the Head of the GSMA Intelligence team. I’ve got with me Tim. Hey Tim, how are you doing?

Tim Hatt (00:20)
Good, Peter. How are you doing? Happy August.

Peter Jarich (00:22)
Happy August. I’ve got my tennis, I’ve got my Lacoste on because we’re in the middle of your home country. We’ve got the National Bank Open.

Tim Hatt (00:34)
We do.

Peter Jarich (00:34)
That’s on ESPN, the Ocho, or whatever minor ESPN channel. It’s the National Bank Open, sponsored by Rogers, which I think is probably the best way I’ve seen—

Tim Hatt (00:45)
That’s right. They’re bad fellas.

Peter Jarich (00:49)
Yeah. How do you get as many sponsors’ names in the title as possible?

Tim Hatt (00:52)
Yeah.

Peter Jarich (00:52)
But it’s through the middle of August, obviously setting the stage for the US Open. So I just felt a bit in the mood. My countryman Djokovic is set to do doubles with Sabalenka, which, I mean, stack in the deck, right? If I’m going to do doubles, I might as well get together with the number one. Gotta give him credit for that.

As you all know, we generally form the context of this podcast by talking through the CEO newsletter that a lot of you are hopefully seeing because you’re all super important people.

But it’s on break. We’re in the middle of summer, or the end of summer, or the part of summer where no one’s around, and so it’s on break for a bit. Which means Tim and I are unleashed. We’ve gotten rid of the shackles of whatever is in the CEO newsletter to talk about whatever.

And I think the expectation is, in the middle of summer, there’s not much to talk about. But I think we’re wrong, right, Tim? I think we’ve got some good stuff to talk about this week. So I’m excited for it.

You excited to talk some non-tennis things?

Tim Hatt (02:07)
You can put some tennis in there as well, but there is a lot going on. August sometimes people think—I think I’ve heard in political circles they call it a month for warm champagne. Lots going on in the tech and telco sector, so let’s get it going.

Peter Jarich (02:21)
Yeah. So as usual, I like to kick off with a little bit of a context story.

It was interesting because I was going through, as is often the case when you get to the sort of quiet part of summer, and I’ve got some holiday coming up, so I was going through my laptop, sort of cleaning it up, just trying to make sure that I get things straight. And I ran across some old documents.

It reminded me of the early days of my career. And for those of you who listened last week, you heard a bit about Tim’s early career when he was a would-be brewer, wanted to run a brewery.

My first job out of college, out of grad school, was actually working for a company that was developing stratospheric platforms to deliver 3G services because, heck, it was expensive to build a bunch of towers and deliver 3G.

And this was going to be a much better way to do it. Get up into the stratosphere. Maybe start with aerostats. Hopefully, you don’t clip too many wings off airplanes that are flying nearby. Go up to the stratosphere. Giant, giant tower.

It made so much sense.

And in many ways, I think the company was a bit like current ones that had a bunch of things on paper, but God knows it would actually work out. And that company didn’t necessarily work out.

But it was interesting because after that, becoming an analyst, I remember telling the story of saying, “Hey, early days of my career, big tower, covering a lot of folks.”

But the downside is you’re spreading all that capacity, all that spectrum, unless you do super-advanced beamforming. And let me tell you, back 20 or 25 years ago, super-advanced beamforming was not super advanced.

You lose a lot of density, you lose a lot of capacity because you can’t get that reuse. And what we really want are small cells.

And I remember doing a lot of work with the Small Cell Forum, attending all their events. It was interesting to see that.

And then that buzz went away a little bit. Until last week, when the small-cell buzz came back.

And we ended up with a lot more small-cell buzz than I would have expected.

So Tim, what’s brought small cells back into everyone’s consciousness?

Tim Hatt (04:53)
It is the SpaceX results that brought small cells and femtocells back into play.

So we saw the SpaceX Q2 results, their first quarterly filings as a public company. Some notable numbers we can pull out and get into, but net-net, there is very high growth, as you would expect from a company in growth markets.

We’ll get into the mobile side and their direct-to-cell ambitions as a potential mobile provider, but that’s where the femtocells come into play.

But yeah, over to you, Peter, for your take.

Peter Jarich (05:35)
Yeah, I think that’s the part that we’re going to dig into most, right?

The big numbers: revenues were up, as you said, huge growth, like twofold, almost twofold growth. Was it $7.8 billion? Spending, sixfold growth?

So we’re going through BP27 planning right now, and my finance guy wouldn’t let me do that. But it’s a little bit more ambitious, to be clear, than what SpaceX wants to do.

That’s why you saw the stock take a bit of a dip, was it 9%?

The AI of it all is obviously driving a lot of the revenue, driving a lot of the growth, right? But what’s interesting is a couple of quotes here.

You had this quote out there that it’s not out of the question that at some point Starlink will operate most of the world’s internet.

So I think with that kind of comment, those are the kind of comments that help people go, “Yeah, okay, revenue is up twofold, but spending’s up sixfold. That’s fine if that’s exactly what you’re doing.”

And if you look at it, the AI side of things is still generating around $2.5 billion, as opposed to $4.3 billion for the Starlink side of things.

So it is the business.

But I think, when the IPO was filed, it was pretty clear—and we were saying this—look, they say they’re going to take on the telcos.

You read between the lines, you don’t even have to read between the lines that closely. They’re clearly going to go directly after all of the telcos, after sort of saying, “Hey, we’re partners, we all work together.”

It’s pretty clear what they want.

Give a bit of the story of how they said they want to do that, because I think a lot of us were focused on this idea of direct-to-device. They’re going to do direct-to-device. That’s how they’re going to target all of these users. That’s where the competition lies.

And there was a bit of a different angle here, at least per the Q2 earnings, explaining how they might try to take on all the carriers.

Tim Hatt (07:40)
Yeah, that’s right. And I listened to the earnings call. I was there. I know a number of the financial analysts who were asking questions on there.

It was a momentous event, it being the first quarterly filing. But the question came up: “How do you guys plan to get into mobile? Can you give us some more detail on that?”

And while management can’t give, or was not going to give, huge amounts of detail, they made it pretty clear that they are going to try a CapEx-light deployment model that is going to pair femtocells—which are a type of small cell—and deploy them within or around, as in close proximity to, Starlink broadband dishes.

The types of dishes that, just so everybody knows, you need to have on your premises as a Starlink broadband customer.

The other part to the story, so that’s the infrastructure part, is then to use the terrestrial spectrum, right? The EchoStar spectrum that they bought, in order to roll this out and have the suitable capacity so that Starlink Mobile is essentially pairing their satellite constellation with a terrestrial femtocell network.

Peter Jarich (09:00)
And it’s okay—and just, I mean, the big number-one question is, I think we all sort of thought they’d acquired spectrum for satellite use. It’s totally okay for terrestrial, right?

Tim Hatt (09:11)
That’s how they’re positioning it. That’s how they’re positioning it.

And I think it raises interesting questions.

So, number one, if you step back, you talked about the numbers. Starlink right now is growing very strongly, about 65% year-on-year. The broadband subscriber count doubled to 12 million.

And if you look at it as a contributor of growth, it accounted for almost half of the SpaceX revenue growth in the second quarter. The other 40% to 45% was from AI. Space is actually relatively flat, their rocket launch, or less so.

But when you look at how that plays in terms of capacity, I think there are really interesting questions around: can a femtocell provide the same type of service that we would see from terrestrial base stations?

They seem to think so.

I think there are a number of questions there. How do you think about that, Peter?

Peter Jarich (10:09)
Yeah, I think it’s interesting.

I think, as always—and I’ll come back to the bigger question I have about all of this—but actually, before I come back, while I’m at it, if you look at the earnings call, some of the comments, and it’s always hard to know how to read tone, right?

But if we look at the comments where this first came out, “Let me give you a little hint. You can put a cellular base station on the gear that holds a Starlink broadband dish. You can have little femtocells around the country and deploy it as you need.”

This doesn’t sound like a fully fleshed-out plan. And I know it’s a little bit of a hint—we’re not trying to give too many details—but it reminds me of the other one: “I anticipate that we’ll be able to acquire quite a few of their customers because I think our service will be better.”

This reminds me so much of when Musk was talking about Tesla and said, “Hey, we’ve got all this AI compute in the Teslas, and we could take it all and basically develop an AI grid and provide, compete with AWS, given all the AI capacity we have.”

Which is a really cool idea and concept, but how that gets implemented, or if that gets implemented, is another question.

So the tone of it, and sort of coming out of nowhere, makes me go: is this real, or is this really a plan?

But I think, assuming that they’re all okay with the spectrum—and we’ll assume that they’ve done, they’re smart folks, they’ve got smart lawyers, I’m sure they’ve done the right work to at least make that seem plausible—I guess it is the question of what does that model look like?

Because the original consumer femtocell model really was about: how do I deploy this inside a house so I now have better service in my house?

If we’re talking about deploying this at the dish, either you can connect the dish to something inside the house, which would be great, right? But you’re probably not going to provide coverage or build out a nationwide network.

And that’s the implication, right? We’re going to build out a nationwide network by throwing these things on top of roofs.

Or if you put it on the roof, you may not have as great in-building coverage, and you have the power, right? Because how much power can you actually get?

And then, of course, you’ve got the sticky question here of where are people using this?

They’re not using it in the cities where you need a lot of capacity. Your average Starlink user isn’t putting a dish on their flat in downtown Manhattan. They’re out places where the solution for bad coverage was supposed to be direct-to-device satellite.

So I think it is a wonderful classic Elon Musk thing when you’re the chief engineer of the company. You go, “This is a really cool concept. We could do this, and wouldn’t this be great?”

And the tech nerd in you goes, “Yeah, wow, that’s great.”

But how it all comes together, I’ve got some other questions. But that’s my take. Those are the parts where I’m still scratching my head.

Tim Hatt (13:27)
Yeah, and you’re right. I think there is—you’ve got to split out the logic from the aura of Musk.

And you see it even on the call. A number of times, he sort of went into, very much unlike traditional CEOs or top management, explanations at a physical level of how things will work and why their innovation is going to translate into never-before-seen types of propositions or success.

On this one, he also chimed in to say that it was a really important move, that it was going to change the game, just like when Shotwell was talking about it.

But just to build on what you said, let’s just go back to the cold, hard realities of it.

Coverage, point number one. A femtocell has a link of perhaps 10 to 15 metres coverage radius. A typical macro cell is up to two kilometres. So potentially a 200x kind of difference there.

Now, when you scale that up to needing to meet capacity requirements of a national carrier, that’s a big difference.

Two: what happens if customers don’t have a Starlink broadband dish? How are you going to provide service in areas if you’re looking to co-locate the femtocell where there isn’t a Starlink broadband dish?

There are only 12 million of them, and you’re looking at a mobile market of 300-plus million.

So—and then you have, that’s to say nothing of the spectral capacity limits.

I think what they’re playing on, Peter, is that they talk up the V3 satellites having a 10x capacity gain from the spectral efficiencies and the innovation, and that that will solve everything.

What the missing link is, is how that translates into terrestrial capacity, which is a completely different ballgame. And I’m not sure people get that.

Peter Jarich (15:14)
But I’m surprised that they didn’t call this out.

And I think a lot of listeners will probably know that we’ve done some work around capacity calculations—what Starlink, or really NTN capacity, looks like given a certain amount of spectrum.

The one thing that I was surprised wasn’t called out is that this kind of changes some of those calculations.

If we’re talking direct-to-device, and direct-to-device is now being delivered by a femtocell connected to a Starlink dish, it changes some of those calculations because your link budget on a giant Starlink dish is much better, right? So you get much better efficiency.

Tim Hatt (15:51)
Hundred percent.

Peter Jarich (15:53)
Connected.

And so if we’re talking about the efficiency connected to a dish, a Starlink dish, which is going to be better, then in theory, even those 10x numbers, you could maybe bump those up.

Because we’re not delivering direct-to-device from the satellite. We’re delivering direct-to-device from something a lot closer.

And the actual satellite link is much more efficient.

So I will give them, as an engineering exercise, it makes sense. But I think, to your point, in the end what you end up with is much better experience for a lot of backyard barbecues in rural America.

And two parts of that.

One, don’t @ me. I own a farm in rural America, so I’ve had those parties. I know those people. It is important, right? And I think, if I still lived on that farm, I would probably have a Starlink dish right now.

But that’s the other part. This is just one market.

When we talk about the global implications here, it’s a particularly important market, don’t get me wrong, but we also shouldn’t be getting too out of sorts here when we think about what this means.

The last thing I think was kind of interesting—we could talk about this. We’ve had some back and forth with one of our colleagues who does a lot of our NTN satellite work, and he was saying, he wonders: is this just a bargaining chip, a negotiating chip, to put the US carriers into a better mood to allow for an MVNO?

Is this just a ploy to say, “I mean, you may not want to give us an MVNO, but you know, we have this backup option, which is going to come right at you.”

And to your point, the coverage range of those residential femtocells, which don’t really exist much anymore—you don’t see many of those. You see more enterprise-side of things—is pretty small.

But if you’re Musk and you’re talking about building this giant manufacturing capability, “I’m going to have silicon, I’m going to have manufacturing, I’m going to do all sorts of cool stuff,” you can say—and no one’s—because it doesn’t exist—you can say, “Hey, look, I’m going to build those.”

And the old ones that you’re used to, yeah, those were 3G, 4G, and they only had a 12- to 15-metre range.

But I’m going to build some that are going to be super powerful and cover you and your neighbourhood, if the regulations allow.

So I think there’s a very classic sort of Musk story here. The engineering side is there. There are so many questions that you sort of go, “Is it real? Is it random?”

I guess we’ll find out.

Tim Hatt (18:31)
We’re going to find out.

And I think a number of analysts have asked the US carrier CEOs if they would countenance an MVNO with Starlink, and I think we know where the jury resides on that.

But I think whether the point here is: is it bargaining leverage, whether in spectrum or MVNOs, or is this real?

And you just don’t quite know with Elon Musk.

We’re going to see, but it’s interesting to follow.

Peter Jarich (19:02)
Well, my favourite, before we leave this—my favourite sort of, I don’t know if this was prescient or if this was just bad luck—but I love it.

The end of that last quote I was talking about, about acquiring other customers, goes—I think it was—“It’ll be better during any sort of natural disaster because, surprisingly, even though space movies make space look quite dangerous, it’s a pretty quiescent environment.”

That was right before they crashed a rocket into the Moon.

It’s pretty—I mean, space is big, right?

Tim Hatt (19:32)
Yeah.

Peter Jarich (19:34)
You hear this a lot from SpaceX every time, and everyone, and they say, “Don’t worry about how many satellites we’ll bring up there. There’s a lot of space. Space is super huge.”

So it would seem like it should take a lot to crash something into the Moon.

Like, that feels like that shouldn’t be that easy.

But yeah, pretty quiescent.

So what is the—again, although it’s a bit quiet in the summer—second thing we want to talk about? Because I think it might be just a regulation in and of itself.

We’ll probably talk a bit about AI and what other folks are doing on AI.

And so I think the second piece—and this was from the end of July, so it’s not quite as fresh, but there are some interesting angles here that made me think about it—is the Korean AI Data Center Alliance.

So this was one of three national projects, and the whole plan is 18.4 gigawatts of large-scale AI data centre capacity, developing AI data centre-related upstream and downstream industries for import and export.

So it’s not just internal, but obviously external as well, and sort of establishing these data centre clusters and test beds.

I’ve got some reasons why this, including some left-field reasons, piqued my interest.

Important? Not important? The kind of thing that just takes place in the summer and we forget about it?

Tim Hatt (21:19)
No, I think it is important.

I think part of it comes down to scale. I think there’s a sovereignty point, and I think there’s a point around it taking a village, and sort of how this plays into the geopolitical dynamics around AI.

Just for our listeners, as Peter was saying, the announcement here comes out of Korea and it is a national mission to build a huge amount of AI capacity.

As you said, they’re targeting 18 gigawatts. That will be split, phased in over time, and they’re going to put an investment of close to $680 billion.

So a huge amount for context.

The US, probably the private announcement after President Trump came in for his second term, is about $500 billion. So very, very big numbers.

If you look—you step back right now, just to give you some context—Korea accounts for about 3% to 5% of global installed data centre capacity.

The US is number one, China is number two, by far and away the two biggest.

But the interesting thing is Korea punches above its weight. If you standardise that metric per million of population, Korea is actually number two. The US stays number one, China is number three.

And when you install this additional capacity, if it comes online as they planned, they’re going to go up 5x there.

So I think there’s a scale point.

The other thing, Peter, before I come back to you, is the point that it takes a village.

What do I mean by that?

They’ve been very clear in the announcement: this isn’t just about bringing on the hyperscalers, or just bringing on the telco operators, or just the chipmakers—SK hynix, Samsung and others.

It’s all of them, along with power and cooling companies.

And I think that really speaks to how this is a kind of national mission here, which is the way we’re starting to see AI investments across the rest of the world.

Peter Jarich (23:17)
Yeah, I think so.

Going back to your point, what’s interesting about those calculations, the capacity per pop, is it kind of goes back to a bit of—if you look at how they’ve set this thing up, right? They’ve got these three working groups: foundation, working with the local ecosystem, working on regulation, which we’ll come back to is important, but also supply and demand.

The supply side of things, I think, is the thing that we’re all understanding.

Building the IT side of things, building the power, cooling, all of that.

But it’s the demand side of things, which can’t just be domestic. It’s going to have to be external as well.

And that’s what those numbers point to, right?

Those numbers say, be off the charts in terms of how much capacity per person and how much does one country need.

Now, you’ll assume that those numbers will go up. If you grow the ecosystem, then you’re going to have more demand internally.

But it is also this sort of external focus, which I think is interesting, and how do you match that?

Then, going back to your point about taking the village, the power side of things is something that sometimes, if you’re looking at it, certain people are talking about it. I think more people than not aren’t talking about it and how we power these things.

My understanding is that it’s particularly sticky in Korea because the national power company has some debt burdens.

So I think trying to figure that out, and how that comes back to the regulatory side of things, where some of that money actually goes, might be the interesting side of things.

Not surprisingly, I think the reason there are a couple of reasons why this piqued my interest.

If you go and you look at SKT, who filed results recently, second quarter AI data centre revenues were up 92.5% year over year.

So, not surprisingly, right? One of the faster-growing—and that’s why they built SK Hyper, their data centre business.

So you definitely see this reflected in how these companies are—

Tim Hatt (25:25)
I’m not sure.

Peter Jarich (25:27)
It is just the question for that’s going to face everyone else, right?

You heard this with the investments in the US. You talked about the size of the US investment versus the public-private versus this.

Slightly larger than what’s going on in the US.

But the follow-on to that is: does this stuff actually take place? Is it commitments? Is it real?

What happens if a bubble bursts? What happens when someone realises they spent a whole lot of money on last year’s silicon that’s now less powerful than this year’s silicon?

I think it’s the same issues, right? It’s got to be the same issues in Korea.

And I guess the big touchstone will be how do we get to 2029, when they’re going to have that extra five gig deployed before they get to the longer term?

Tim Hatt (26:12)
Yep.

Peter Jarich (26:14)
What was interesting, though, for me—and I said there were some slightly random reasons why this caught my attention—is that in tandem we saw last week, and this was actually from August 4th, Ericsson put out a release about being named partner with SK Telecom for their Hyper AI Assort consortium.

And this is some work around AI RAN. It’s supporting a government-backed AI RAN initiative.

And there are a couple of things here that are interesting.

For me, the one part is, as always with the vendor considerations, it’s a question of what do we mean by AI RAN?

Is it AI silicon powering RAN operations?

Is it AI and radios?

I think in this case, if you look at some of what Ericsson is going to be doing, they’re going to be including their EIAP, so their automation platform, as well as extending demonstrations.

So, cool. Full raft of things.

But Ericsson made a really big deal—I say a big deal, it’s a big deal when it’s your press release with your title and your bullets underneath that summarise it.

One of those summary bullets is: “Ericsson named sole global vendor among main partners.”

They’re making a big deal that they’re the only non-Korean company, which is interesting because if we go back to earlier in the year, SK Telecom was talking up an AI RAN trial with Nokia.

And so, one, how we’re defining AI RAN and what AI RAN means as part of that work that they’re doing as part of the consortium.

It’s a question, two, what this means for different vendors.

I mean, if Ericsson says it’s the only global vendor, does that mean the work with Nokia isn’t taking place? Does that mean that they’re focused somewhere else?

Those are all the fun intrigues.

But I think for me, the biggest one is, let’s go back and look at what the numbers are.

You talked about the US public-private partnership around $500 billion. You’ve got $683 billion coming out of Korea.

This Hyper AI Network Infrastructure Initiative, $10.7 million.

I mean, that’s minuscule.

That’s like a drop in the bucket compared to the broader AI side of things, which should benefit it.

But as always, it feels like we’re making a lot out of something that is, in the grand scheme of things—

And maybe it just means we’re not investing in the right places. Maybe it means that we should be investing more in AI RAN, where there’s probably less of a bubble because we understand the value versus other things.

But yeah, it’s a fun one.

So what else?

What else? This would be the part where, when our content folks get the animated videos, the Willie Nelson—we call this the Willie Nelson section, right? The “On the Road” and “On Our Minds.”

And we’re not on the road because it’s—I mean, I’m on the road soon to go on holiday, but we’re not going anywhere that we want to talk about.

So what else is on our minds? What’s on your mind, Tim? What else other than this big news is capturing your attention?

Tim Hatt (29:53)
Well, I wanted to shift gears a little bit.

We talk about AI all the time. We talk about satellite a lot. Just a story on towers, actually, at the moment.

IHS Towers, big global tower company with an erstwhile footprint in Africa and South America, has sold—they’ve divested about 25% of their portfolio globally in South America to a bank, an Australian bank called Macquarie.

This may seem—and most of those were concentrated in Brazil, some in other parts of the continent.

But I wanted to bring it up because as much as we have seen the trend of towers being sold off and increasingly invested in as part of a digital infrastructure asset class by private equity groups and banks, as we see with Macquarie, who owns a lot of tower assets, I think the interesting thing here is that—

And the other side of the story is IHS is being sold to MTN at the moment, and they will concentrate the rest of their 28,000 towers in Africa under that ownership.

And I think it’s interesting, Peter, because we’ve seen so much in the way of sell-offs and tower leasebacks of telco operators the last 10-plus years as a way to deleverage and get economies of scale.

That kind of is a win-win for both tower companies and operators, or any other investors in the passive infrastructure.

But MTN’s kind of going the opposite way and realising—and their CEO, Ralph Mupita, is on record as saying, “Hey guys, we’ve got a big plan to support Africa’s digital ambitions. We think we need both the passive and the active layer of the value chain to do that.”

And it’ll be interesting to see whether we start to see other moves in this direction to take control of tower assets for things not just normal telco service, but things like edge inferencing at the towers, a lot more mobilisation of new types of AI-based applications.

If you control that full stack, you’re in a better position.

Peter Jarich (32:02)
Yeah, it’s the fun swinging pendulum of everyone rushing to get rid of the towers because you could unlock more value and get value out of those assets.

And I think we’ve all sat through the big management consulting presentations of “de-layer, de-layer, de-layer,” and here’s the value, as if that’s telling you something super, super genius, right?

Because you see some examples out there.

And now we’re seeing that, if we think about them not just as towers but as real estate—real estate you can do something with, right? Real estate that you can put edge network into, to your point.

Or maybe that’s building an AI grid, as long as you’ve got power there and you need to be able to do that because you’re in markets like MTN is.

And they need to be able to support that.

Then it’s not a one-size-fits-all story.

So it’s definitely good for me. Also kind of niche, I think, potentially.

Also, back in early August—it was the fourth—there was an announcement that Intel joined the Linux Foundation’s ACUDU Foundation.

It’s the Open Central Unit, Distributed Unit.

I think for the record, it is pronounced “ACUDU.” I hear sometimes people question that.

Intel joined as a premier member.

So for those folks—and I think we talked about ACUDU here in the past—this is the initiative kicked off with support by the US government to basically create open baseband solutions.

And you’ve got Intel joining.

It’s interesting because I’ve seen some folks—and at first, when I read this, I think this was the way our folks at Mobile World Live put it—“Intel’s entry brings technical weight to the initiative, particularly across cloud-native infrastructure.”

Like, technical weight.

I mean, it’s Intel. Yeah, they’re important.

But you’ve already got AMD, Ericsson, AT&T, Nokia, NVIDIA, Verizon, Cisco, Marvell, the Open RAN Development Company, Red Hat, T-Mobile.

There are plenty of folks who understand cloud-native infrastructure, silicon platforms.

And so at first I was kind of like, yeah, I mean, I guess.

But you really can’t—from my perspective, if we’re talking about open, and we’re talking about radio access networks—Intel was one of the biggest drivers of Open RAN, right?

One of the drivers of the Open RAN Alliance, even if they weren’t one of the founding members.

And it’s probably not too much to say that, given the US government’s stake in Intel, where the US government is interested in driving this ACUDU development, it makes sense for them to join.

Of course, my one bit of lamenting is the same thing about how much money’s being spent on AI RAN versus AI data centres.

I think it’s interesting because I went to see what folks were trying to say, or were saying, out there.

Looking for some snarky—I was hoping for at least those folks in the media that we know that always like to cast everything in the most snarky, cynical way possible.

There was something out there of someone saying, “Yeah, of course Intel’s there,” or whatever it was.

And LinkedIn’s always a great place to see those things.

But when you go to the LinkedIn post about this and there are five comments, and some of the comments are like, “Whoa,” or “This is great,” it sort of highlights that this is kind of the thing that’s either flying below the radar or not that exciting to folks, and maybe not as exciting as it should be.

So yeah, I think maybe that’s the way they want it. Maybe they want to fly under the radar while they do big things.

Anything else?

Keeping anything else on your radar, Tim?

Tim Hatt (36:04)
I think we’ve covered most of it. I know we’re going to be back later in the week before we sign off, but lots happening.

Peter Jarich (36:12)
Yeah. I would—the only other thing I would throw, and I’m just going to throw this out there really quickly before we go, because it’s near and dear to our hearts here at GSMA Intelligence and the GSMA.

Also from early August: Optus advances Australia’s 6G future with landmark research trial.

And this is interesting because this is really about upper 6 GHz use.

And a couple of angles here, right? They were testing what could be done.

And it begs the whole question of what do we mean by 6G?

So is 6G just new spectrum? Is 6G new waveforms?

It seems like it was less about that and more about showing we could match the 5G grid, match 5G coverage, as long as we’re talking about more advanced antennas.

And why it’s interesting, though, is—I mean, shout out to Nokia for making that happen. Shout out to MediaTek for powering the test device.

But it becomes one of those things where you can sort of ask, “Hey, why do we need yet another proof point?”

There are other proof points out there that you can match the grids, and that’s what we do whenever we go to new spectrum.

It goes to one of those things where, if you’re talking to national regulators, they want to see: does this work here?

If I’m in Australia and I’m trying to make a decision about how this spectrum gets used, I probably want some demonstration, some commitment to show that it’s taking place here.

So I think that was—and beyond that, I think the thing that was the really fun part—and yeah, for the Elon Musks out there and all the engineers, the fact that we were talking about an antenna with 768 antenna elements is pretty impressive.

But in case you can’t forget the physical AI of it all, there was a robot dog.

It was a robot dog that was being used to actually map the radio environment indoors.

Which is novel, interesting.

I think it’s an interesting idea of how, going forward, we’ll probably see new ways to actually measure network performance and signal strength.

So that was pretty cool.

It doesn’t look like a dog, if you actually see pictures of it.

Tim Hatt (38:35)
Is it Bart?

Peter Jarich (38:36)
Does it bark?

Whenever it gets a good signal—okay, that’s the API.

Whenever it runs out of signal, it sits there in the corner and barks.

I mean, I hope it’s better than my robot vacuum cleaner, which falls over.

Tim Hatt (38:50)
Ha ha.

Peter Jarich (38:50)
We are now at the end of time.

Tim, thank you very much. We’ll be talking to you again next week.

Everyone out there, thank you as always for joining us.

Tell your friends, tell your neighbours. Tell any CEOs that are on the CEO newsletter. Give us a review.

And as always, I would like to remind you, if only because it is this podcast, to embrace, seek out intelligence, but embrace the noise.

And with that, we’ll see you all next week.

Tim Hatt (39:26)
Cheers, everybody.