Episode 14: The next AI wave:Social media regulation, AI data centres & the rise of humanoid robots

Tim Hatt (00:00)
Hi everybody, and welcome to this week's edition of Intelligence and Noise, the GSMA Intelligence podcast covering all things great and good in telco and tech, where we look to sift through the huge amounts of noise happening in the industry to pull out the intelligence that matters.

Great to see you all again. We are coming back from the summer holiday, and so you can probably notice first off that I am not Peter. I may sound a bit like him because we're from the same continent, but he is still enjoying his last week of break. So he's got the week off this week.

I'm here in his place and very delighted to be joined by one of our great analysts and longtime GSMA Intelligence members, Radhika Gupta, who is going to be our guest on the podcast this week.

So, Radhika, how's it going?

Radhika Gupta (01:02)
It's going well, Tim, and thank you for having me. I am delighted to be on this one.

Tim Hatt (01:08)
Excellent to have you here. Radhika is based in India, but she has fantastic insights to share across the whole sector and the world.

So, I think when we left off in the last show a couple of weeks ago, we were talking about a number of the key stories related to AI, satellite and others. You know, the world has moved on, but it kind of hasn't, in the sense that those topics still remain very key, very prescient and very top of mind.

We're going to be talking about a few stories today because the CEO newsletter, the newsletter that our Director General sends out to CEOs across the industry, is still on holiday for one more week. So we get to use our own discretion for what is really mattering amongst the huge wave of news.

We're going to be talking about the social media settlement that Meta made last week. We'll then switch gears to talk about a new data centre investment and vehicle out of Korea, which SK Telecom is using. And then we'll talk about a few other stories that caught our attention, and sort of wrap it into what it all means going forward as we hit the ground running in September.

You probably remember that Peter always starts each episode off with an anecdote to preface our conversation. So I'm going to do the same this week.

Actually, I'm going to go back to Canada, where I was on holiday for the last two weeks. And it's an interesting story.

Canada is a very large country, as you would imagine. It's got great cities, but really it's known for its huge forests and rural areas. For part of the time, we were at a cottage in Ontario, where I'm from, and we were doing a boat tour with some friends of ours who had come with us from the UK.

The guide was a local gentleman who had lived in the area for years. It was on Georgian Bay, which is one of the huge bays in Lake Huron.

While we were on the boat trip, he talked about some of his hobbies and the natural world there. At one point, he ended the conversation by talking about what he does on his fishing trips and how, at the end of that, because the Great Lakes are full of clean water that is, in most instances, safe to drink, he finishes his fishing trips with a glass of "Georgian Bay champagne" — i.e. water from Georgian Bay.

Now, I am not necessarily advocating that you all go out and have a glass of water from a Great Lake, even though I think that would probably be safe.

But it was an interesting point because what he was really getting at was how a lot of the natural world has used very advanced processes for hundreds, if not thousands, of years. Think of migratory birds coming down from Ontario to the Caribbean or Florida in the winter.

And it got me thinking that there is kind of a connection to what we talk about in AI and other technologies here. As advanced as those technologies are — and they most definitely are — at the end of the day, you still can't replace the need for fundamental questions and planning as to how you use the technologies.

Those kinds of simple questions remain just as prescient now with AI as they did 20 years ago, 40 years ago or 100 years ago.

So, when we talk about how you look at an AI transformation, the technology can only get you part of the way. You've really got to understand the fundamentals of what you're trying to do, who you're trying to do it with and why that makes sense at the given time.

So it was kind of an interesting prompt for thought: stepping out of technology and into nature every couple of weeks and thinking about the important questions that I think we should be asking ourselves.

Anyway, as I said before, we're going to get into the stories this week.

Coming out of summer, it's always an interesting time because people have been off, they've been taking time off, but equally, we are straight back into things in a very busy and eventful autumn period.

And so I think the first story that we're going to start off with is the Meta settlement that was made last week.

I will give the context here and then, Radhika, I'm going to come to you.

So, as a reminder, Meta reached a very large settlement in the US with a consortium of states who had pursued litigation against the company for the alleged effects of its platform on young people.

Details of the settlement come to around $18 billion in total. However, that is split between two different tranches. There is an initial approximately $12 billion payment, and the rest — $6 billion — is contingent on other competing social media platforms abiding by the same rules.

It has a number of rules attached to it, like daily time limits for young people, whether you can use algorithmic feeds versus more basic chronological feeds, the use of Like buttons and how those appear on the feeds of young people.

And then I think it has interesting questions in terms of age verification and how we go into this, which we'll get into.

But, Radhika, maybe your perspective on this at a high level, because I know we've talked about social media many times in the past, not just in the US but in light of social media bans in place in other countries.

We now have this US approach. What's your reaction to this?

Radhika Gupta (07:37)
This one is particularly very interesting, Tim, because, one, it's not the first time or this is not the first case against Meta. This is not the first time Meta might be settling.

But this case is interesting because there was a consortium of states coming together. There was a consortium of attorneys general coming together and suing Meta. And there were allegations against Meta.

Again, the $18 billion figure is making a lot of noise, but as you already highlighted, it's split between two tranches. Another interesting fact is it has to be paid over the course of 10 years.

So while $18 billion sounds like, wow, this is probably something to learn a lesson from, it has to be paid over the course of 10 years and it's probably close to just 1% of Meta's revenues.

So yes, it's a good amount to be paid, but is it really? I think there was a $200 billion consideration as a penalty that Meta would pay if they lost this case, and now they have settled for $18 billion.

So I think those numbers are really to be kept in mind when trying to evaluate whether Meta has really won or lost the case.

Now, coming to the actual question: social media bans. I think this topic is gaining more and more momentum and traction every single day, every single month.

And we have seen a lot of activity in this space, whether it's with the EU Digital Services Act or the UK Online Safety Act, or the recent ban from Australia on under-18s.

So this is a good step in the sense that this is not just a regulatory move. This is a move in the form of a penalty, as well as Meta having to comply with some obligations in the form of product design itself that it has to implement to ensure those child safety controls are in place.

So this is interesting, but there's a lot more that needs to be done.

And I would really like to give it back to you to talk about the second tranche of $6 billion, which is dependent on the other players in this space making similar changes to their products.

Tim Hatt (09:51)
Yeah, 100%. And you brought up a number of really salient points there, because I think as much as the $18 billion is what appears in headlines, in reality there's so much more beneath this that is arguably more interesting than the financial amount.

Because as much as that is a huge number, as you point out, it's like 1% of their turnover. And so even if that gets booked in one or two years, that ultimately is not a showstopper for a company of that size.

And so the question then becomes: what is the actual implication?

I think, just going into a little bit more detail so our listeners understand the rules that have been negotiated as part of this settlement, there is a two-hour daily limit for teenage users of Facebook platforms, including Facebook and Instagram, excluding Messenger.

There will be a number of modifications made to the Like system so that people can't see automatic Likes for posts. This is often seen to drive excessive use for people coveting Likes.

There will also be, I think importantly, a move to allow feeds that are not driven by algorithms. In other words, not what Facebook wants you to see, but rather just what comes up in the timeline of your friends.

And then various time prompts. In other words, if you've been using it for 15 minutes, 90 minutes, you will automatically get served a prompt.

The interesting thing here, for me, is that all of those, I think, make sense. And I think you've seen the reaction from a number of the attorneys general in the various states that were part of this investigation and lawsuit.

But there are key questions around how that gets enforced, particularly if the competitors don't play ball, and what happens to that additional tranche of $6 billion in that case.

And secondly, how you ensure that you know who is actually on your platform.

You know, we talked about this before, Radhika, Peter and I did, around the use of age verification and potentially the tie-in with mobile network APIs for that.

We know there are age-verification APIs within the Open Gateway suite, for example, that telecom operators are actively using with providers across a number of industries, including social media.

But how that gets enforced at a national scale in the US in this instance is still to be determined.

So I think there's a number of interesting questions.

One other thing I wanted to flag: you talked about the different types of action governments are taking on social media. Some are going for a ban, some are doing nothing, and then some are kind of in the middle here, like the US.

Where do you see that going? Do you think we're going to see more countries landing on one of those models more than others? Or do you think we're going to see a wide spectrum of interventions?

Radhika Gupta (13:08)
To be honest, I think first I would like to add one more restriction, which is my personal favourite: cosmetic filters will not be allowed for teenagers, which is very, very interesting because that somehow, I believe, boosts false ego for a lot of teenagers and kids.

So that is quite interesting for me.

Coming back to your question, this topic has certainly been gaining global momentum. So I do believe that we will see a lot more action from additional countries in this space, whether that would be similar to how the US has done it or similar to how some other countries have been doing it in terms of enforcing bans on usage.

We will see that with time.

But one thing that remains particularly important, no matter what approach you adopt, is the age-verification and age-assurance mechanism.

And I think that is the underlying fundamental, and that relates to your story, that anecdote you were talking about earlier: the fundamentals have to be in place.

So no matter which approach you take, no matter which strategy you take, unless and until your age-verification and age-assurance mechanisms and capabilities are rightly in place, then whatever approach you take is not going to work.

And Meta has been asked to develop those capabilities. As part of that, it has actually been allowed to use, to an extent, the data that it collects from children.

Otherwise, there is this law in the US that does not allow data for children under 13 to be used without their parents' consent.

But as part of this settlement, it has been allowed to use, to a certain extent, the data collected from those children to verify their age and provide age assurance.

So, one, I believe this is going to only get more momentum and traction globally.

Second, I think strategy is less important. What is more important here is that you have the right age-verification and age-assurance mechanisms in place, and they have to come from a combination of attributes.

And one of them, as you said, is the mobile network APIs that are giving these age-identification and KYC-verification solutions.

One of them is obviously available through network APIs, and similarly, digital identity wallets could be another.

So this is how I look at it.

Tim Hatt (15:51)
Yeah, I mean, and Radhika, I think we've talked about this before, we're talking about it now, and we're going to talk about it again.

Because as much as this is a US story this week, it will be another country's story next week, and the month after that, etc.

And so I think it's maybe the last point I'll just flag before we move on to the AI story is that part of this, there's an implicit point here that it's very difficult to regulate social media.

If it were possible, perhaps this lawsuit goes to completion without a settlement. But I think implicitly there's a point that says that was going to be difficult.

And to me, it reminds us that there are a lot of facets of the technology world for which the regulation is muffled or undefined or in grey areas.

I think we're going to see that a lot with respect to some AI use and questions we haven't even started to grapple with, let alone those we have, and we see it in social media.

So it's a really evolving situation.

Okay, I talked at the start that the other story we really wanted to give some credence to here is coming out of Korea, South Korea, where we've seen some interesting news in GSMA Intelligence on how telco operators finance and operate AI data centres.

So, give us the details here.

Radhika Gupta (17:20)
Sure, Tim. So, for our listeners, I think some context on this story.

In the month of July, we saw SK Telecom announce its new subsidiary, SK Hyper, for its data centre initiatives, where it would be responsible for developing its data centre operations. Its main role would be to acquire land, develop sites and so on.

Now, in another interesting development from SK Telecom, what it has done is it has announced that it will split its wholly owned subsidiary, SK Broadband, into two different entities.

The one will be called SK Broadband, which will continue to focus on its fixed-line, media and enterprise business.

The second one, which will be newly created, will be named SK Horizon. It will house its AI data centre and submarine cable business, and this is expected to be completed by Q1 2027.

Now, the interesting bit in this one is the newly formed SK Horizon, which will house its AI data centre and submarine cable business.

SK Telecom will retain only a 51% stake in this new entity, as against its 100% stake in SK Hyper that it created in July.

And the remaining 49% will come from two different entities: KKR and the IMM Investment and Stonebridge Capital consortium.

Now, this 49% investment from KKR and IMM is interesting because SK Telecom had highlighted earlier how data centre economics are enormous and how you really need money to fund those data centre capabilities and capacity.

So for me, this is a very interesting bit in this particular transaction, Tim. What do you think about it?

Tim Hatt (19:12)
Yeah, I do think it's a really interesting story. I think on a number of levels, you highlighted a couple of them up front.

In terms of there being third-party investment, I think as much as that's not a surprise because private equity and financial investors have long crowded into infrastructure as an asset class, it's still interesting here in terms of the ownership structure.

As you say, SK retains formal control with over 50%, so it doesn't have to worry about any operational confusion on that level, despite having the financial investment to support the build.

So I think the partnerships with the investors are an interesting angle.

I also think there's an interesting point in terms of where this whole deleveraging question that we've long been talking about comes in.

Operators have spun off towers for years. They've spun off submarine cable assets, in some cases fibre and dark fibre networks. We see Verizon, among other operators, working in that milieu.

So I think there's an interesting point here around just understanding the scale of investment that is required to operate AI compute and the data centres that power that, and that for a telco means you need to do that in a lean and efficient way.

And this is a good example of that.

And I do wonder, Radhika, whether we're going to see more attempts like this.

I think one of the reasons I think it's an interesting story is that we see — and you know this, you're very deep in AI data tracking, and maybe we can talk to some of that — we see in our telco AI research, we just published the Q2 tracker a few weeks ago, that as much as the customer-care channel and network automation and optimisation still take the lion's share of overall telco AI activity, increasing momentum is moving towards the data centre, towards the edge, to monetise assets using new AI business models.

But to do that, you need to have a very efficient investment model to be able to do it.

So, do you think we're going to see other examples of operators operating this type of model in different countries?

For example, India, where I know Jio and Airtel, amongst others, are very heavily moving into the AI infrastructure domain.

Radhika Gupta (21:56)
There are two aspects of looking at it. One is investing more into the data centre space, and second is delayering the data centre arm, analogous to how you explained the telcos' tower position.

They had de-layered for close to 10-plus years. Now they have de-layered their towers arm and then leased back those tower operations or their passive infrastructure.

So there are two ways of looking at whether or not others will follow.

Investment in data centre infrastructure, I think yes. As you mentioned about the Telco AI Tracker, the numbers are clearly evident of this. We have seen this from a lot of other telecom operators.

The interesting thing is there is a growing realisation among telecom operators that in this AI era, they want to capture a larger part of the value chain.

They don't just want to remain pipes, like in the data traffic era. They don't just want to remain pipes through which data traffic flows. So they'd want to capture a larger part of the value chain.

And in that particular attempt, we have seen distinct models emerging. Not only for this discussion, but tokens are another example we have seen. The distribution layer is another example.

And to that effect, in that attempt, we have seen a lot of telecom operators moving into the data centre opportunity because, one, it gives you monetisation opportunities and allows you to capture that part of the value chain.

Secondly, there's a lot of growing emphasis on the sovereignty angle also.

So when you talk about the sovereignty angle — sovereignty in AI — a large part of the AI value chain sits at the infrastructure layer.

Obviously, there's a data layer and there is a model layer, but a large part of the value chain sits at the infrastructure layer.

So if you can control and govern that layer, where the inferencing is happening and where the data processing is happening, you control a large part of the sovereignty layer as well.

So there is a sovereignty angle also at play here.

Certainly, a lot more telecom operators are going to foray into this.

But whether everyone is going to split their data centre arm, we will only figure it out with time because not everyone has the same amount of capital and the same amount of expertise that SK Telecom has.

But SK Telecom has a lot of in-house capabilities to do it, to de-layer its data centre arm and run it separately.

Tim Hatt (24:41)
Mm-hmm. Yeah. And you're right, there are going to be different approaches.

I think it's almost as interesting just to lay out what those possible approaches could be.

And we've seen in some recent research that we published — it's on our website — you can see on AI factories, we looked at different go-to-market options and what the potential revenue upside would be.

And we came out at an addressable opportunity of up to a 5% increase in cloud revenues through AI factories of the services we see now.

Which isn't just compute-as-a-service or GPU-as-a-service. It's not just inference. There are a number of vertical-specific solutions that telcos can offer within that portfolio, to your point, on the full stack.

Through to other plays like we see in Europe of getting government funding to support the builds as an anchor tenant to the data centres over time.

We're looking at Deutsche Telekom, we're looking at other operators pursuing that type of model. I think they're all in play.

The other thing I think just is worth mentioning is just reminding us that Korea is a big player in the global AI race.

You have the US and China, which are the two kind of undisputed behemoths, but Korea is really kind of betting the farm on AI here.

When you look at the size of the national investments they announced in the summer, it's in the many hundreds of billions of dollars, admittedly spread over time, but it co-opts multiple industries from their semiconductor champions, Samsung, SK Hynix, etc., through to the telco operators and others.

And so this is very much a national effort.

Radhika Gupta (26:41)
Yeah, 100%.

They have this national AI G3 strategy, as part of which they want to be among the top three countries — the AI powerhouses in the world — by 2030.

So obviously there's a national backing and a motivation for telecom operators to do it.

Also, it is worth mentioning that SK Telecom has kind of been the leading operator and pioneer in almost every technological innovation or advancement.

I'm just trying to recall my booth visit at MWC Barcelona every single year in the last two or three years.

You remember, post-COVID, there was this metaverse hype and I remember going to the SK Telecom booth and trying to get into those immersive worlds. They had those demonstrations.

If I remember correctly, at Barcelona 2024, they launched this Global Telco AI Alliance, where they launched an alliance to develop a telco-specific LLM.

And so this AI data centre initiative again was part of the same activity.

SK Telecom has kind of been a pioneer and a leading player in almost everything happening in the technology world.

So while they are setting a good leading example, whether every telecom operator or whether more telecom operators can replicate it will depend a lot more on whether they can attract investment, whether the national ambitions really back it up, and third, whether you have in-house capabilities or not.

As I mentioned earlier, SK Telecom houses 500-plus subsidiaries, and with the help of those 500-plus subsidiaries in-house, they have capabilities that include power generation, construction, semiconductors, computing and networking, just to name a few.

So when they have SK Hyper and SK Horizon to run the entire data centre operations, from site acquisition to power acquisition to construction, they can really do it end-to-end.

And for them, probably it makes sense to keep the management and the real estate part separate.

For others, whether or not it makes sense will depend on many more factors.

Tim Hatt (29:02)
100%.

And I think we'll come to this at the end.

But speaking of the topic of national AI agendas, we're going to be at IMC in your home country, in India, in early October.

And I think no doubt we're going to hear a lot about India's AI plans and approach there.

So we can come to that.

But let's transition, Radhika, into the other stories that caught our attention.

I know you had circled one that relates to India and Singapore. So tell us about that.

Radhika Gupta (29:39)
Yes, very interesting story and probably one of the weaker signals, as we call it in the GSMA Intelligence team, but a very interesting one.

So Singtel Singapore has recently made an application to the government here to offer satellite services directly to enterprise, public and private enterprise consumers.

They already operate in India through their arm that is known as Singapore Telecom Private India Limited, I believe.

And they have made this application because, I think in 2024, the Indian government did open up its space sector and did open up its FDI in the space sector, where they allowed up to 100% FDI in the space sector.

Up to 74% is allowed without any permission. If you go more than 74%, obviously you have to take government permission.

And because Singapore Telecom Private India Limited is wholly owned by Singtel, this will be 100% owned and they need to seek this permission.

Now this news is interesting because there's a lot of satellite news making the rounds in the Indian telecom market and in the global market, from Starlink to, just a few months ago when Jio was announcing its IPO, Jio announced its constellation plans.

So there's a lot of satellite news that has been making the rounds in the Indian market.

But this plan and this application filing from Singtel is not another constellation announcement, and that is the interesting bit.

It is not trying to compete with the likes of Starlink, Amazon, Jio's constellation plans or Eutelsat OneWeb.

Basically, what it is trying to do is just act as a satellite integrator, probably trying to use its own GEO capacity and aggregate capacity from other satellite players and offer it to enterprise customers.

So that is a very interesting and new model emerging when you talk about opening your satellite sector, when you talk about opening your space sector to other players in the market.

Until now, the conversation was mainly around constellation, D2D connectivity, and this is now shifting gears to offering direct connectivity to enterprise customers without even running your own constellation or without launching your own constellation.

And because you do so much in this topic, Tim, it would be really interesting to hear your take and your perspective on this.

Tim Hatt (32:12)
You've set it up beautifully.

I think you hit on two important points. One, it's not the traditional approach, and two, it gets to how you sell into enterprises, particularly for IoT connectivity, which often gets lost in the mix when we talk so much about Starlink and AST and various other direct-to-device go-to-market possibilities.

And I think when you talk about India in this instance, the interesting thing is when you look at the capacity aggregation, India not only has a burgeoning business market to sell satellite into, just as they do consumers.

They are now obviously in the top economies of the world, the largest population, a number of high-growth sectors — manufacturing, agriculture, even others such as oil and gas — for which we know there is satellite demand.

About 20% to 25% of buyers from our enterprise survey indicate that they would be interested in using satellite as part of their connectivity portfolios.

But India is also a very highly regulated market.

It's not clear how some of the big global titans will be able to enter it. There are a number of, I think, regulatory discussions ongoing there.

And so I think this approach, to your point, is potentially a kind of deft way of navigating that landscape to be able to offer capacity without launching a whole fleet of new birds and potentially facing different regulatory questions.

So, yeah, a very interesting story, and I suspect we're going to see more of that as India talks about satellite for its broader connectivity ecosystem.

I wanted to flag one here as well, and it's kind of half humorous, half meaningful.

And it's the Robot Olympics, the humanoid Olympics, which we saw in August over the last couple of weeks in Beijing — not surprisingly, given that it has hosted multiple Olympics before and is the robot capital of the world in China.

And I think what's interesting, I mean, when I look, Radhika, at the stories, I was on holiday and I got a news flash, as probably many people did, saying, "Robot beats Usain Bolt's 100-metre record."

And my first reaction was like, who cares? Why? Obviously a robot is going to be able to run faster. Tell me something I don't know.

And then you see all the bloopers of robots crashing into crash mats and kind of the hilarity of it.

But it's also an interesting point where I think China accepts that it's willing to take on board the jokes and the humour of these early trials to push humanoids further into how we look at everyday life.

And so I think we see examples of this at conferences, we see it elsewhere, but we're probably going to see more demonstrations like this, not just in sports, but in other contexts, as humanoids grow in their sophistication.

Radhika Gupta (35:29)
We do. And when you were mentioning the robot beating Usain Bolt, I was thinking, okay, fine, your heart rate doesn't go up, so you can do that.

I'm really bad at cardio, so the first thing that comes to my mind is how fast your heart rate goes up.

But coming back to the actual discussion, it indeed is interesting.

And I don't know how much fake news my LinkedIn feed shows me, Tim, but of late my LinkedIn feed is showing me so many humanoids and robots that are very human-like.

So, you know, they are not pieced together as a static robot or moving like this or that. They have a silicon face, they have silicon hands, and if you're only looking at the face and not the remaining body part, they pretty much look like humans, and they can make facial expressions also.

So while, one, this is an interesting development, second, we have seen so many, as analysts, presentations from vendors.

And one thing that constantly pops out is physical AI, which is robots and humanoids and all these things, and agentic AI is going to contribute to the next exponential rise of data traffic.

So obviously, this is one part of the story, and another part of the story is that it is going to have wider implications on how we are going to see the rising demand for data traffic and consequential implications for network demands or network requirements from telecom operators.

And obviously, accordingly, demand for AI infrastructure as well.

Tim Hatt (37:22)
Yeah. And all of those questions are relevant.

I think as you talk about physical AI starting to move into a wider range of device categories and form factors, we're going to see more on that.

And so I think it's almost as much as the story is about the sheer effect of it and the performative value of that.

As we start to get deeper into humanoids, we're going to need to go through these kinds of demonstrations.

I think the question for me is, as much as how the humanoids work and the productivity benefits they can have, where you draw the line between when you have a humanoid and when you don't.

And there are these questions around, I think a lot of big tech figures from government to industry are asking searching questions around what needs to be sort of off-limits for humanoids, or preserved, as some people have called it, versus what isn't.

And I think that's a live question that we're going to be grappling with for some time.

But there'll be many other podcasts to discuss that.

So, we will move on.

That, I think, brings us to the near conclusion of this week's podcast.

It being the end of summer and into September, we will be on the road again.

That great Willie Nelson song — we still don't yet have a jingle for on this podcast, but I am going to keep pushing for one.

As much as we had a relative break in July and August, the GSMA Intelligence team will be out and about in force through September, every week, in all parts of the world.

We'll be in Malaysia next week for a 360 event that the GSMA runs with industries there.

We'll be in other parts of the world through late September.

And then, of course, Radhika, we will be joining you in Delhi for IMC in early October.

So we've got a lot to look forward to.

As always, we are very interested in your feedback. Please post comments, please let us know what you want to hear more of, less of, what is getting your attention.

We are always very interested in that.

Feel free to give us a rating. We will definitely look at that, and we want to ultimately make sure that we're putting out content that is most valuable to our listeners.

But next week Peter will be back, so you'll see him again.

But for now, just to remind you to seek out the intelligence but embrace that noise across the industry, because there is a lot to be made from it.

Radhika, thank you again so much for being on this podcast.

Radhika Gupta (40:25)
Thank you for having me, Tim.

Tim Hatt (40:27)
It was great to have you, and everybody else, we will see you next week.

Have a great rest of your week. Bye-bye.

Radhika Gupta (40:33)
Thank you, everyone.