Guy Daniels, TelecomTV (00:09):
Welcome to our special webinar, The Fast and the Wireless: Connectivity, Network APIs and the Race to Scale Autonomous Mobility. Hello, I'm Guy Daniels. Robotaxis are already on public roads. Autonomous logistics vehicles are moving goods. Connected mobility platforms are scaling fast. But as the autonomous vehicle industry accelerates, can the network keep up? What does it really take to put the right connectivity infrastructure under the next generation of mobility services? And what roles do network APIs, programmable connectivity and cross-industry collaboration play? Well, we intend to address these questions over the course of this webinar, and we also have a presentation and a live demo for you, both coming up very soon. But first, let's meet our guests. I'm delighted to be joined today by Sridhar Gollapudi, who is Global Telco Market Lead for Google Cloud; Kurt VonEhr, who is Founder of Strata Wireless; Tuuli Tolmats-Aia, who is COO and Co-Founder of Elmo Remote; and Dinos Katsaros, who is R&D Manager at ICCS/NTUA Greece and Vice Chair of the 6GIA Connected and Automated Mobility Working Group.

(01:38):
Hello everyone — it's really good to see you all. Thanks so much for taking part in today's webinar. Now, before we start our discussion, I think we should establish why this is a GSMA Open Gateway webinar and the role being played here by GSMA Fusion. Earlier this week, I spoke with Natasha Nayak, who is Senior Director at GSMA Fusion, and I asked her to elaborate.

Natasha Nayak, GSMA Fusion (02:07):
GSMA Fusion is the outside-in demand engine for GSMA Open Gateway. While Open Gateway focuses on exposing network capabilities through standardised APIs, Fusion starts from the opposite direction. We begin with enterprise problems and work backwards to determine what network capabilities are needed to solve them. Our role is to ensure that the voice of the enterprise is heard within the telecom industry and translated into globally scalable network solutions. We work closely with enterprises across four priority verticals: automotive, aviation, media and entertainment, and financial services. In each sector, we engage business leaders to understand their operational challenges, identify where connectivity can create real value, and convert those requirements into industry-wide demand signals. A good example is the automotive industry. Through our work with technology service providers, OEMs, industry bodies and others, we've identified a critical need for applications like autonomous vehicles, remote operations, and critical software updates to receive predictable, reliable network performance across markets.

(03:18):
Fusion has translated these business requirements into requirements for Quality on Demand APIs which address these pain points and are now being enabled globally through collaboration across the Fusion ecosystem. Similarly, in financial services, we are working with organisations focused on fraud prevention, where solutions like SIM Swap, NumberVerify, and Scam Signal can help reduce fraud while maintaining a seamless customer experience. Where existing capabilities are insufficient, Fusion helps shape the next generation of requirements that can become future industry solutions. In short, Fusion identifies enterprise demand, aligns it with operator capabilities and channel partner enablement, and helps transform real business challenges into globally scalable network solutions.

Guy Daniels, TelecomTV (04:06):
That was Natasha Nayak, Senior Director at GSMA Fusion, discussing the role of the organisation. So let's now have a chat with our guests. To set the scene: autonomous mobility is moving from pilots to commercial deployments — we all see this. I'd like to ask each of you from your own perspective: what's the biggest challenge to scaling these services, and why does connectivity sit at the centre of that challenge? Sridhar, I'll come to you first if you don't mind.

Sridhar Gollapudi, Google Cloud (04:48):
Yeah. So first of all, there are quite a few challenges in scaling from pilots to commercial deployments. First and foremost, user safety is a big challenge that the industry has to work through. There are others like regulatory compliance, infrastructure build-out, partnership build-out, and so on. But one that stands out within this set of challenges is reliability at scale. And connectivity sits at the centre of that challenge, because you need reliable performance for service continuity for the fleets deploying this autonomous mobility. What that means is connectivity needs to move from a best-effort service to a dependable, automated network performance that measures and constantly enforces metrics like latency, bandwidth, and determinism. And what this enables is a model where you have network-based performance continuity combined with the vehicle-resident safety operators — and that combination enables the systems to scale confidently.

Guy Daniels, TelecomTV (06:07):
Sridhar, thanks so much indeed. It's great on this particular webinar to have so many people from such diverse parts of the ecosystem to give their views. Tuuli, I'm going to come across to you next. What do you see as the main challenge in scaling out?

Tuuli Tolmats-Aia, Elmo Remote (06:23):
I very much agree with what was just said. It is absolutely essential to have a reliable network, but the maturity of autonomy and remote driving comes across four different segments. First, the technology itself has to be ready and mature — and remote driving, as I'll show you in a minute in the demo, is there already, but it relies on the network. Then, after we have the technology and the network, there's the legal framework that has to allow for it to be widely scaled — also a challenge. Then there's public acceptance: people need to be comfortable sitting in a car that's remotely driven or fully autonomous. And we mustn't forget that it also has to make financial sense — the ROI needs to be positive as we start doing this at wide scale.

Guy Daniels, TelecomTV (07:18):
Tuuli, thanks so much indeed. Obviously a lot of challenges going from pilots to scaling out. Kurt, are you seeing a similar set of challenges?

Kurt VonEhr, Strata Wireless (07:28):
So we see in pilots that it's quite easy to operate using existing LTE and 5G services, but as you scale up you start to experience long-tail problems with larger fleets. You get those edge cases — you find congestion, you find day-of-week and time-of-day effects, holidays, different events that emerge that typically aren't forecasted. And so having a pilot in a very narrow operating condition does not necessarily translate to connectivity at scale and reliability at scale for autonomous systems. You start to see a lot of challenges emerge as you have a higher volume of operations. And so one of the things that can be improved for large-scale deployments is having reliability and homogeneous network deployments with APIs that provide predictable and reliable data bearer information — so that an operator has confidence when deploying in that operational domain that the vehicle will have consistent connectivity throughout the entire navigation.

Guy Daniels, TelecomTV (08:48):
Great. Thanks very much, Kurt. And let's round out this opening question by going across to Dinos. What do you see as the challenges?

Dinos Katsaros, ICCS / NTUA (08:57):
I pretty much agree with what my colleagues have said. Service reliability is very hard to achieve at great scale, because at the intended scale of global deployment and operation, we need coordination — coordination across the actors in the field — and coordination points to communication, and communication points to good connectivity. Then comes another dimension: the wireless medium is by definition unreliable. So we need to address the increased complexity of the policy setup. In all, I think reliability is the major obstacle, and overcoming it will generate trust — user trust in the services — which is what drives adoption.

Guy Daniels, TelecomTV (10:10):
Dinos, thanks so much. Thanks everyone for that opening perspective on where we are and the challenges ahead. Let's go a little deeper now. I'd like to ask you about what the network needs to deliver. Are we talking about speed? Reliability? Low latency? Or something else entirely? Kurt, I'll come to you first.

Kurt VonEhr, Strata Wireless (10:42):
A lot of what the network needs to provide is reliability, so that the UE itself is able to understand what the network can offer at any point in time. Right now it's a lot of guess-and-check from the UE to understand whether there is a reliable and stable link. Being able to provide consistent APIs across multiple network operators enables scaling across cities within a particular nation or across various markets internationally.

Guy Daniels, TelecomTV (11:15):
Thank you, Kurt. And I'm going to put the same question to Tuuli. What is it specifically that you need the network to deliver?

Tuuli Tolmats-Aia, Elmo Remote (11:26):
Building on what Kurt just said — we know today that if you look at the network, it changes every second. There is no network today that we can rely on to deliver smooth driving throughout hours and hours of the day. That's the problem. It's the stability at every single second that we would need.

Guy Daniels, TelecomTV (11:52):
Great. Thanks very much. So you talk about stability there. One of the issues that we hear an awful lot about at the moment is the difference between uplink and downlink. How important is uplink performance compared to downlink in real-world autonomous or tele-operated mobility? Tuuli, maybe I can come straight back to you first.

Tuuli Tolmats-Aia, Elmo Remote (12:14):
Uplink is the core. As you'll see in the demo, we have a car and we have a station. The car is the one moving around in the city, and the car needs the uplink to make sure that the video feed from the car goes to the driver's station. So uplink is absolutely critical — it's the thing that needs to be maintained every single second to ensure safe driving.

Guy Daniels, TelecomTV (12:37):
Great. Thank you. And Kurt, the uplink — there's a lot more emphasis on uplink performance when it comes to autonomous applications.

Kurt VonEhr, Strata Wireless (12:46):
Yes. Uplink is primarily broken into about five different traffic classes: command and control, telemetry, remote assistance, payload data, and lifecycle data — and each one has different network behaviour requirements. So you can triage based on network conditions which types of traffic classes need to be focused on by the network end-to-end. The UE can classify them based on application demands, and each has a set of criteria regarding bandwidth required, latency, jitter, and availability. A command and control signal, for example, needs to take priority over pretty much any other signal. But there are other things at lower priorities — such as offloading payload data — that allow you to be flexible about when that is uploaded. All of these are focused on uplink specifically.

Guy Daniels, TelecomTV (13:48):
Kurt, thanks very much indeed for clarifying that. This really is an exciting sector and it continues to amaze us all. We're going to pause our conversation for just a moment, because as Tuuli alluded to, she has something very exciting to share with us. This is going to be good — Tuuli is going to give us a live demo of Elmo Remote, and it promises to be very hands-on. Over to you, Tuuli.

Tuuli Tolmats-Aia, Elmo Remote (14:16):
Hello, hello. I'm just walking into the car, so bear with me. What we do at Elmo is we make old cars do new tricks — we take existing cars and we make them drive around. Now I'm in an Elmo car, and my colleague — I've asked him to give me a spin around a public road. So this is remote driving. It means there is a person controlling the vehicle — and I'll show you the driver as we come back from the little tour. He's driving me around through a screen. It's a road-legal technology. The Estonian Traffic Department decided four years ago that we can be on public roads without a safety driver and drive at normal city speed. Now I'm going to fasten my seatbelt to be a proper citizen and get back to the screen. Now there are three important things about this technology that make it work very well.

(15:16):
Number one — and for this discussion, the most critical thing — is the unique use of parallel networks. We don't take only one LTE channel; we take multiple and make them run in parallel. And this is to overcome the network fluctuation that happens every single second. We have a proprietary algorithm that then picks the packets at the station side and reassembles them. There's no switching in between — it's an actual parallel use of networks. And you can only drive when you're connected. If you're not connected, you cannot drive — as simple as that. So that stability is the absolute essence of this technology. The second thing is that if you have good connectivity, you can then be driving from anywhere with anything. We have no limitation on where the driver is and where the car is.

(16:19):
Our current record is driving from 4,300 kilometres away. Really — if you have a good network, you can drive over any distance. And by the way, at speed. So if you have a good picture and good connectivity, there's also no limitation when it comes to speed. As long as the car can go fast, you can too as a remote driver. Our current speed record is 157 kilometres per hour — we achieved that in a partnership at one of the Mobile World Congresses with GSMA, on an F1 circuit, driving the same car I'm sitting in at 157 kilometres an hour. So from a technology point of view, there is no limitation. On public roads, of course, we keep to the permitted speed. And there are many, many use cases where you can use the efficiency of this technology.

(17:19):
For example, in car sharing, you can do doorstep delivery of the car, or you can park it for the customer — they don't need to do it themselves. It gives a better service to the customer, and at the same time you can redistribute the car-sharing fleet to where the demand is, improving turnover and achieving efficiencies. The same applies in logistics. We know there is high demand in logistics for better working conditions and greater efficiency. But we are also war-proven and active at the frontline saving lives — which is very critical today in Ukraine — and I'm sure we'll need that capability in the future too. So from a technology point of view, there are many, many use cases and the demand is only growing. So now you've seen the car side, and the car is just parked.

(18:14):
I'm going to quickly run inside and show you how it looks on the station side. Bear with me — we'll do a quick changeover, going to the workshop. And here at the workshop we work on many different platforms. As I said, it can be really any type of vehicle, and there are different stations. We're going to the one that's just here, and we say hello to our driver now. And this is how it looks: you have a screen, you have a driver, you have a wheel and pedals — as simple as that.

Guy Daniels, TelecomTV (18:55):
Fantastic. That was so much fun. Tuuli, thank you very much — that was terrific, and it really was a live demo happening as we were watching it. Right, as Tuuli gets settled, we will move on to talk about programmable networks. Let's now bring Sridhar and Dinos into the conversation. As vehicles become increasingly software-defined and AI-driven — as we've seen in the demo — how do network APIs and programmable connectivity change what's possible for developers and enterprises? Sridhar, let me come to you first.

Sridhar Gollapudi, Google Cloud (19:43):
Yeah. So first of all, software-defined vehicles — what that enables in terms of programmable networks is: beyond the autonomous mobility demo we just saw, which was super interesting, there are other use cases already being deployed. Simple things like reliable, safe and secure OTA updates — most vehicles are now operating in that environment, and connectivity is a key underpinning for those use cases. And then you have rich, interactive in-vehicle infotainment use cases that are coming up, all enabled by these connectivity services. They encompass a lot of different use cases within the SDV paradigm. So what are these APIs and programmable network features being supported by these use cases? You've got plenty of them. You've got the Quality on Demand API that we talked about at the beginning of this session.

(20:50):
There are others that are also published and live across the operator ecosystem: edge discovery, location verification, connectivity insights, identity — something that we heard about in the introduction from GSMA Fusion. So all of these are important APIs that are readily available today. And that brings the programmability aspect of the network. But what really kicks it into a higher gear is that these APIs are completely abstracted away — the network layer is completely abstracted away from the developers — with agentic AI built on top of the network layer. What that enables is for agents to act on behalf of developers, essentially acting as an orchestration layer to manage this complete end-to-end system. And that is super important, because at this point what this enables is for developers to deploy their applications in an intent-driven, automated fashion — where you have these systems actively monitoring and observing network performance and adjusting it to match the intent that the developer intended for their service.

(22:20):
So that is key both in terms of discoverability of these applications and also the deployment of the services that developers want. On top of that, what these programmable networks and the underlying APIs enable is guaranteed trust — making sure that the network performance that has been promised or committed as part of this programmability is truly being delivered, via cryptographic attestation at the API layer. So you have policy-compliant deployment versus a best-effort connectivity paradigm. That's key — a key feature of how programmable networks are enabling these use cases. And lastly, interoperability: with standardised APIs, you have a broader ecosystem being built out across geographies, across different types of telecom operators — both from the wireless and fixed sides of the network.

Guy Daniels, TelecomTV (23:46):
Sridhar, thank you very much. Great insight there. And Dinos, I'm going to come to you as well. How do you see network APIs and programmability changing what's possible for enterprises and developers?

Dinos Katsaros, ICCS / NTUA (24:01):
Well, I think the previous answer made the benefits quite clear. This is a different paradigm because it allows direct communication and interaction between the services and the network. And as correctly said, it abstracts the complexity of the network — which allows the developers of automotive services to make decisions without needing to be mobile network experts. So this gives the service the opportunity to get information from the network — useful information, for example the location of certain users — it can provide information to the network to trigger some AI inference, and it can ask the network to be reconfigured.

Guy Daniels, TelecomTV (25:16):
Great. Thanks, Dinos. Well, let's hold it there for a second. We're going to pause our conversation because Dinos has a presentation to give us. The European ENVELOPE project looks into the evaluation and validation of connected mobility. Here is Dinos to explain more.

Dinos Katsaros, ICCS / NTUA (25:36):
Okay. So this is about a 6G SNS JU project called ENVELOPE, focusing on the evaluation and validation of connected mobility in real, open systems beyond the 5G system. The project focuses on exactly what we were talking about before regarding programmability and the exposure of APIs — giving developers of CCAM services and automotive services the opportunity to get information and data from the network about current conditions, provide information to the network so it can configure itself better, and guide the dynamic reconfiguration of the network subject to their needs. This project is particularly supported by the SNS JU and the 6G Industry Association, which fosters collaboration between stakeholders in the broader ecosystem. On the APIs and programmability — we are talking about support and exposure of APIs ranging from Quality on Demand, device location, geofencing, predictive connectivity data, and so on, including other types of APIs standardised by ETSI.

(27:32):
Each one comes with a series of benefits and facilitates the programmability not only of the network but also of the service itself. We've built these APIs and the corresponding network functionality across three trial sites in the Netherlands, Italy and Greece, and we've put them into practice with real use cases — such as advanced service delivery for automated driving vehicles at the Italian trial site, vehicle testing with mixed reality and tele-operation in the Netherlands, and applications for data sharing for real-time situational awareness. We have several demos — not live like the previous one, but to give an example of what we are talking about regarding the operability of APIs: we've implemented a demo for a C-V2X service, which means the front vehicle streams video of what is happening ahead to the vehicle behind it. Then we get a predictive QoS notification.

(29:01):
This means that the network — based on input data from the vehicle — can predict that for some reason, in this case a background traffic increase, the performance of the service will deteriorate, and takes proactive action to request reconfiguration of the network. In one case, this can be Quality on Demand for the uplink of the front vehicle — more resources for sending video towards the network. In other cases — for the following vehicle in this example — we have the activation of ATSSS, which is a 3GPP feature that allows the use of multiple network interfaces from the same vehicle. So in this particular example, the vehicle can also use an ITS-G5 radio access network, or even a satellite link if integration with non-terrestrial networks is available. This provides more resources to the following vehicle, maintaining a stable experience — all made possible through the interaction with the network, via the QoD or ATSSS-based solution.

Guy Daniels, TelecomTV (30:35):
Great. Thanks very much, Dinos. Substantial work going on there with the ENVELOPE project — a lot to digest. Right, let's return for the final part of today's webinar. We know that no single organisation can deliver autonomous mobility alone. So I'd like to ask you all: what does successful collaboration between automotive companies, telcos, cloud providers and solution partners actually look like, and where do you see the biggest gaps today? Kurt, if I could start with your views.

Kurt VonEhr, Strata Wireless (31:13):
Projects like the CAMARA API will allow UE hardware developers on the OEM side to scale across markets without having to do bespoke implementations of their connectivity manager. Right now, in order to access the network and understand link stability, you have to do a kind of guess-and-check. But these APIs provide Quality on Demand services, they provide link data, they provide information that allows you to intelligently have multi-carrier implementations. We discussed this earlier — having multiple MNOs in each vehicle to improve latency, jitter and availability. And as you scale across multiple MNOs, you end up substantially improving those link connections and data bearers. In order to understand link quality, you typically have to do an implementation at the application layer. But these APIs can provide that and allow intelligent switching in advance, providing much better availability and reliability for each one of the service operators.

(32:32):
So for any OEM that's implementing an autonomous service.

Guy Daniels, TelecomTV (32:36):
Great. Thanks, Kurt. So improving link quality — very important. Tuuli, let me come across to you. How do you see collaboration working and benefiting everyone, and where are the biggest gaps you see today?

Tuuli Tolmats-Aia, Elmo Remote (32:50):
I think a lot of the things we need have been developed. The capabilities are there, the APIs are there, but we are not using them widely or smartly enough globally. So there's a big challenge around them being available everywhere and at a good price. When you are a business trying to deploy as many vehicles as possible across as many use cases as possible, you want the capabilities to be actually available — not just theoretically developed and existing — and to be available at a good price.

Guy Daniels, TelecomTV (33:25):
Global availability — we hear that so often. Thank you very much, Tuuli. Sridhar, let's come across to you in terms of collaboration and where we need to work harder.

Sridhar Gollapudi, Google Cloud (33:37):
Yeah. In terms of collaboration, it starts with shared value creation. We need the experts in this field — the application developers and the automotive OEMs — to tell the rest of the ecosystem what is important to them. It starts from the customer experience, it starts from what the service deployment scenario looks like, and then the rest of the value chain builds an alliance on that approach and develops the appropriate technology. The technology is not the key here — it's the user experience. That's where I think a good collaborative model works. And we are seeing some of that actually coming together now. Of course, the corollary is that once you have the shared value creation, you need to align on the business models as well — that's a key follow-on step.

(34:38):
Another aspect of collaboration is where, at each of these layers, you have telcos making sure the APIs are standardised for global deployment, hyperscalers making sure the infrastructure is ready for global scale-out — both in terms of AI infrastructure and the broader compute and storage — and then automotive OEMs making sure that, A, the service requirements are being properly defined for the rest of the ecosystem to work on, and B, that once the rest of this technology is ready, they are ready to go live with their services. The third aspect is that for this to work at scale, you absolutely need to build a trust model for data governance and sovereignty across this value chain.

(35:43):
That's important because, looking forward, what we see is intelligence being essentially embedded across each of these different aspects of the value chain. And for true intelligence to be built in end-to-end, you need these data governance models to be in place. In terms of the big gaps — first and foremost, you have different players in these ecosystems each operating at their own velocity in terms of how they develop products and go to market. And clearly that creates a gap in terms of what different players believe is ready — as Tuuli mentioned: while the APIs are already there, getting to the next step is something the industry needs to work towards.

Guy Daniels, TelecomTV (36:45):
Sridhar, thank you very much. The idea of matching velocities between sectors — that's really interesting and one to watch. And Dinos, to you — what does successful collaboration look like, and are there any gaps you identify?

Dinos Katsaros, ICCS / NTUA (37:00):
I would like to highlight Tuuli's point. In several cases — this depends on the actual application and service — the technology is not the limiting factor. It is mostly about the business aspects and what actually gets implemented. I recall a case from a previous European research programme where we were working towards cross-border mobility solutions, and we had the following interesting situation: the discussions led to the conclusion that vendors can implement a solution — there is a solution and they can implement it — but they won't do it, and operators will have difficulty deploying it, unless there is demand from the automotive sector. And then the automotive sector would not engage in adopting these kinds of solutions if they are not readily available and deployed. So what I'm trying to get at with this example is that in several cases, how demand meets supply — or the other way around — makes a significant difference.

Guy Daniels, TelecomTV (38:40):
Absolutely, Dinos — thank you very much. It certainly does. A follow-up question then — Sridhar, if I can come straight back to you: how important is consistency across operators and markets for the automotive OEMs?

Sridhar Gollapudi, Google Cloud (39:00):
It is very important. In fact, we've been talking about it and demonstrating that use case a number of times with our telco partners — where it's not just between different wireless providers that you need interoperability, you definitely need interoperability between wireless and fixed telcos as well. And now we're slowly getting into NTN — non-terrestrial networks — and satellite operators as well. It's important for a number of reasons. First and foremost, it provides the guarantee of network performance for fleet operators and enables a develop-once, deploy-everywhere model. And it's also key for ensuring reliability, because if you have interoperable networks, you have the optionality of failover mechanisms for automotive OEMs to use in case they need an alternative path for service assurance.

(40:08):
So it is very important indeed.

Guy Daniels, TelecomTV (40:10):
Thanks very much, Sridhar. And Dinos, same question to you — how important is consistency across automotive OEMs and operators?

Dinos Katsaros, ICCS / NTUA (40:21):
It is of paramount importance. This also points to service continuity issues — I'm referring now to the pan-European ecosystem, where we need not only some APIs and features implemented, but we need the same features, following the same standards, all across Europe. And there is a potential role for regulatory frameworks. The market can lead to some developments, but to some extent there could be space for regulatory authorities to try to affect, boost and support the market — through some minimum set of requirements regarding the available features and available APIs — so as to achieve this consistency across member states, across PLMNs, and so on and so forth.

Guy Daniels, TelecomTV (41:50):
Thank you very much, Dinos. And I'll put the same to you, Kurt, on this follow-up. From your perspective and what you need, how important is consistency across not just markets, but also operators?

Kurt VonEhr, Strata Wireless (42:02):
To echo what everybody else has been saying — it's incredibly difficult to design bespoke APIs and applications for every single municipality, state and nation. Having that consistency across markets enables a streamlined approach to software development, hardware development, and enables a much faster deployment that is also much more reliable. Having a consistent interface to develop against becomes of the utmost importance. There is an inherent trade-off right now — as was mentioned around the business case. There is a need to understand the growth of the autonomous market in the context of consumer handsets, which are primarily focused on downlink. That tension does exist. But as we start to see more adoption, there's a positive feedback loop: more autonomous vehicles on the road drives stronger demand and a higher use case value from these fleet deployments.

(43:11):
And so that will build on itself and create increasing demand for an uplink-centric and harmonised API ecosystem.

Guy Daniels, TelecomTV (43:21):
Great. Thanks very much, Kurt. And picking up on what you've just said, let's look ahead. For our final question — if we are sitting here three years from now, what capability or industry shift will have had the biggest impact on scaling connected and autonomous mobility? Kurt, I'll come straight back to you.

Kurt VonEhr, Strata Wireless (43:47):
I think one of the biggest impacts would be having dedicated DNNs, network slicing, and Quality of Service — like Quality on Demand — being able to have that predictable data bearer and know that it's available in advance, rather than hoping it is as you drive into those operational domains. Right now, if you're triplicating across multiple carriers, it's a very operationally expensive endeavour. And so if a network operator is able to deploy that consistently and enable roaming partners to participate in it, you can reduce operational costs and improve the overall performance of the autonomous vehicles themselves.

Guy Daniels, TelecomTV (44:42):
Great. Thanks very much, Kurt. And Tuuli, let's come back to you for a final comment. What would you like to see? What do you hope will be the biggest capability that makes a profound shift in how we can scale out this market?

Tuuli Tolmats-Aia, Elmo Remote (44:59):
Dinos said it very well — it's today like a chicken-and-egg situation. And Kurt explained very nicely what needs to happen: that Quality on Demand APIs actually need to be there and work at scale. I believe that in three years' time, we will be in that new world where the demand has materialised, where the hype has moved into everyday practice. On one cell tower, you won't have one car driving remotely or autonomously — you'll have a hundred of them. And then it's not even a choice whether we have Quality on Demand available. You have to have it for the business case to successfully work.

Guy Daniels, TelecomTV (45:43):
Thanks so much, Tuuli. And Dinos — if we are going to scale, what has to happen in three years from now? What capability or industry shift has to be in place?

Dinos Katsaros, ICCS / NTUA (46:00):
I will use a very popular term: AI. This is the key to taming the complexity that comes with scale. We will have to build trust in AI-based solutions — which do come with some risk in solving difficult problems — particularly in what concerns the management of the network itself under the dynamic conditions of mobility. What I'm trying to say is that a QoD request may come from multiple vehicles simultaneously, as was correctly said. And we need to ensure that this is feasible, and we need clever ways to achieve that. AI will be a key enabler — with whatever frameworks or schemes. It doesn't matter. AI will deliver the automation needed to solve complex problems so that networks can actually support these QoD — or whatever requests come from the service layer.

Guy Daniels, TelecomTV (47:32):
Great. Thank you very much, Dinos. And finally, Sridhar — we all want to scale this sector. Looking ahead three years, what do you think has to happen? What's going to be the biggest shift or development that enables this scaling?

Sridhar Gollapudi, Google Cloud (47:49):
I think Dinos already touched on it briefly — he called it AI. We'd like to call it intelligence. Autonomous anything, for that matter, is only as good as the underlying intelligence of the systems that underpins the autonomous operations. And that is key for autonomous mobility as well. So what I see in three years is this: the scale-out is going to happen — I echo what Tuuli already said — but for that scale-out to happen, you would truly have intelligence built in, essentially embedded across these different nodes in this ecosystem, from the user interface to the autonomous vehicle operators to the OEMs, the hyperscalers or cloud providers, and finally the telco network as well. And so you have intelligence built in to make sure that each of these different nodes is operating at optimal performance and efficiency in an autonomous manner.

(48:59):
And then also communicating with each of the other systems and making intelligent decisions in real time. In the end, it's all about having a seamless, intent-driven experience for the end user. That's really the shift we see coming — we are already seeing it — but that's what is really going to make this scale up.

Guy Daniels, TelecomTV (49:20):
Great. Lots of work to do by all partners over the next few years. Well, we must leave it there. Thank you all very much indeed for taking part in today's webinar. And if you'd like further information on the topics featured in this webinar, please just follow the links below the video. For now though, from all of us here, thank you for watching and goodbye.