B2B2X is everything to 5G (with Amir Rao of AWS)
Podcast episode 42
The growth potential for the cloud under 5G relies on Business to Business to Everything. If we're to find the next big thing, telecom operators need to partner up and be flexible. AWS’s Amir Rao knows that the next killer app requires CSPs to recognize their role in this new ecosystem.
Below is a transcript of this conversation. Some parts have been edited for clarity.
Michael Hainsworth: The growth potential for the cloud isn't just in B2B or B2C, but business to business to everything. How critical is a platform-agnostic approach to the success of B2B2X? It's everything. Amir Rao knows that as the Global Head of Solution Portfolio in tech alliances for telecom at AWS, but ensuring developers of the next killer app can become hyperscalers like Netflix, Uber and Airbnb requires telecom operators to partner and be flexible.
Amir Rao: Partnerships are going to be in a very, very important aspect of building the 5G businesses across the globe. And for the use cases, the operators would have to partner with different ISPs, different CSIs, maybe at times, different enterprise customers who may want to build B2B2B or B2B2C use cases that would benefit from 5G latency and other parameters such as high throughput.
MH: And we will bring all of that together, the idea of B2B2C or B2B in general, that sort of goes out the door because to your point, this is a business to business to everything now.
AR: I'd have to agree. B2B2X, business to business to everything is going to be critical to deliver the key promise of 5G, to enable the innovation in the industry and the society to development of new use cases and new solutions that can benefit from cloud, and that can benefit from 5G network attributes or network qualities, which differentiate it from the 4G or some of the legacy technologies.
AWS innovation on behalf of our customers are providing them both the depth and breadth of cloud services to remove the complexity of our large number of ISPs and developers in building the new innovation solutions. And therefore, how we are thinking is to be able to extend AWS cloud services to 5G networks by a service called AWS Wavelength.
The key here is to be able to provide the ISPs and developers the tools that they need to benefit from the capabilities of 5G by extending AWS cloud services to AWS Wavelength. We can differentiate, we can help the ISPs, developers differentiate and build the use cases, which could be relevant to everybody.
It could be relevant to a healthcare system, it could be beneficiary to actually autonomous vehicles, it could be beneficiary to some of the computer vision work that's been going on in the industry, and it could be for a number of type of users. It could be for enterprises who want to build industry for auto solutions, it could be for common public safety, who are just trying to monitor traffic conditions or congestions, or for law and order maintenance, as well as imagining a B2B2C use case such as gaming, a low-latency gaming application which could sit and benefit on AWS cloud and actually can benefit from the capabilities of 5G.
MH: So let's talk about those new innovation services, automotive, smart city technologies, healthcare, they're all built on this next generation cloud technology. Give me some examples here. How is the healthcare industry leveraging edge cloud?
AR: The AWS Wavelength, which are AWS cloud services in the 5G networks, they enable the 5G application and use cases across a range of industries, and it does it by offloading data processing tasks from 5G devices to the network edge, how to conserve resources like power memory and bandwidth, the standard compute capabilities of the end user device.
And this allows the ability to run machine learning inference at the edge and provide results back to the requesting application in a near real-time application. One of the examples is healthcare. I mean, of course we've talked about autonomous vehicles, smart factories, with healthcare, you can really use machine learning for assisted diagnostics. You can use machine learning for interactive live video streams and other interesting use cases.
Specifically, we've been working with Avesha Systems, an AWS customer who has built an AI assisted model and inference at 5G edge to assist medical doctors to identify polyps during endoscopy procedures. This particular solution takes a video feed from the endoscopes during the procedure and run inference, machine learning inference at real time on that feed of the 5G edge, and then use the pre-trained models to detect the anomalies and highlight those areas to the doctors.
This particular case is also demonstrating that data and applications have to invariably use the services both in the cloud and at the edge as AI models are trained in AWS Cloud using SageMaker and other services while the inference is performed at the Wavelength zones at the 5G edge. Does that make sense?
MH: It does, it is fascinating because you're talking about some really core stuff, whereas everybody else seems to focus on the ultra reliable, low latency communications that brings us things like robotic-based distance surgery, but something as complex yet really important as dealing with endoscopic results is something that is really a foundation of healthcare-based AI.
AR: Yes, it does. And that is not to say that we are not working on an ultra low latency sort of mission critical nature of 5G. There are multiple use cases which can be enabled by 5G. And of course, low latency is a very, very important piece of it. But just the fact that that is not the only possibility that it opens up. I mean, just having AWS cloud at the 5G edge allows you the ability to open up different traditional use cases such as machine learning or augmented reality. And I say traditional of course, with the tongue in cheek because these things have not been there for decades. It's just like four or five years, but these things they have existed for the last five, seven years.
Now 5G is opening up and AWS Wavelength, at the 5G edge is opening up using some of these capabilities closer to the end user, which are of course going to benefit from the low latency, just by virtue of 5G, but also by having AWS Wavelength at the 5G edge, you are reducing the latency even further from a compute capability perspective.
MH: One technology though that does require URLLC is smart card technology. How is the autonomous vehicle industry approaching the cloud?
AR: The autonomous vehicle industry is approaching 5G and then cloud in multiple ways. So of course, it all starts from having very robust and large platforms in cloud, that could benefit from the scalability, reliability and the elasticity that the cloud offers. So that's one piece, but then it also requires some real time decisions to be made, and that is where 5G and potentially AWS Wavelength comes into play because 5G is going to offer extremely low latency for split-second decisions actually, millisecond based decisions to be made by autonomous vehicles.
And what can enable those decisions to be made very quickly, rather than traveling very large distances to get to cloud, you can really get to AWS Wavelength, which is sitting right at the network edge in the 5G network, and you can offer all of the compute capabilities that are required for the applications and those decisions to be made.
Now, whether those decisions are AI based decisions or whether those decisions are computer vision based decisions for a vehicle to navigate themselves, or whether those decisions are just to provide the entertainment experience, which is another part of the autonomous vehicle industry, because it's just not only autonomous, it's going to be a fully embedded experience for a passenger where they can... While the car's driving itself, they can pick and choose anything that they want to listen or watch. So I think the automotive industry in general is using both 5G and cloud for not only autonomous vehicles but to also offer a fully immersive experience to the passengers, which would define the next generation of car travel.
MH: See that's interesting too because if we spend all of our energy thinking about how you're just going to get the vehicles to talk to each other, that completely ignores a key component to the autonomous vehicle experience, which is that it relieves the driver and the passengers of focusing on the safety aspect of the commute, and they can focus on other things.
AR: So the tough process here, I think, at least I perceive it, or we perceive it at AWS is, it's going to be a function of once the vehicles become autonomous, you want to provide the passengers of those vehicles an immersive experience, whether those experiences are work-related or whether those experiences are just entertainment-related. For those passengers to be able to... If they too completely trust the fact that they are riding an autonomous vehicle that are just going to take care of the navigation and course correction to whatever it's worth, then naturally they will be able to use their time more productively, and productivity has its own definition depending on the persona of the passenger, persona of the customer that's riding that vehicle.
If you're a road warrior who want to be able to latch on to your emails and conference calls, then you'd need to have that capability from an internet service vision perspective and having the high speed bandwidth that is required to conduct your office work while you are traveling to and from, whether it's a city to a city or traveling from your home to the office.
And you could also be a younger passenger or any person who would want to just watch the latest show and consume content on demand. So we see both of these major types of categories but especially once autonomous vehicles become a reality, there's an all-immersive content experience is going to become a more of a need, not to say there's not a need today, because even when you are attending to all of the navigational needs, it's only the driver who is attending to the navigational needs and there could be other passengers in the car, especially with the apps like Uber and other ride sharing apps. So you want to offer the passengers of the car an immersive experience. What use cases will exist in parallel?
MH: Use cases with cloud vary dramatically depending on we were talking about here. For example, when it comes to public-centric B2B, how do you ensure flexibility for the end user when there's such a large diversity of the kinds of applications that come with say, a smart city, you're going to have a whole lot of sensors, you're going to have camera systems, you're going to have systems that require an immediate ultra low latency, but then you're going to have other systems that are going to require a lot of horsepower in the back-end.
Sure, yes, the cloud offers the capabilities for different use cases that I've mentioned before. I'd like to cite another example of a use case, which could be a great use case for public sector customers, which is we've worked with in AWS ISP called Cloud Vision. It's just focused on monitoring... I mean, I'm referring to it because you just mentioned crowd congestion, right? And monitoring crowd congestion with video cameras and LIDAR detection among multiple route sequence.
And so every time when congestion exceeds certain thresholds, alerts are sent off to staff, signage, public addresses and smartphone applications recommending folks, whether you are fans at a major game or weekend game, or whether you are trying to monitor traffic during a public protest or a public gathering, or whether you're just trying to, in a commercial setting, just provide workforce management services. So all of these are realities.
This addresses one piece of the question. So public safety could be a significant beneficiary from use cases like these. But then this specific application, whether you've worked thousands of sensors, which are providing temperature, let's suppose you are a firefighting agency. So not only you'd want to understand the congestion once you're entering into a building, into a large complex as to what's happening where, but you'd also want to understand where the fire is.
To your point, you want to check where the humidity is, and therefore all of that data through multiple sub applications can actually be taken to cloud, AWS Wavelength, processed there, and different types of insights could actually be managed. One insight would just take care of the congestion, control the congestion aspect. The other types of insights would actually be able to navigate the firefighting staff going into a building as to what to avoid and where to look for the fire or the source of the fire that needs to be extinguished in a jiffy. Does that make sense, Michael?
MH: It does. Five years ago under 4G, telcos, were building their own data centers. But let's talk about the partnerships aspect to all of this though, because over the last 12 to 18 months, we've seen CSPs recognize they need to partner with hyper scale cloud providers like AWS, but that requires a culture of change. What does that look like?
AR: I think it's more than culture of change, it's embracing each other and embracing each other's capabilities to do what's the right thing for the end customer. For AWS, the customer obsession sits right at the center of our business, of our innovation, of the scale that we have been able to create over the last 10 to 15 years.
And in that sense, what we really want to... The journey that we want to embark on with the CSPs is just to focus on the customer obsession. And I think, and I hope, and that's what the signals that we've got from different CSPs is that the CSPs are thinking, "Let's just think about who's our end customer, who's our enterprise customer or developer that we want to bring the capabilities of 5G?"
And if that requires working with an AWS or ensuring the AWS Wavelength can deliver the cloud services at the 5G edge, then they are very open because they want to leverage the capabilities of cloud that AWS has built, and also bring in the large number of ISPs and developers that AWS already has on its platform to be able to use the 5G capabilities.
So in a nutshell, I think the thesis of any potential partnership should be based on working backwards from the customer. And we are seeing a lot of signals off that, and actually that probably tipping the... Or breaking the ice, in fact it has already broken the ice many many years ago, but both sides are accelerating together, AWS and the service providers to create those immersive experiences for the end customers.
MH: Let's expand on that because you're accustomed to partner relationships, but CSPs are used to being in charge. What's your best practices advice for a CSP who then partners with those developing applications that leverage their cloud infrastructure?
AR: The key here is really to focus on understanding who are going to be the builders. I think you mentioned right at the beginning, who are the folks who will be building the Netflix of 5G, and other applications. So we need to cover all the bases. And you also already see different CSPs have set up innovation centers to try to open the talent across the nation here in the United States and also globally so that whether you can open startups, you can open ISPs into your programs or to expose to them the capabilities of 5G.
Now, if you want to do that, you have to appreciate that those companies are going to be much more nimble, they would not have the hundreds of people that generally a large operator or a CSP would have, or even an AWS would have.
So we need to appreciate, and both sides need to appreciate the difference of persona that is trying to meet in the middle to create a successful partnership. Specifically for CSPs, they have to appreciate that these smaller ISPs or developers are looking for faster results and they need to be provided with the resources.
So if both sides are meeting each other's demand, and if both sides are aligned in terms of the end customer result that needs to be driven, which can actually show both sides the path to not only meeting the customer need, but a path to revenue, a path to scalability. These are going to be at the center stage of some key tenets which will define whether a relationship or a partnership between the CSP and an ISP, or a CSP and a developer or a startup is actually going to be successful or not.
MH: So what you're saying is any partnership can only scale if it's solving a real customer problem and the CSP has a role to play in determining pain points?
AR: Right from the outset. Reality is, working backwards is in our DNA at AWS, and we believe it works for everybody, and especially two or more entities coming together to solve a problem, because if you eventually will not be able to solve the problem it just becomes a science project, eventually the parties would stop their interest, they'll stop investing, and it will not go anywhere.
As we said, success breeds success, generally in our society. So we need to look for those early proof points for any partnership to be mutually inclusive, and then to work backwards from a customer need so that we are focused on solving that need and being able to solve that need will actually take us into the next level of our journey as partners together, whether you are any two companies or three companies coming together, then you scale the business.
MH: The next startup may be the 5G equivalent to the next Netflix, Uber, or Airbnb, but how does CSPs ensure they're giving their partners the right tools to become the next disrupter?
AR: They'd have to offer, as you said, the right tools. The right tools would include capabilities, exposing 5G capabilities. Potentially providing and leveraging partnerships, for example, the one that AWS has been forging or building with the CSPs to offer AWS Wavelength and AWS cloud services through AWS Wavelength at the 5G edge. We need to expose those capabilities, whether it's the CSP or whether it is AWS.
And we need to ensure that not only do we expose those capabilities, we also do a bit of handholding in terms of understanding the art of possible, and actually being open to getting the feedback from those ISPs. Maybe they actually help us unearth some of the attributes of 5G and cloud services which we are not thinking.
And this is part of the working backward process, because you are trying to think of a customer in two dimensions. One is the end user and the other one is the actual builder. And in that case, I think really what we would need to appreciate as industry stakeholders, and especially as CSPs, that in order to be able to build those next generation killer applications as we call them or hugely successful use cases in future, we'd need to be open to listen to the creators of those use cases, and then being able to offer them a flexibility and platform agility, whether it is 5G network itself, or the cloud services from AWS perspective, that will determine our success together.
MH: What you're saying is we need to ensure that CSPs don't get stuck in their own bubble, their own echo chamber, or have blinders on about this transformation that they know they have to undergo.
AR: I would say this is true for anybody. Anybody entering into a next era of technological innovation, whether you are a CSP or an AWS, or you are just a healthcare company. We can't afford to limit our set of potential permutations and combinations. That doesn't mean we change the business outcomes, we could still be chasing the same business outcome or, as I was explaining earlier, the working backwards outcome that you want to achieve, but not limiting the means to get to those business outcomes is going to be key and being open to understand that while we, as the heavyweights of the industry are thinking at the CSPs, they are thinking that for example, the killer application is going to come from, as you talked about earlier in the discussion about the low latency aspect.
Maybe it's something else, who knows? And that requires still the 5G use cases. So yes, In short being open to the feedback that we get from our developers and ISPs who are actually going to be building those use cases is going to be hugely important.
MH: From your personal perspective, what's the one 5G killer app that you're most looking forward to?
AR: I'm not a big technology geek, my colleagues -
MH: Wait a minute, you were employee number one at AWS, how are you not?
AR: No, I was not employee number one at AWS, I was one of the first guys hired for the telco business unit. But still taking on that question, I would just want to see a lot of new use cases emerge, which could actually meet the real time requirements. And these could be like, for example, on a personal level, I can tell you I've got a son, he's 14 years old, he has cerebral palsy, so I really hope that autonomous vehicles will be a possibility because I know he would not be able to drive. So you got me there, but this is... If in five, six years, I don't know how long. If autonomous vehicles would be a possibility, it'll greatly personally benefit me and my son, that'd be a good thing.