Amazon Web Services (AWS) will, at some point, replicate its managed private 5G service offering in global markets, it says. The product, currently limited to CBRS installations in the US, has underpinned 200-odd private cellular deployments so far – mostly LTE-based, mostly using componentry from JMA Wireless and Athonet, always pegged to AWS. Its availability, formalised as AWS Private 5G in late 2021, was a major statement in the market. But it remains a parochial US affair – in a rapidly globalised private 5G market.
In conversation with RCR Wireless, Ishwar Parulkar, chief technologist for edge and telecoms at AWS, confirms that the Seattle-based cloud giant will make the service available, including with ‘vertical’ spectrum access, in other geographies in due course. He notes the CBRS model – differently adopted in Germany, the UK, Japan, notably, and elsewhere, by degrees – provides enterprises with the chance to own, even to manage, their own bespoke private cellular networks without interference from the traditional telecoms operators.
But AWS is in no rush. It acknowledges the value of independent private cellular to a part of the market; but it recognises as well that the enterprise space is practically endless, and that it has a primal business to transform the carrier sector itself, and to optimise public operators as suppliers of new digital enterprise services. To this end, AWS announced a parallel offer at MWC a couple of weeks back, alongside a tranche of operator change-projects, to enable carriers to offer the same managed private 5G service in local markets.
The difference is telcos can take charge of the customer relationship, including the private 5G service billing and management. Parulkar is sanguine; the new Integrated Private Wireless service uses telco infrastructure but is pegged to AWS, ultimately so AWS is supplying the ’over-the-top’ applications, and likely gets the workloads as well. It says, in email, it wants to build “use cases that offer more industry-specific private wireless approaches which leverage AWS compute services to help drive outcomes”. P5G is a more “DIY approach”, it says.
And amid all the jostling in the supplier market, as different sides jockey for position, AWS just sounds magnanimous; like it will make its own offer, under its own steam, in non-US markets when it gets around to it, and is just pleased the game is afoot.
Well, of course it is, and of course it does; the exercise is not about selling networks, but about enabling enterprises, as Parulkar makes clear. AWS most wants to carry the workloads, wherever they are from and wherever they go. Which makes it a good partner, presumably. It sounds like one; it has no interest in buying-in cellular componentry to make its own network proposition, as some others, says Parulkar. It will go on selling others’, he says, and manage the service where the use case is simple, and enable the ‘over-the-top’ supply to make it tick.
We are beginning to ramble; the conversation with Parulkar is a good one, and printed below; it works as a useful counterpoint, as well, to the IT narrative that has developed in the private 5G space of late, and which has featured so heavily in these pages recently. All the answers below are from Parulkar; a parallel interview with AWS about its IoT game is available here.
…
If we can zoom-out for a moment; just talk through the AWS strategy in the telco space, and explain where private networks fit into that.
“So, it has been a journey for us, both ways: to help the cloud community to understand telco, where the protocols are different, the ecosystem is different, the go-to-market is different; and on the flip side to help telcos to understand how to bridge the gap to the cloud. We have built capabilities to run telco workloads and solve telco use cases in the cloud, and we have engaged with vendors to build their network componentry and functions in a cloud-native manner, and with operators to start adopting it. That has been the journey.
“But there are other pillars to our strategy, as well. Enterprises are moving their workloads to the cloud to reduce costs; we are bringing that to telcos, as well – not just in enterprise operations, but in network operations. A third pillar is about using AI/ML to harness all the data telcos have – in the network, the devices, the applications – to reimagine the customer experience, and to help them build new applications and services, and to reduce churn. And the last pillar is around monetization, which is where private networks are important.
“The value has moved above the connectivity layer in the last decade, to services on top – to devices and apps. We are helping telcos regain some of that value in a couple of ways. One way is with private networks, by offering cloud services that enable telcos to get into new verticals; another is by integrating the cloud into the network at the edge – to enable apps and services that use both. Which is the last way – to build new services and solutions that use both of our assets, so they can go beyond simple connectivity and grow their top line.”
So, just explain that private-networks monetization piece – is the new money for telcos in supplying the private network, or in somehow offering services on top?
“Private networks are about more than just connectivity. Enterprises don’t really care about the network; they don’t care about which core or which radio, and they don’t generally want to manage the network. They want it to run apps on top. Manufacturing, say, wants the network to run digital twins and predictive maintenance. And telcos can get into new verticals because of the applications on top. That is the key part here. That is where the monetization is.
“The network is not just for mobile phones; it is for key applications for enterprises. Which is why partnering with AWS is a differentiator – because we bring the cloud to the edge. It is not just an edge server for a private network; it is the cloud. Which means you can run AWS services on top; from day one, as soon as you build the network. It is very easy to build these applications on top of the private networks with AWS – which is where the monetization piece for telcos comes in.”
Clearly for AWS, you are just seeking to enable the service providers in the enterprise space, and it is up to them to make good on the opportunity with enterprises. But are operators doing a good job at this – at supplying and monetizing the enterprise market? Because there is an argument that this is not an ‘operator’ discipline, but a more specialist consultancy sale.
“The private wireless space is very large. We’ve been working on it for three years. We have over 200 private deployments in the field, created as solutions using AWS services – [private wireless] is a solution and not a service. And none of them feature operators; they use partners like JMA Wireless and Athonet, and various independent software vendors (ISVs). And through this, we have learned a lot. And the conclusion is that it is a really vast space – not only in terms of the variety of deployments, but also in terms of use cases.
“Deployments go from small two/three-radio sites to whole cities with hundreds of small cells. Use cases vary hugely – a factory is different to a campus, which is different to a stadium, and different to a city. Some require mobility between public and private networks, which is where operators start coming in. And there is a big category of use cases where operators have a role. But there are also use cases, like these 200-plus, where deployments are simpler, and you don’t need [traditionally-licensed] spectrum. So, yes, they have a role, just not every time.”
And clearly they all have different approaches and capabilities, but, generally, is the message that mobile operators are best placed to supply private or integrate public/private cellular solutions to cities and campuses where applications run across public, private, and shared infrastructure?
“Yes, it is primarily that way. Not strictly, but primarily, yes. City-wide or campus-wide use cases, or stadiums, are primary use cases for telcos to go after because they require private/public network mobility, and larger-scale macro deployments and radio planning. They have a play in smaller closed loop enterprises, as well, just less of a play. But some of these are already their customers – for broadband connectivity, say. But the compelling cases for them, where nobody else can do it, are in large campuses and cities, and things like that.”
Just explain the various AWS offers in the private wireless market – the difference between the AWS Private 5G offer in the US and the new Integrated Private Wireless proposition with operators, launched last month at MWC.
“So there are two pure-AWS flavours. One is the 200-plus deployments we discussed before – which are networks built on existing AWS services. So the enterprise is using Outposts or Snow Family, and an ISV or an SI gets a radio from JMA Wireless and integrates the network. Which is where we’ve done all our learning over the last three years. And based on that, we created the other offer, AWS Private 5G, which allows customers to specify their requirements and receive hardware, which can be configured on-prem and managed in the cloud.
“And neither of these requires an operator; they use shared CBRS spectrum, limited to the US currently. The other offer, Integrated Private Wireless, addresses global markets via partnerships with the telcos. It uses the same technology as the AWS Private 5G service, but it is offered by the telcos. It enables enterprises to pick from their offerings – which have been vetted by our solution architects as architecturally sound and operable. The telcos do the pricing and the billing, and manage the service, but it runs as a managed service on AWS.”
So, the Integrated Private Wireless proposition does not replace the Private 5G model outside of the US; it is an alternative route for enterprises which prefer to go with a carrier. Is that correct? And is it right, then, to think the Private 5G model in the US will be replicated in ‘vertical’ spectrum in global markets – in time, and as the spectrum becomes available?
“Yes. The space is still opening up. There is a bill on the floor in India, for example, and countries in Europe are at various stages. That CBRS-like model – which allows enterprises to own spectrum in certain geographies – opens up the whole field for private networks, so an operator does not need to be involved. But, like I said, it is a vast market of deployment types and use cases, and we have complimentary offerings.”
And on the managed private 5G service, in the US today, and as it launches in other markets: are your vendor partners – from memory: JMA Wireless for RAN, Athonet for core, Federated Wireless for spectrum access – completely settled, or are you signing new partners, or working with other partners?
“We are not announcing specific partners in that space. We are working with all of them [actually], and open to using any we see a need for. But the whole idea with managed services is you really don’t care about the underlying radio or packet core. Which is one of the reasons we do not break it down in terms of which core is running [in the system]. But our strategy is to work with all partners, and we will work with all of them based on customer demand or feature requirements.”
Indeed, the AWS strategy on enterprise networks appears, on the face of it, to be more collaborative. (Microsoft, notably, has bought-in cellular capabilities.) Talk about your partnership approach, and say whether AWS intends at some point to own a 5G business. Or is AWS only interested in what compute loads above and below the network layer?
“Our strategy is to work backwards from the customer – to offer choice about the best options for them. We work with all the leading ISVs to understand their applications at an engineering level – so we can enhance the services on top, to run with the scale, performance, and reliability that telcos require. We’ve brought telco capabilities to the cloud and cloud capabilities to telcos. This industry has succeeded and thrived because of its ecosystem of partners, standards, and alliances. We want to support that and bring our assets to the table.”
On ISVs, as well; there are a whole bunch of new vendors in the private 5G space that have grown up with AWS, and a whole bunch of big legacy vendors that grew up before AWS. Are there differences between them, in terms of their adoption of cloud technologies?
“There is definitely variation in terms of maturity. But it is a journey – you don’t just switch, suddenly, to being ‘cloud-native’; it happens by degrees. The good thing 3GPP did with 5G was to make it microservices-based. The telco industry took the first step to do away with all the old niche-telco interfaces and put these packet core applications on a cloud-native path. And all of the ISVs have containerized applications; some started a little earlier, going from virtual machines to containers. But everyone is going the same way.
“At MWC, we showed both Ericsson’s core running on AWS and Nokia’s RAN stack running on AWS. All of them are working very actively; all of them have cloud native software on AWS. In the end, we are driven by customers, and some [operators] are pushing hard to determine which ISV to go with to go to the next level. That is where you see little differences. We still need to work at an engineering level to get this built out. But a couple of years down the line, you’ll see push-button offerings [to dial-up networks on AWS].”
Stepping back, a moment; explain how AWS sees this Industry 4.0 migration towards the edge. Is the contention that the centralised cloud remains where the action is, and that the edge is really just for parochial breakouts for niche use cases?
“A large percentage of workloads will be in the public cloud in the large regions. That is clear. The edge came about, partly, because we wanted to serve telcos – and networks are essentially about connecting endpoints to large data centres. The edge is fundamental to a cloud native network. Some workloads have to run at the edge, right? You have to stitch them together to build connectivity. So the edge is key for that, and that was the genesis – to be able to start to build and host networks.
“But the edge is important for some low-bandwidth IoT apps and low-latency cases. Equally, some parts of workload can run in the region. Like with gaming: rendering is latency sensitive, and has to run at the edge, but game stats and sessions can be done cheaply in S3 buckets in the cloud. Take machine learning: the inference happens in real time at the edge, but the large training models can run in the region. There are some unique cases coming up, and telco is one, but lots of the work will always run in the region.”
What is the biggest challenge in this market – for digital transformation of the industrial sector?
“The mindset, really. The technology works. We have the proof points. There are always early adopters, and there are always people who don’t want to take ‘the risk’ until it is proven – even though we don’t see any risk; just perceived risk. But that has been the cloud disruption journey from the start, and we have come a long way. Two years ago, people were still asking, ‘Why cloud?’ Now they just want to know how to do it, to start their journey. So it is just about the mindset. We have seen massive acceleration in the last two years.”
