Major telcos are enjoying some success partnering with hyperscalers in the burgeoning Mobile Edge Compute (MEC) field, even if they risk being treated as junior partners just providing the underlying connectivity. There is also a sense of urgency among telcos to recoup 5G spectrum and infrastructure investments by tapping emerging demand for applications and services that require low latency. At least with backing from the hyperscalers they are more likely to convince enterprises that they need such applications.
There has been a flurry of such recent partnerships between MNOs and hyperscalers, the latest being BT’s declared “multi-million-pound investment” in MEC which will embed Amazon Web Services (AWS) compute and storage services in its network infrastructure. The UK operator, whose mobile network is branded as EE, will deploy the AWS MEC product, called Wavelength, in its network. This choice was made at least partly because Wavelength has been gaining momentum in a battle with Microsoft Azure in particular for combining local MEC capability for applications that require latencies down in the single digits of milliseconds, with access to more remote cloud based services for higher volume analytical, archiving or other processes where more delay can be tolerated.
BT seems particularly intent on targeting what it calls “high-bandwidth IoT use cases in the field”, such as autonomous vehicles, police cameras, media production, industrial robots, and community healthcare. With these in mind another reason for the choice was that BT wanted to line its 4G/LTE and 5G infrastructure heavily with AWS hardware so that enterprise processes connect directly with services hosted in the cloud directly without the data having to leave its network. This will help convince potential customers that this combination will yield the highest performance and especially lowest latency possible, while avoiding the security risk involved crossing into domains controlled by other parties, as Alex Tempest, managing director at BT Wholesale, contended.
“Launching the AWS Wavelength service for our business and wholesale customers is a hugely important step on our journey,” said Tempest. “It is set to unlock use cases like IoT cameras to help first responders keep communities safe: a real-life example of using tech to connect for good. By building cloud edge services into our network, we can accelerate innovation across industries, and bring fast, secure data processing closer to where our customers need it most.”
Wavelength is a fitting name for the product since it is at the leading edge of Amazon’s strategy to make telcos dance to its tune, rather than Microsoft’s as the closest equivalent, or Google’s as a relative latecomer to the party. There are also other bystanders in the hyperscale cloud business, such as IBM, and also China’s Alibaba.
There are various ways of counting but there is a broad consensus that AWS is number one with almost exactly one third of the global cloud services market, followed by Microsoft Azure on 23%, Google 11%, Alibaba 5% and IBM 4%, with the remaining 24% or so divided among sundry smaller players. As the field matures those smaller players, as well as IBM and Alibaba, appear to be squeezed slightly, while AWS is consolidating its position at the top with Google gaining slightly and Microsoft up about 2% over the last year. It looks like AWS may also be edging up on the back of wins such as the BT contract.
AWS Wavelength combines the bandwidth and ultra-low latency of 5G with its own services, especially compute and storage, to yield this composite infrastructure for hosting different use cases. It is based on the concept of Wavelength Zones rooted in the data centers of telcos, like great cuckoo eggs as one cynic suggested.
The idea is that such deployment of its resources means end users, including those attached to private enterprise networks, can reach application servers without ever leaving the telco’s network. This supposedly keeps latency down and also allows the telco to control security with help from AWS as appropriate and argue that it can match the levels of a private network.
Just before the BT announcement, AWS unveiled T-Mobile US in February 2023 as one of the first operators to use its cloud compute and edge products in a 5G-Advanced portfolio, alongside a private wireless product AWS had just launched.
This allowed T Mo to claim it was the first US operator to work with AWS on delivering customizable 5G edge compute offerings. However, Callie Field, president of T-Mobile Business Group, admitted it had to raise its cloud game to overcome slower than hoped uptake of edge compute and private 5G to-date. Indeed, a backdrop to those hyperscale market share percentages is a slight global shrinkage in the market.
Alongside the new Integrated Private Wireless product, AWS has announced general availability of its Telco Network Builder designed to help operators run and scale their networks on its cloud, again aiming to lure them deeper into its web. This allows operators to use their own networking language to describe various elements of their architectures in a template which they can then upload to as a prelude to moving network functions into the AWS cloud.
Wavelength was launched late in 2019 and Verizon was the first US and global operator to deploy it, starting from the end of that year. Then in June 2021, Vodafone was the first so deploy it in Europe, starting with the UK and then setting up Wavelength Zones in three German cities.
Microsoft has been heading in a similar direction, although with AT&T in the USA it was almost the other way round. AT&T in June 2021 announced it was moving its whole 5G network into the Microsoft cloud, rather than instead deploying the latter’s inartistic inside the network, as the likes of Verizon, Vodafone and BT are doing with AWS Wavelength.
AT&T wanted to outsource management of its network traffic to Microsoft Azure, representing an even more full blown transfer of control. The process began with AT&T’s 5G core, comprising software connecting users and IoT devices via the RAN into the internet and other services. AT&T was under the cosh from its rivals and this move was motivated by the quest for operational cost savings and a determination to focus on delivery of new applications and target specific verticals, such as the connected car.
As part of the deal, Microsoft gained access to AT&T’s intellectual property and technical expertise to expand its telecom flagship offering in the hope of eating into Amazon’s lead. This included AT&T’s carrier-grade Network Cloud platform technology, so this clearly represents a transfer of power from a major telco to a hyperscaler.
Meanwhile, Google has been trying to slip through by offering alternatives to such full blown interlocking of the cloud and mobile network. This is playing to the sensitivities of operators about handing over too much power to the hyperscalers, and probably helped win the contract at France’s Orange for example. In France there is particular concern over US technology giants generally, and yet the operators are under pressure to engage with the hyperscalers to gain the economies of scale and access to critical technologies in automation and analytics.
Orange has continually tried to frame the relationship as being one of cooperation over edge cloud technology development, rather than any dependency. The partnership was announced in July 2020 covering data analytics, with Google designated as provider of expertise in cloud technologies, analytics and AI upon which Orange would build its own edge platform incorporating machine learning based control of traffic and routing.
This was followed by the opening of a joint lab in Chatillon near Paris for app developers to test their products over edge computing services. This has an edge node based on Google’s technology coupled to Orange’s 5G infrastructure, resembling a microcosm of the Amazon and Microsoft approach, but only at the test level. It is not surprising with this reticence that Orange has lagged behind over edge deployment, raising the suggestion it is getting the worst of both worlds. It is dependent on Google and yet not getting the full benefits of its technology.
Indeed, we heard the likes of Yves Bellego, Orange director of network strategy, worrying that “some years ago, everyone was saying we would have vendor lock-in with Ericsson, Huawei and Nokia and no one mentioned Oracle and Cisco and now the light is on hyperscalers. In fact, that risk is something we have always been very concerned about.”
Telcos elsewhere have taken a fuller dose of hyper scaling medicine and it is certainly true they risk being consumed by it, but at least they are better placed to target edge opportunities at present.