Telco edge manifesto

STL Partners has been working with telecoms operators on developing their edge computing strategies since 2015. The following web page summarises the key learnings we have taken from those engagements and acts as a manifesto for how we feel the industry should be addressing the edge computing opportunity.

Weathering the trough of disillusionment: Network edge will take off

The explosion of network edge hasn’t happened… yet. Our estimations show that fewer than 250 network edge sites had been built by the end of 2022, and most of these are concentrated in the US and China.

While these deployments do represent progress, we believe that edge computing is currently firmly in the “trough of disillusionment. To progress beyond this, we believe telcos must focus on four key areas.

Weathering the trough of disillusionment: Network edge will take off

1. Telcos should be their own customers of distributed compute first. If a telecoms operator is struggling to identify use cases for edge computing, then they should look no further than their own network workloads. As networks become increasingly cloudified these workloads will require distributed compute. Network workloads, particularly those required to operate the RAN, are truly demanding – telcos that can handle these latency-sensitive, mission-critical workloads should have a compelling case for then telling enterprise customers they can do the same for them. This holds particularly true for industries whose own OT (operational technology) workloads also have stringent performance requirements, such as those in the industrial sector.

2. Telcos should finalise critical decisions that are slowing 5G SA rollout. Various factors have slowed the deployment of 5G SA networks including the regulatory uncertainty surrounding the use of Huawei equipment and the delaying of spectrum auctions thanks to the COVID pandemic. This impacted markets across the world including India, Malaysia, Spain, among others. However, another source of delays has stemmed from the apparent lack of clarity from telcos as to the best strategy for rolling out 5G networks. Telcos are still debating key architectural questions such as pooling CUs in CU hotels versus deploying CUs alongside DUs and RUs and the relative benefits of O-RAN versus vRAN. While edge computing can be deployed with 4G networks (and is also being deployed in fixed networks), for most mobile operators, the two are interlinked. This is because it is in 5G standards that we see control and user plane separation, enabling the distribution of user plane functions doing packet processing to the edge and enabling local breakout of traffic.

3. Telcos should stop seeing edge computing as a minor innovation programme and start to embrace distributed computing (for their networks and for other customer workloads) as operators’ defining identity. Almost all telcos focused their resources in the years 2020-2022 on their networks and core connectivity services, as they sought to handle the unparallel levels of network traffic during the pandemic. Logistical challenges prevented site upgrades and innovation programmes, such as those focused on edge computing, were deprioritised (more than 30% of respondents to an STL survey indicated a delay in their edge programme during the pandemic). For telcos to stand a chance of being market leaders in edge, they will need to change their investment priorities, putting the orchestration and management of distributed compute as the single most important priority.

4. Take actions on key developments in the market suggesting customer demand for edge computing is maturing. These developments include but are not limited to:

• Mature and commercialised anchor use cases that can be exploited today. The clearest one of these is content delivery and optimisation. STL estimates that edge content delivery will be the most significant revenue generating use case and will reach over US$58 billion in 2028. Proof points for this use case include announcements by telecoms operators that are hosting new, disaggregated CDN platforms in edge data centres. For example, Qwilt (a US-based edge content delivery platform) has partnered with multiple telcos in the last two years, including Vodafone, BT, Telefonica, stc and Bharti Airtel.

• Major technology companies are forming strategies and making investments that will require edge infrastructure to support them as they scale. Examples of this are Meta’s metaverse announcements and Apple’s Vision Pro announcement, both of which signify the growth of AR/VR and immersive applications. Limitations of these products indicate a need for edge computing, including the need to have a battery attached to the headset to prevent it from being too heavy and the considerable price tag for the device.

Number of network edge data centres globally, 2022

Source: STL Partners Network edge capacity forecast

By following these steps, telcos can effectively tap into the $445bn edge market

Total addressable revenues from edge computing, 2020-2030

Source: STL Partners Edge Computing Market Sizing forecast: Second release

If these four recommendations are followed the market opportunity for edge computing will be significant, and telcos will be in a strong position to address it. STL Partners forecast US$445 billion in total addressable revenues globally from edge computing by 2030.

It is important that telcos make informed and proactive investments in distributed compute infrastructure to ensure that they can scale fast enough to support software growth. Software applications can scale quickly due to the maturity of the cloud infrastructure and the tools underpinning them. Building the infrastructure, on the other hand, takes time and requires overcoming various logistical hurdles including the acquisition of land and hardware to start physically building it out. The speed at which software can scale cannot be better illustrated than by the explosion of use of OpenAI’s ChatGPT which, according to Reuters, reached 100 million monthly users just two months after its initial release and became the fastest-growing consumer application ever. Telcos need to act now to ensure they are in a position to serve the growing edge market.

If not edge, then what? A comparatively low-risk way to move beyond connectivity

Many operators have experience running cloud-native workloads in a distributed manner (there are more than 75,000 live cloud native distributed compute sites running telco workloads today). Telcos have developed know-how and expertise that they can build on to unlock synergies and develop network edge propositions ready for commercial use by third parties:

• Operators are in the process of building the processes, skills and teams to run distributed workloads, troubleshoot, add new hardware, and ensure operational control and management of the infrastructure.

• Automation capabilities as well as vendor maturation, pre-integration and new tools will reduce integration, deployment and service iteration timelines going forward.

• Cloudification is evolving traditional network sites (e.g. central offices) into data centre environments and freeing up physical space in these sites to make it possible to add infrastructure (servers) to host third-party workloads.

Telcos are already building a distributed edge cloud to support their network workloads. They can build on these capabilities with limited incremental investments to deliver commercial network edge services to their customers.

Incremental investments for network edge are comparatively low

Source: STL Partners analysis

Telcos can benefit from existing infrastructure while keeping workloads separate

Telcos are hesitant to use fully converged infrastructure platforms

Source: STL Partners survey, n=191, 2022

It is not true that telcos can only benefit from complementary investments if they use the same hardware infrastructure (and management systems) for network and edge workloads. From a survey of 191 stakeholders across the telco ecosystem we know that this is not a well-favoured deployment model – over half of respondents stated that the two domains remain separated.

Despite a lack of physical convergence, we believe that synergies will still be achieved through common tools, processes, and learnings which will unlock considerable efficiencies for operators as they build out network edge.

To scale and monetise edge, interoperability must be guaranteed

Interoperability is key to making the network edge attractive – we have highlighted how both technical and commercial challenges need to be overcome to enable a scalable network edge. Technically, this means allowing workloads to easily move between edge clouds, regardless of their network service provider and who owns the edge nodes. Commercially, it must be easy for edge customers to use edges, without interacting with each and every edge service provider.

Interoperability is critical to creating a consistent global experience

Source: STL Partners

Three scenarios for how network edge will be built and how interoperability will be achieved

Three potential scenarios for building the network edge

Source: STL Partners

There are three credible scenarios for how network edge will be built out and how interoperability will be achieved. We would argue that telcos put themselves in a stronger position to own their own destiny and be in the driving seat for network edge monetisation if they focus on making a success of scenarios two and/or three. This means investing time and resource in making federated edge possible, not just technically but commercially as well.

A significant cultural change is necessary to help telcos achieve the goal of a truly interoperable network edge. Telcos must reconsider their approach working with partners to build ecosystems around them and commit to openness via an API-first strategy.

Three scenarios for how network edge will be built and how interoperability will be achieved

Scenario 1: Hyperscalers invest in edge platforms as an extension to their cloud offerings and create consistent global experiences for developers through their tools (portals, CLI, APIs etc.). In this scenario, telcos provide the physical sites for edge workloads and enable local breakout of network traffic to a hyperscaler who provides the full stack of hardware and software – essentially the telco plays the role of a colocation provider. This is essentially the model of AWS Wavelength deployments (though in most instances the telco is also acting as a sales channel as well). In this case, interoperability at the edge works in much the same way as it works today in the cloud. Hyperscalers would need to invest in partnerships with most telcos to provide a consistent global experience for enterprises or developers (with the need for complex interoperability between edge nodes solved by the hyperscalers’ internal resource management and orchestration tools). In some ways, this scenario seems likely – most of the network edge deployments globally today (at least in mobile operator networks) have been done in conjunction with hyperscalers. However, the caveat is that this would require a mighty investment from the hyperscalers. While they seem driven to invest resources in working with selected tier one operators today, it is yet to be seen if they will be motivated to strike deals with the long tail of smaller operators as well.

Scenario 2: Telcos provide consistent experiences through federated edge programmes that allow developers to seamlessly “roam” onto other telcos’ edge clouds. In this scenario, telecoms operators provide the physical sites, as well as the hardware equipment (racks and servers) required to run the edge workloads and either partner with a third-party edge platform provider to install their software on those servers or the telco builds its own platform to provide the edge service. To allow developers to access edge resources globally, all via one operators’ portal, telcos would need to enable the customer to “roam” onto another operators’ edge cloud and pay them for it in a similar mechanism to how they do intercarrier roaming settlements today. This is often referred to as federated edge and trials for this have already been completed by telcos, such as Singtel and Telefonica. All POCs so far have always focused on the technical rather than the commercial mechanisms and have been between non-competing telcos. It is yet to be seen whether competitive dynamics will stop the likes of Verizon being comfortable allowing customers to seamlessly leverage AT&T resources.

There are industry organisations and initiatives that are trying to address the commercial and technical challenges associated with network edge federation. Some of the notable initiatives include the following:

* GSMA’s Operator Platform Group
* eGATE initiative involving Deutsche Telekom, Orange, Telefónica, and Telecom Italia
* 5GFF’s APIs to ‘enable interoperability and seamless integration’
* Bridge Alliance’s Federated Edge Hub (FEH) to help interconnect member operators’ network edges

Scenario 3: A third party provides a combination of datacentre knowhow and investment to ensure that network sites are able to support third party compute workloads. Different models for this are already being explored – primarily:

– AtlasEdge: a joint venture between Liberty Global and DigitalBridge in Europe. Here, the edge hosting is provided by the neutral third party entity of AtlasEdge. The involvement of Liberty Global means that edge traffic can be locally broken out to the AtlasEdge sites. While this seems a promising model, and has been fast to create available network edge sites, many of the challenges associated with edge federation still exist and it is yet to be seen whether AtlasEdge will be able to appeal to other CSPs in a way that would turn their sites into carrier-neutral locations.

-Vapor.io’s partnership with Comcast in the US. In this instance, Comcast is still providing the edge hosting but rather than working with a hyperscaler to provide the edge node and the edge platform they are instead working with Vapor.io.

Telcos should understand the limitations of working with hyperscalers

For many, when thinking of edge computing partnerships, the series of high-profile telco and hyperscalers partnership will come first to mind. It is not that these partnerships are not valuable nor that telcos should not consider collaborating with a hyperscaler, particularly when looking to seed the market and learn from the hyperscalers cloud sales expertise.

However, we do recommend that telcos put less emphasis on creating a hyperscaler strategy in and of itself and instead first identify their own business priorities and then allow these to inform whether a particular hyperscaler is in position to support this path.

Humble but ambitious: A new approach to partnership models

Telcos should understand the limitations of working with hyperscalers

Telcos should think about their own value-add with hyperscale partnerships. For example, applications that are best-suited to run specifically at the network edge are usually “hardcore” – they will often be mission critical, require ultra-low latency, and are highly distributed in nature. Examples of this would include M2M communications and a variety of “autonomous” use cases including cars, drones and more. Cloud providers are a long way from being experienced in supporting these kinds of workloads and so may not make the strongest partners in these domains. In comparison, these workloads should be telecoms operators’ bread and butter (very few workloads are as demanding as RAN).

Secondly, we recommend rethinking the question itself: instead of asking “how can the hyperscalers support my network edge strategy”, consider asking “what are the specific strengths and weaknesses of individual hyperscalers and how does that impact how work with them?”.

It is counterproductive to think of the cloud providers as a single entity. Bundling all hyperscalers and treating them equally creates the false impression that they operate in the same way, have the same strategies, offerings and attributes. We recommend evaluating each cloud provider’s unique abilities independently. Telecoms operators should ensure before entering a partnership conversation that they have already considered the following key questions:

• How will working with this player help us advance our network edge strategy?

• What is the expected commercial model for the partnership? If revenue share, how should the exact percentages be determined? What is the minimum take-up of the infrastructure to make revenue sharing an attractive approach?

• What are the parameters we expect from the partnership? What are the go and no go areas?

• How will we know the partnership has been successful? What are the KPIs of success?

Partner widely with solution partners and nurture them by minimising typical telco bureaucracy

Telcos are generally well-respected players in their local markets, with scaled access to many enterprises. This means that they have unique sales channels that many are eager to access. However, it is equally well-known that telcos are corporate organisations with lengthy decision-making processes, complex legal requirements and bureaucratic systems. This environment is very different to the lean, agile environment of software companies and the disconnect causes problems.

Telcos should therefore shield partners who are building solutions on top of telcos’ network edge infrastructure from legal and bureaucratic hurdles. They should also ensure that internal commitments have been made early so that, if successful, the partnership can gain more investment quickly.

Leading operators have already taken steps to engage more closely with crucial solutions partners and provide them with what they need to deliver value via edge infrastructure. One example of this is Verizon’s 5G Edge Developer Portal, built in collaboration with AWS. Verizon’s portal has clearly been built with solution providers in mind and demonstrates several areas of best practice, including:

• Clear explanations of the benefits of edge computing avoiding overly technical network jargon.

• Offers routes for companies to access freemium edge services in order to test the impact of edge on their application ahead of any monetary investment.

• Focus on making edge accessible in an automated fashion (e.g. via APIs) which means solution providers do not have to spend significant amounts of time interacting with the telco to stand up an edge instance.

Telcos should also consider ways to tap into ecosystems built up by parties beyond just the hyperscalers, including hardware and chipset players like Dell, Nvidia and Intel.

Multiple routes to monetisation should be pursued

We have identified five models that telcos should pursue for monetising the network edge.

The right monetisation strategy will depend on the operator, its position in the market, its strengths and weaknesses, and its investment model. However, the value operators can expect to capture will increase significantly moving from left to right, from simply providing the physical space and connectivity all the way to applications and solutions. We forecast end-software application revenues to account for just over 50% of the total addressable value stemming from edge computing by 2025

Telcos should have ambition beyond pure infrastructure monetisation

The traditional telco approach has been to start by trying to capture the infrastructure and connectivity revenues with the potential to move up the stack if things are working well. Our recommendation is to rethink this paradigm. Telcos that first consider which edge-enabled applications could deliver most value to their customers and work with partners in order to get close to the inner workings of these applications set themselves up for success. This is a demonstrated strategy by leading operators in telco edge including:

• SK Telecom and AWS’s jointly developed computer vision application

• Verizon’s smart stadium solution and international business unit

• Vodafone’s partnership with solution providers such as Dedrone

The role of a multi-edge and cloud orchestrator could also be an increasingly interesting role for operators as the market matures. It complements well with the multi-vendor orchestration capabilities that telcos will need to develop to handle best-of-breed network environments for a start. And while we are not yet in a position where enterprises are looking to deploy different edge workloads on different cloud platforms, multi-cloud optionality is increasingly central in cloud strategies. Assuming that edge computing follows a similar pattern at maturity there could be an interesting role for telcos to play as a neutral third party able to orchestrate workloads in a vendor agnostic manner.

Five models for network edge monetisation

Source: STL Partners

What should telcos do today?

Invest in anchor use cases beyond network workloads to seed market adoption

Form solution-specific cross-domain teams

Take practical learnings from participating in on-premise edge deployments too

Build industry leading expertise in running and managing distributed cloud

What should telcos do today?

Invest in anchor use cases beyond network workloads to seed market adoption

It is hard to gain traction for a horizontal edge platform while the market is still in its infancy. Once network edge establishes itself as the technology of choice for powering certain enterprise and consumer applications, it will be possible to sell generic enabling platforms or capabilities. Until that point, our recommendation for telcos is to pick a vertical opportunity (e.g., content delivery or video analytics) and build relationships with customers and solution providers in that domain. In other words, telcos need to have a clear proposition for solution providers and enterprises proving a tangible business case. This is much harder to do with a horizontal platform than it is with a vertical-specific solution.

Form solution-specific cross-domain teams

Currently, many telcos have fragmented teams that focus on specific technologies rather than solutions. For example, operators might have separate edge computing, private networking, cloud/security and IoT teams – all working in siloes. The risk is that telcos will come up with solutions that don’t incorporate the full value that they can provide for customers. For example, solutions might incorporate edge computing but not private networks and vice versa. Instead, telcos should create cross-domain teams focused on a specific area or vertical and encourage them to use a combination of whatever underlying technology will best solve their customers’ needs.

Take practical learnings from participating in on-premise edge deployments too

While on-premise edge has a smaller long-term revenue opportunity, it is a more mature technical and commercial option today and gives solution providers a chance to experiment and establish themselves within the edge computing space. B2B verticals offer the biggest revenue opportunity for on-premise edge, in particular manufacturing which accounts for 49% of the on-premise addressable revenue in 2022 and continues to grow to 60% in 2030.

While telcos will face stiff competition from specialised systems integrators, a stronger right to play can be established through bundling on-premise edge with a private network solution. This connectivity-led play can be a helpful route to bridging the gap between networking expertise and expertise in cloud and edge services. It is, however; worth noting that to do this successfully telcos must be highly proactive in identifying applications that will most benefit from the unique combination of on-premise edge and private networks and that have the biggest potential to create value for customers. As this technology carries a relatively high price tag, it is important telcos are able to make a solid business case with tangible business outcomes for an enterprise to justify the investment.

Build industry leading expertise in running and managing distributed cloud

The core thesis of this manifesto rests on the idea that telcos should redefine their purpose around becoming experts in running and orchestrating distributed cloud platforms, both for their own workloads and for others. To do this, telcos will need to recruit and retain cloud and infrastructure experts and work with partners who can help them to further innovate and upskill. Telcos will need to prove themselves as industry leaders to effectively pursue the two-pronged strategy of first solving the pressing internal need and then commercialising the solution obsessively.

What next: Get in touch to discuss your network edge strategy

STL has built a centre of excellence around edge computing and has been advising businesses for over six years

We have supported clients across 90+ projects to develop their commercial strategy for edge computing. We have worked with more than 15 major telecoms operators including advising on a hyperscaler partnership negotiation. Our team of experts includes leading industry analysts and consultants, IoT practitioners as well as cloud specialists.

Our industry-leading edge research service offers subscribers the following:

Database of 50+ use cases for MEC, private networks edge and on-premises edge
• Interactive model forecasting the edge computing market
• Interactive model forecasting the buildout of network edge sites
• Ecosystem tool mapping 350+ vendors across the edge value chain
• In-depth analysis of opportunities and threats via our research reports