How Telcos Outsourced Their Brains!

The decision by AT&T to host its network cloud, including the core network, on Microsoft Azure Cloud is the single most important milestone in the telecom industry over the past few years. It’s an experiment and a leap of faith that is set to have major ripples on the market. The core network is the essence of 5G that will differentiate among mobile network operators. Cloud providers are intent on playing a key role in the re-architecture of the mobile network to provide edge computing services. In this arrangement, AT&T appears to have ceded its nervous system and outsourced its brains. So, what are the prospects, and could they be so negative?

Background

Microsoft’s Azure Cloud will host AT&T’s network cloud deployment and development starting with the core network functions (see the June 2021 announcement here or here). In this process which is expected to last 3 years, Microsoft is acquiring AT&T’s network cloud technology and the design, development and support engineers who maintain and operate the network. They will become part of Azure for Operators service which Microsoft will offer to other network operators.

Cloud providers go after telcos

Dish Networks has outsourced the hosting of its network elements to AWS. The major difference from AT&T is that Dish is a Greenfield with no legacy network. For its part, Google announced partnerships with Vodafone and Bell Canada to host some of their IT workloads and network functions. The Google case flags the lag of telcos in processing large amounts of data for AI applications. Other cloud providers and vendors like VMWare and Rakuten are also in the game chasing the telcos with their own platforms and telecom network functions.

Interestingly, just as some telcos moved their network infrastructure onto public clouds, cloud providers have been developing network functions on their own, or making acquisitions of network function developers and vendors. Last year, Microsoft acquired Affirmed Networks (EPC/5G core) and Metaswitch (IMS). This makes Microsoft a competitor to the likes of Cisco, Ericsson, Nokia and Huawei.

Other cloud providers such as AWS and Google have their own core network development projects. Facebook does as well (or did): they recently open-sourced their Magma Core.

The big prize

The mobile network hosting business is not the end-prize for the cloud players. There are so few network operators around. The core network elements and operation and business support functions scale proportionally with a few factors. Hosting mobile network functions, in and of itself, would not bring a major uplift in revenues for the cloud players. However, it could provide a platform to sell additional services in the future to telcos and other enterprises (example of Google above).

The real opportunity is in hosting edge computing applications (see here). This is possible if cloud providers move downstream from their mega data centers to secure a foothold in the network edge. The telcos own the access networks and connectivity to the end-customers, especially the enterprises. Telco central offices are prime real estate for edge computing data centers.

Another opportunity is that for enterprise private wireless networks market. There are different paths to this market which is yet to develop. In the first path, enterprises could deploy their own private networks independent of any other entity. In the second path, enterprises could sign-up a mobile network operator to operate and manage these complex networks. But there is also a third path: the cloud providers host network functions, and edge computing applications, for the enterprises. This could be done with or without the collaboration of mobile network operators.

Telco blunders

For a telco to junk its infrastructure and move its network elements onto a cloud provider infrastructure signals a major failure. This failure is preceded by a series of failures to develop an effective cloud strategy. Telcos have struggled in their cloud migration activities. AT&T is the first telco to raise the white flag and declare that it cannot achieve the cost economics and technology agility through its own infrastructure development.

This is incredible because AT&T is not just any telco. It was one of the first telcos to recognize the importance of cloud migration. It kicked-off Domain 2.0 in 2013 with the aim to virtualize its network. The program was much advertised by AT&T who announced the completed percentage at different intervals. The deal with Microsoft Azure effectively means that programs such as ECOMP and its ONAP successor, Airship, Akraino and others did not succeed. AT&T cannot develop, scale, maintain and upgrade its cloud infrastructure on its own. They are not the only telco to fail (Telefonica Unica is just another example). The cloud providers have built a huge technology gap that telcos could not bridge.

Vendor failures

The failure in cloud technology migration is also that of vendors who struggled to deliver robust cloud-native telecom core network functions. Vendors first delivered software versions of their core network functions that runs on standalone servers. After several iterations, they gradually built their containerized cloud-native core network functions. The quality of these architectures varied as these vendors – Cisco, Ericsson, Nokia and others – struggled to upgrade the performance. This led to a very slow and arduous process for networks upgrades including the migration to 5G standalone core which is necessary to deliver on the full promise of 5G. The failure is not limited to the telecom network function vendors, but extends to private cloud providers such as HPE Helion among others.

New dynamics and threats

In outsourcing the infrastructure to cloud provides, telcos risk losing control of different aspects of their network and technology roadmap over the long term. Most certainly, Microsoft has committed to certain KPIs to ensure the performance AT&T requires. Performance would also include the ability of AT&T to get the features and services it needs in the timeframe that it wants. One could argue that the cloud providers are better able to deliver on new requirements, but this is not certain since AT&T will be in a pecking order of Microsoft priorities.

More important than performance is perhaps the loss of technical knowhow related to both the infrastructure and the network functions that ride on it. We already see a difference between operators who are highly engaged in technology development, and those that don’t. Such engagement separated the leading operators of the size of AT&T from other smaller operators who don’t have the scale to invest in R&D. On this specifically, AT&T is one of the few operators that has a separate line item for R&D on their financial statements – a value that’s been shrinking over the past few years from 1.65 billion in 2016 to $1.21 billion in 2020. The desire of AT&T to reduce R&D costs are explicitly stated in the announcement.

Yet another critical point: outsourcing cloud infrastructure raises the risks for vendor lock-in since it is difficult and very costly to move workloads among different cloud providers. Mobile network operators are bemoaning vendor lock-in in the radio access network where few vendors with low-profitability remain. I wonder how vendor-lock with a few highly-profitable cloud providers will shape up?

Hosting telecom network infrastructure with public cloud providers raises the question of the long-term ability of telcos to provide value added services to their customers – individual subscribers and more importantly enterprises. To what extent would the telcos be restricted from developing value added services and new revenue steams? The risk is very high, especially where they don’t have control over their infrastructure.

New supply chain

Threats could also develop over the long-term for the network function vendors. There’s nothing that prevents cloud players from developing similar functions – and clearly, they did. But this could allow for new vendors into this space as well that would pose a challenge to the incumbents.

Vendors are likely to face a challenge in having different sales process with the operators than what they are used to. The cloud provider will need as well to validate solutions on their cloud which requires a 3-way collaboration. Support and maintenance also require the good will of all 3 parties.

Lost opportunity?

The public core hosting model effectively yields potential edge computing service revenue to the cloud players. The significance of this loss varies since the telcos were at a disadvantage versus the cloud players to start with. However, the outsource model indicates that telcos would have little role in future edge computing services. The cloud players will offer their cloud infrastructure to enterprises as well as any additional services such as those related to AI/ML.

On the face of it, the cloud providers like Microsoft would be reaping most of the benefits from such partnerships. Microsoft would compensate AT&T for the opportunity to access the edge, but the long-term impact is hard to assess and does not look promising.

Financial markets provide some evidence of this. Take for example the tower asset outsource model that operators followed. We have evidence from the financial markets of the value unlocked by the tower companies whose valuation keeps on increasing while that of telcos remain stagnant.

From dumb to dumber

Several telcos are likely to follow AT&Ts lead, especially those without much R&D capabilities. Economists may call this comparative advantage where telcos can focus on doing what they do best. But where infrastructure determines your destiny, it’s an approach for short-term gain. In the long term, AT&T may lose control over its destiny.

In the process, the cloud providers continue to amass the expertise to develop, deploy and operate telco network functions on their clouds, the telcos turn into entities akin to MVNOs who should command MVNO-type margins!

This is ironic since telcos developed their own infrastructure equipment in the telephony era (Lucent/AT&T for instance, or Nortel/Bell Canada). It is also ironic that the cloud players are vertically integrating to gain better control of their infrastructure. Hyperscalers such as Google, Microsoft, Amazon and Facebook have built thousands of submarine cables miles crisscrossing the globe to connect their data centers. Aside from partnering with telcos, cloud players are selling them connectivity services as well. AWS developed the ARM-based Graviton processor which competes with solutions from Intel and AMD; and Google developed the TPU to target AI/ML applications in competition with Nvidia’s GPUs.

Cloud players are vertically integrating key technologies while telcos appear to be outsourcing their brains! This will ensure that telcos become not just a dumb pipe, but a dumber pipe. Time will tell if this was a good move by AT&T, but for now, the odds don’t seem to be good.