E-Book

FutureNet Asia 2023 Report – Network Automation: 6% EBITDA improvement possible, but wider buy-in needed

Andrew Collinson, Principal & Founder, Connective Insight, chaired FutureNet Asia, 18-19th October 2023 in Singapore. These are his key findings and recommendations. Connect with Andrew on LinkedIn.

Network automation in telecoms is starting to become a reality, although only a few CSPs are yet realising benefits. To accelerate progress, building solid business rationale and working out how to get organisations to successfully adopt the technologies are bigger current challenges than the technologies themselves for many, as complex as those technologies are. The report contains high-level recommendations on how to address this.

 

Executive summary

 

Key findings and recommendations from FutureNet Asia, 18-19th October 2023 in Singapore.

 

Key findings: proof of value is starting to emerge

 

  • Network automation was claimed to offer up to 30% TCO network savings or around 6% EBITDA improvements, with the potential impact of AI & Automation overall being estimated between 5-25% EBITDA. These are promising indications, although only about a third of operators are currently embarked on network automation programmes.
  • While the realisation of most benefits is estimated to be about 3 years away on average, about 1 in 5 of those we surveyed report benefits already.
  • Telcos expressed increased confidence about moving to the ‘techco’ model and getting the fundamentals of promised operational improvements and cost efficiencies in place.
  • While new business models offer hope of growth, the details appear somewhat vague to many, leading to lower understanding of requirements and less confidence in their delivery.
  • The top barriers cited to progress to network automation overall were ‘people issues’ (43%) and ‘business rationale’ (26%) above all technological concerns.

Overall, telecoms network automation is still relatively early on the road to being a mature and fully implemented strategy and there is a wide spread of maturity in players across the market.

 

To accelerate progress, harness automation to the business context

 

To accelerate progress beyond ongoing technological adoption and trials:

  1. CTOs/CIOs and proponents of network automation should develop and share a business value framework covering operational improvements, cost efficiencies and new business models, putting them in context and quantifying their potential effect to get even clearer buy-in and traction in their organisations and ecosystems.
  2. Automation proponents and practitioners should keep sharing case studies of success (and failures) to accelerate collective learning. It is a complex art, and the industry stands to benefit from collective improvement as much as individual players can muster competitive advantage.
  3. In particular, the industry should carefully study all approaches that can be used to accelerate the development of successful developer communities and implement those with the greatest chance of success. This is a mission-critical development area to enable dynamic innovation and adaptation.
  4. In the field of new business models, additional development is needed to clarify the models and their requirements they place on technology. This requires the extension of discussions and understanding to include a wider commercial and strategic audience within the CSP community.

 

Contents

Executive summary. 1

Key findings: proof of value is starting to emerge. 1

To accelerate progress, harness automation to the business context 1

Table of figures. 2

Introduction. 3

FutureNet Asia, Singapore 2023. 3

Methodology. 3

Network automation offers increasingly promising benefits, but they are 3 years out on average. 4

About a third of operators have started network automation. 4

Operational and cost benefits: up to 30% network Opex; 5-25% EBITDA total from AI & automation. 4

Timing: 3 years out on average. 6

There is growing confidence in new operating models and technologies. 7

There is higher confidence in network automation’s impact on operations and cost efficiencies than new business models. 7

Results in operational improvements are expected first 8

New business models offer hope of growth, but the details are still vague. 9

Persuading the whole business to embrace network automation is currently the hardest challenge. 10

Conclusions: even ‘no brainers’ need to be thought through to be invested in. 11

Recommendations: get more of your business on-board. 11

 

Table of figures

Figure 1: 52% of survey respondents were from CSPs, and the geographical focus was primarily Asian or Global 3

Figure 2: About three-quarters of responding event participants said their organisations were already embarked on automation. 4

Figure 3: Ericsson’s estimates of NOC and Field costs for network opex. 5

Figure 4: Online survey respondents estimated the benefits of network automation to be ‘definite’ and about 3 years out on average. 6

Figure 5: 89% of respondents were highly confident that network automation is vital to operational improvements. 7

Figure 6: 17% of respondents said that network automation had operational improvements now or within a year. 8

Figure 7: There was very little difference in expectations of the impact of network automation across a selection of new business models. 9

Figure 8: 72% of audience who voted said ‘people issues’ or ‘business issues’ were the hardest challenge vs 28% for ‘data & process’ and ‘technology issues’ 10

 

Introduction

 

FutureNet Asia, Singapore 2023

FutureNet Asia, a conference specialising in Network Automation and AI, was held at the Marina Bay Sands Hotel, on the 18-19th October 2023 in Singapore. It gathered nearly 400 participants, including leading C-level participants from telecoms operators, their technology partners, and analysts, primarily from the Technology function. The agenda and speaker listings can be found here.

I was kindly asked by Giles Cummings, MD FutureNet, to chair this event. This report summarises my observations, findings and recommendations.

 

Methodology

 

As a telecoms strategy analyst, I am primarily focused on the value of the application of technologies and other strategic assets towards delivering business ends. Hence that is the narrative focus of this report rather than the technical and technological angles.

For the report, I draw on my own notes and observations, checked where possible with speakers and panellists. My apologies to anyone who’s contribution I missed – there are bound to be many as it was a very rich and dynamic event.

I also draw on two further research resources:

  • An anonymous online survey sent to all participants and completed by 132 of them in the week prior to the event (respondent profile below).
  • Ad-hoc ‘slido’ surveys that we ran during the event by showing QR codes to link to votes, with varying respondent numbers.

I look at the results of these surveys as good general indications of the collective view of the conference attendees who represent a rich mix of well-informed industry professionals in the technology function, primarily operating in Asian markets. They should not be regarded as forecasts or treated as ‘absolute’ findings or assumptions in any plan.

Figure 1: 52% of survey respondents were from CSPs, and the geographical focus was primarily Asian or Global

 

NB, We asked respondents to think about the operator they work for or know best when answering questions such as ‘Operator scope’.

 

Network automation offers increasingly promising benefits, but they are 3 years out on average

 

About a third of operators have started network automation

Huawei’s Stephen Zheng[1] shared statistics that around a third of operators had deployed network automation: 34% in Europe, 40% in Asia Pacific, 30% in Latin America, 25% in Africa, and 31% in the Middle East.

In terms of event participants, engagement in automation was twice this level, suggesting that participants represented a highly engaged audience as might be expected for FutureNet. However, the sample size voting on this poll was relatively low, so it is support for the reasonable assumption that the audience at FutureNet are advanced, rather than evidence that the market is moving faster to higher penetration of automation than Huawei’s research shows.

Figure 2: About three-quarters of responding event participants said their organisations were already embarked on automation

 

Operational and cost benefits: up to 30% network Opex; 5-25% EBITDA total from AI & automation

 

Two speakers presented estimates of the value of network automation and AI to telcos:

  • Ericsson’s Sam Keys-Toyer[2], said that Ericsson estimates that is has achieved up to 30% TCO network savings in its own network operations over a 2-3 year period. This included operational improvements of:
    • A 12% reduction in work orders, 24% reduction in truck rolls, and 5x improvement in compute efficiency
    • A reduction of 6 million human hours vs an increase of 5 million comparable machine work hours
  • Using Sam’s estimates, we can estimate that total possible Opex savings from Network Automation and AI are about 6%, using the following assumptions (see Figure 2 below for assumptions source)
    • Network Opex = 25-35% total opex = (let’s call it 30% total opex)
    • 60-85% network costs can be impacted by AI & automation (call it 70% applicable areas)
    • Total efficiencies possible = 30%
    • Therefore, total cost-efficiencies possible – 30% total Opex x 70% applicable areas x 30% possible = 6.3% total Opex

 

Figure 3: Ericsson’s estimates of NOC and Field costs for network opex

 

  • Ahmed Saady Yaamin, Group Head of Analytics, Axiata Group, estimated based on Axiata’s experience that AI and Automation could have an overall 5-25% impact on EBITDA. Saady said:
    • Some telcos are experiencing around 5-8% currently
    • But if the transformation can be done in the right way, up to 25% impact can be achieved within 3-5 Years

In recent news, Scott Petty, CTO Vodafone announced that it saved €500m on operating costs in the last 3 years[3], so potential savings are certainly material.

Collectively, these estimates give a sense that the business value of network automation and AI is certainly significant.

 

Timing: 3 years out on average

 

In the online survey, we asked respondents how much network automation would impact different business objectives, and how long it would take for the majority of benefits to be realised.

On average, respondents felt the majority of benefits would take around 3 years for operational improvements and cost efficiencies and closer to 4 years for new business models.

Figure 4: Online survey respondents estimated the benefits of network automation to be ‘definite’ and about 3 years out on average

 

  • In contrast to the average, 17% of respondents said they believed the benefits of operational improvements were already being experienced. This suggests that there is a reasonably wide spread of maturity across operators.

 

There is growing confidence in new operating models and technologies

 

There is higher confidence in network automation’s impact on operations and cost efficiencies than new business models

It is perhaps unsurprising that there is higher confidence in network automation delivering operational improvements and cost efficiencies in the sense that this is these are ‘obvious’ impacts or goals of automation.

Figure 5 shows that there is a marked difference in the proportions saying that automation is ‘pivotal’ to operational improvements (57%) vs cost efficiencies (41%) and new business models (37%).

More of a surprise to me were the proportions saying network automation only has a ‘likely’ or ‘uncertain’ impact on any of the objectives. The total proportions expressing lower certainties were: operational improvements 11%, cost efficiencies 20% and new business models 26%.

This suggests that there is still a need to prove network automation’s business benefits, given that the audience at FutureNet Asia likely represents more of network automation’s early-adopters – those in the third of the operators that have started to implement automation vs the two-thirds that haven’t (see page 4).

Figure 5: 89% of respondents were highly confident that network automation is vital to operational improvements

 

Results in operational improvements are expected first

 

There was a similar but more pronounced difference in expectations for the timing of impacts.

  • 17% of respondents said that they were either experiencing or expected to experience the operational improvement benefits of network automation within the next year.
  • In contrast, only 7% said that for new business models, and 45% were either uncertain or believed the benefits were more than three years out.

Figure 6: 17% of respondents said that network automation had operational improvements now or within a year

 

New business models offer hope of growth, but the details are still vague

 

Looking in more detail at different types of new business models (e.g. NaaS, vertical models, etc.) respondents showed very little differentiation in their views on the impact of automation, grouping them all on average 3-4 years out and of ‘definite’ impact (Figure 7).

Figure 7: There was very little difference in expectations of the impact of network automation across a selection of new business models

 

The two most likely reasons for this grouping are that:

  • The audience is perfectly informed, and the impact is the same for all business models.
  • The requirements and impacts on new business models are less clearly articulated and understood, and therefore the audience has equal levels of uncertainty on them all.

It’s impossible to determine this from the survey data because we did not ask why in the short survey we conducted. However, based on qualitative discussions at the conference, including the panels I chaired, I would say that while both are to some extent factors, the broad understanding of the requirements and opportunities are the most important issues to tackle.

There are undoubtedly also detailed issues to resolve in the successful delivery of new business models. One of those flagged by the GSMA’s Henry Calvert[4] is the ongoing need to cultivate new and bigger developer communities to apply APIs such as the GSMA’s Open Gateway APIs.

 

Persuading the whole business to embrace network automation is currently the hardest challenge

 

There was much discussion about the many challenges of network automation, and every time it was broached, the ‘human’ issues came out as more difficult that the ‘technological’ issues – albeit that these are not trivial either (see Figure 8 for one illustration of this).

Network automation is a difficult and complicated thing to do, and it has impacts on many people’s jobs and daily lives. It can fundamentally change how the core business of a telecoms operator works.

It’s not that people don’t like change, but it is hard to do. It takes understanding and persuasion, and this is harder to achieve when the subject is complex and can feel personally ‘threatening’ – you might need to change your job, learn new skills and this can be a vulnerable situation.

It is interesting that in the responses below ‘business issues’ such as compelling vision and clear business case was chosen as the hardest challenge by 29% of respondents.

It is possible that ‘business issues’ are more of an issue for those still deciding on their strategy, and that the people issues impact those in the midst of it. However, as our survey was brief and anonymous, it is not possible to link this to any segmentation on degree of adoption of network automation. This would be a useful and interesting dynamic to explore in further research, as it would point to opportunities to streamline the overall change process.

Figure 8: 72% of audience who voted said ‘people issues’ or ‘business issues’ were the hardest challenge vs 28% for ‘data & process’ and ‘technology issues’

 

Conclusions: even ‘no brainers’ need to be thought through to be invested in

 

In theory, network automation should be a ‘no brainer’ for telecoms operators. Why wouldn’t you want to speed up, improve results and efficiency, and be able to do new things faster? In practice, the problems are (or were) that:

  1. It is a complex technical undertaking to automate networks.
  2. It wasn’t always clear exactly how much there was to be gained by automation (i.e. business returns) – although this is now changing.
  3. Automation is a time consuming and costly investment.
  4. It has a significant impact on people’s jobs and hence their willingness to embrace it.
  5. It requires scarce new skills that need to be acquired, trained and retained.

In theory too, network automation is a prerequisite for many new telecoms business models, such as NaaS, open APIs, and vertical and horizontal plays. In practice, progress on new business models is slower. This is because:

  • The industry overall is less clear on what precise business models are needed
  • This means that it is harder for CTO’s and CIO’s and their teams to figure out what’s needed to deliver them.

Challenges facing overall progress in network automation are led by people issues, for example:

  • Persuading existing teams to embrace totally new ways of operating
  • A general industry shortage of the new skills required to make it work

Operators are making progress in addressing these issues, but some challenges remain particularly in markets with less developed IT workforces.

 

Recommendations: get more of your business on-board

 

  1. CTOs/CIOs and proponents of network automation should develop and share a business value framework covering operational improvements, cost efficiencies and new business models, putting them in context and quantifying their potential effect to get even clearer buy-in and traction in their organisations and ecosystems.
  2. Automation proponents and practitioners should keep sharing case studies of success and failures to accelerate collective learning. It’s a complex art, and the industry stands to benefit from collective improvement as much as individual players can muster competitive advantage.
  3. In particular, the industry should study all approaches that can be used to accelerate the development of successful developer communities and implement those with the greatest chance of success. This is a mission-critical development area to enable dynamic innovation.
  4. In the field of new business models, additional development is needed to clarify the models and their requirements they place on technology. This requires the extension of discussions and understanding to include a wider commercial and strategic audience within the CSP community.

[1] Steven Zheng, Chief Marketing Officer, GD Dept, Huawei Technologies

[2] Sam Keys-Toyer, Head of Business and Portfolio Development, Managed Services Networks, Business Area Cloud Software and Services. I’ve checked these estimates with Sam.

[3] https://www-telecomtv-com.cdn.ampproject.org/c/s/www.telecomtv.com/content/digital-platforms-services/vodafone-cto-network-automation-has-saved-us-500m-in-three-years-46723/amp/

[4] Henry Calvert, Head of Network, GSMA