Skip to main content

Discover how telcos can unlock new revenue streams with AI and 5G

Dennis Lorenzin, Senior Vice President, Network Cognitive Services Unit at Nokia.
(Image credit: Nokia)

Warren Buffett recently claimed that if industry leaders ignore the need to digitally transform their business functions, they will “risk obsolescence.” For Communication Service Providers (CSPs), the foundation that underpins transformations is 5G. But why do CSPs need to transform their operations? Look no further than Bell Labs Consulting’s research which recently found that a minimum of $2 trillion of new revenue could be on offer for CSPs by 2028. This vast potential new revenue stream derives primarily from the sheer number of use cases that will soon be available for enterprises, through 5G technology.

Take IoT for example, McKinsey identified sectors such as manufacturing, connected and smart cities, transportation, logistics and healthcare as industries where IoT will drive significant new revenue opportunities. This is exactly where CSPs need to step up and ensure they are ready to provide the required 5G infrastructure. For example, through the provision of edge computing, analytics and the creation and hosting of new services or applications. These are crucial components that are essential for CSPs to provide in order to support enterprises undergo transformations.

However, CSPs cannot simply transition their operations to 5G overnight and instantly realise its full benefits. Transitioning to 5G brings vast complexities because CSPs have existing layers of network generations in place – from 2G to 4G. This legacy can make the final step to 5G incredibly difficult. So, what do we need to bear in mind when implementing and operating 5G networks

The revolution will be automated 

Well, the first essential step is to leave behind the mindset of monolithic service provision. New networking technology enables a plethora of 5G use cases that can be adopted by enterprises, so CSPs must create a 5G ecosystem whereby all operations are automated. With the sheer scale of 5G use cases, AI will be key for CSPs to run a network that enterprises can truly harness to serve their own needs. In short, AI unlocks a CSPs’ ability to realise new the revenue streams, that as mentioned earlier are in the trillions. But again, implementation isn’t easy. 

Challenges with automation 

Why isn’t it easy? Because CSPs have not yet been able to realise the potential of AI. In a 2019 study of 50 CSPs we conducted, we found the following when CSPs attempted AI implementation: 

  • 56% of CSPs face data quality issues
  • 55% lack the right data science skills
  • Only 20% of AI Proof of Concepts have so far scaled and progressed to live deployments

I know that the migration to automation can be extremely challenging. Some time ago, we helped a CSP implement an AI solution to troubleshoot network issues. The AI solution was trained and very accurate. However, initially senior engineers were reluctant to follow the AI recommendations – despite the fact that the AI was consistently correct and often could help solve the issue in minimal time. This example shows the natural human resistance that we often see when implementing automation, but it doesn’t have to be this way. 

Paths to solving that challenge 

Despite these challenges and resistance to change, we must take heart, there are solutions to these ‘teething problems.’ For example, CSPs can be supported to overcome the previously mentioned challenges: 

1.  Data quality 

Aggregating data effectively can take a great deal of effort to implement AI successfully. CSPs utilize a large variety of data sources, the list is extensive; OSS counters, device measurements, drive testing, probes, complementarity-determining regions (CDRs) from billing systems, customer care records, trouble tickets, external data like weather conditions and social media sentiment (which can give an early warning when problems arise). Interpreting and managing large amounts of data is absolutely critical, as typically only a small fraction of all this data is actually used. The task is to identify where to commit resource. Of course, this is not straightforward. Therefore, we must program AI to identify which slice of data is relevant, so the appropriate remediation response can be taken. Possessing a mature data hub that can automatically collect multiple sources of information will enable CSPs to implement AI into processes, compared to those without data hubs.

2. Data science skills

Moreover, it’s crucial to ensure your team has the appropriate skills and expertise to implement AI. A combination of data science and telecommunication expertise is required to implement AI successfully. However, data scientists are sometimes few and far between and are in high demand. So there seems to be a hurdle we must leap here. I’d advise retraining your high performing telecommunication engineers in the art of AI.  They don’t need to become masters level data scientists but should be comfortable with the basics of big data, analytics. I call this new important group of team members ‘Citizen Data Scientists’.

3. Successful scaling of AI

At a fundamental level AI is about change; change always impacts and is itself impacted by people, process and technology. It is people that are perhaps the biggest single factor affecting success or failure of AI. Implementing AI does require a great deal of revolutionary technology – sometimes daunting – and trial and error, so we must embrace risk, work collectively to unite teams to work towards an objective and be clear with employees about expectations. 

Furthermore, to be successful in introducing AI the CSP needs to have a clearly defined strategy, accompanied with clear business objectives and the support of senior management. AI is so vast that it can theoretically be applied everywhere, so we need a focus on where to deploy. My advice to CSPs is start by asking yourself the following questions:

  • What are the key areas and processes that require automation first? 
  • Which areas will produce the quickest Return on Investment?
  • How will success be measured?
  • How will you communicate the initiative to teams to ensure strong employee engagement and support?

When we can answer these starter questions, we can dive into the task before us. 

Call to action 

So, while the challenges to transitioning to 5G to unlock significant revenue may be daunting, I’m pleased that my industry is gaining further traction in supporting CSPs and enterprises. We do so because the words of Mr Buffett ring true in my ears and it is our responsibility to help CSPs and enterprises to thrive.

Dennis Lorenzin is Vice President of Network Cognitive Services at Nokia. Dennis has held several roles in the Nokia services organization throughout his career, operating fast changing environments, and gathering international experience in operational and financial management, innovation and transformation initiatives. Dennis believes that AI/Machine Learning are fundamental enablers to address the needs for efficiency in the wake of a surging technological complexity and bring value as the industry faces an increasing sophistication of 5G use cases. Dennis holds a MsC degree in telecommunication engineering and still enjoys lecturing in the academic environment on industrial trends and strategy topics.