The Role of Artificial Intelligence in the Transformation of Telcos

This past week, I had the pleasure to moderate a panel on digital transformation and to join my Artificial Intelligence-expert partners at Xona in conducting a workshop on AI at TMT Finance World Congress to a crowd of ecosystem professionals. It is ironic that an industry that is a catalyst to the transformation of many other industries is itself in dire need to change. Telcos increasingly feel they are in a ‘straight jacket’ with few and painful options [some telcos actually still don’t see that: many incumbents still enjoy regulatory protection and sit happy where they are!]. Launching new services in the digital world is exponentially more complex with hardware-based networks. Transformation into efficient ‘digital entities’ through software and data is hard. Financial markets value enterprises with consideration to their future cash flow. Amazon, Google and Microsoft market cap alone stand at ~4x that of AT&T, Verizon, Sprint and T-Mobile combined. With the background, what does AI has to do with digital transformation?

The AI Factor

I’m struck by how many people thing AI is ‘something new’. The roots of AI trace to the 1950’s and has already gone through three hype cycles. While we may well be in another cycle today, but what’s different this time and what the future foretells? In perspective, the combination of big data, compute power cost and performance, and improved algorithms make all the difference. AI is real, so how can one leverage AI to get ahead by improving performance and quality of experience, reducing cost, and/or monetizing new services?

AI Timeline [Nvidia].

Making Use of Data

In the world of AI, data is critical. In fact, many old algorithms proved their effectiveness beating ‘fancy’ algos. Many organizations sit on loads of data but do nothing with it. For example, telcos possess a huge amount of data spewed from thousands of network elements. But this data is used primarily for reactive monitoring by technicians staffing network operating centers. Through AI, this data could be used, for example, to predict faults and lead to a completely different way of managing the network: a proactive approach that will change the whole concept of a NOC!. Similarly, consumers through their cell phones generate much data that the OTTs have leveraged to a great success even as they sit a layer removed from the end customer, while the service providers who are closest to the end customer failed at monetizing a single service through this data.

Will this picture of something of the past thanks to AI?!

The OTT Factor

The OTTs are the leading driver behind the current cycle of AI as the huge investments in this field made by Amazon, Microsoft, Google, Facebook and the others tell. AI is at the heart of many of the services they provide, as for example AWS and Microsoft’s IoT services, or voice recognition functions from Apple, Google or Amazon, leading to future application in autonomous car, or perhaps less futuristic applications in autonomous wheelchairs! Whatever it is, AI is at the heart of many of the services the OTTs are offering. But this did not happen by chance. The OTTs have been investing in AI, developing their own internal capabilities and making acquisitions in this space, often of small companies, but increasingly spending big dollars to make larger acquisitions. Moreover, OTTs have invested in hardware to accelerate AI as for example Google Tensor Processing Unit (TPU).

Performance / watt, relative to contemporary CPUs and GPUs (in log scale)(Incremental, weighted mean) [Source: Google]

Action not Talk!

While the telcos talk about digital transformation, the OTTs are delivering web-scale services at a pace that the telcos cannot match. In part, this is due to legacy, hardware-based networks. In other part, it is due to the culture, people and skillset that resides in these organization. Technology is only one factor, but the other factor is leadership, culture and skillset. Telcos handed over the system integration function to their vendors while the OTTs have control of their own destiny. This widens the gap in the type of people and culture necessary to successfully execute on technology transformation initiatives. AI ‘rockstars’ combine technology, math and domain expertise. Building this expertise is a painstaking process that has to be managed carefully. Additionally, the telcos have to develop the proper incentives to win in the talent acquisition battle.

Investing in AI

Integrating AI into service operations requires a balanced and well thought out strategy. There are many elements to consider. People and talent, leadership support, culture, are a few such elements. Many acquisitions in the AI space fail because the acquirer cannot handle the integration. In fact, to successfully assimilate AI expertise, the acquiring company needs to have such expertise of its own. Developing core expertise in AI is a fundamental first step. This is only now beginning to happen at a few service providers.

It would help the telco service providers to look at the investment strategies in building, acquiring and assimilating AI expertise at leading companies, especially the cloud players. After all, it is their roadmap that drove the explosion in AI activities and they remain the driving force behind AI.

Xona Partners Executive Workshop on Artificial Intelligence

In the Balance

AI is only one aspect in transforming telco service providers. With networks becoming more complex, it will become a necessity. In parallel, using AI to improve revenue is no less important. To successfully leverage AI, the telcos need to develop the proper strategy, including the culture, people and skillset, among other elements. This brings me back to the Xona Executive Workshop on AI. Part of the workshop focused on providing the basics: highlight AI developments, its trajectory and salient points. Then, we looked at how successful acquisition and integration of AI capability need to happen. The space of AI intersects with multiple technological developments such as edge compute and blockchains. How could one leverage these developments? We address these in case studies in open discussion forum.