The latest report from GSMA Intelligence has explained how telecom operators can boost efficiency with the deployment of AI (artificial intelligence) solutions.
Peter Jarich Head of GSMA Intelligence, said: “Telcos play a role in how AI evolves and proliferates – both in terms of their own operations and the AI-enabled services they offer.”
The report has the following statistics:
81 percent of operators globally are testing genAI solutions.
42 percent of operators see today’s early stage of genAI technology development as the top obstacle to deployment (followed by uncertain RoI).
4 percent of Opex savings can be achieved if energy costs are reduced by 20 percent.
AI achievements
UK-based Vodafone said it will enhance customer experience with Microsoft’s genAI tools as part of its partnership signed in early 2024. Vodafone says it has already achieved multilingual virtual assistant improvement and 50 percent boost to customer care enquiry resolution.
Use of energy accounts for 45 percent of network Opex and 15–20 percent of total operational spend for telco. Telecom operators are using AI in RAN, site location and cooling systems. Telecoms have noticed that Al can help lower energy consumption and improve profit.
SK Telecom at the DTW24 event said its CS Agent has achieved 77 percent response rate using a telco-specific LLM, compared to 44 percent with a general LLM. Telco-specific LLMs enable agents to access information from operational support and business support systems, resulting in accurate responses to customer queries.
China Telecom says its AI brand Xingchen has enabled its enterprise clients to achieve costs reduction and efficiency enhancement. China Telecom has launched 32 large vertical models in fields such as industrial, education, healthcare, as well as government administration.
Main developments
Indian telecom operator Reliance Jio said its AI service called Jio Brain accelerates AI adoption across Jio, driving faster decisions, more accurate predictions, and better understanding of customer needs.
China Mobile has launched 23 AI+ products and 20 AI+ DICT industry applications. China Mobile has deployed industry-specific large models in 40 sub-sectors. China Mobile has released 17 self-developed Jiutian industry-specific large models in the first half of 2024.
China Unicom has upgraded the UniAI large model system, pushing the number of parameters from hundreds of billion to trillion. China Unicom accelerated model application and formed over 35 industry-specific large models, empowering areas such as urban governance, economic operations, information consumption, and industrial manufacturing.
South Korea-based telecom operator KT announced three AI technologies to reduce power consumption: Equipment Temperature Management, Energy Saving Orchestrator, and Server Power Supply Optimization.
Indian telecom operator Airtel and Google collaborate on cloud and genAI. Airtel and Google will work together on sales efforts, pairing Airtel connectivity and Google AI assets such as an Airtel IoT solution and managed services.
US-based telecom operator Verizon unveiled AI tools to transform customer experience. Verizon is deploying humanaided genAI to simplify user experiences, aiming to reduce cognitive load on store and customer-service partners.
Switzerland-based Swisscom launched Swiss AI Platform building on previously announced work with Nvidia.
Middle East-based Ooredoo has become an Nvidia Cloud Partner, developing an AI-ready platform for users in Qatar, Algeria, Oman, Tunisia, Kuwait and the Maldives.
Telecom operators such as e&, SK Telecom, Deutsche Telekom, Singtel and SoftBank — at the DTW24 – Ignite event — signed an agreement to co-develop a multilingual, telco LLM.
Singapore-based Singtel has partnered with Hitachi AI business to start a pilot at a US factory for Industry 4.0 use cases.
Japan-based KDDI will work with Datasection, Supermicro, Sharp and Nvidia to build an AI data centre, supplying construction, operations and infrastructure support.
Germany-based Deutsche Telekom is using AI in its data centers and networks to increase energy efficiency. In the mobile network, an AI solution puts antennas into sleep mode when they are not needed, thus saving energy.
Norway-based Telenor, as part of AI-first strategy, has partnered with Google Cloud to deliver a unified data and analytics platform that will enhance customer experience, support new product development, and simplify operations.
Japan-based SoftBank has tied up with Nokia for the development of a communication system that utilizes AI aimed at the development of AI-RAN and 6G technology.
Telefonica’s AI strategy focuses on responsible and ethical implementation of the technology across its operations. Telefonica has deployed more than 500 AI systems and is committed to improvement and ethical AI development.
Cost for the deployment for AI
Cost for the deployment for AI is a big challenge for telecom operators. GSMA Intelligence notes that the creation of a telco-specific LLM requires substantial capital. This is prompting operators to explore cost sharing. In 2023, SK Telecom invested $100 million in AI research company Anthropic to develop a telco-focused LLM.
GSMA Intelligence says upfront and ongoing costs such as licensing associated with developing genAI are significant.
Around two thirds of operators are viewing troubleshooting and maintenance as having the biggest impact from genAI, GSMA Intelligence survey data says. Hyperscalers account for 1 percent of global energy use. AI training and inference is likely to drive that up to 1.5–2 percent in the next five years.
Energy accounts for 15–20 percent of Opex for operators. Telecoms aim to use AI in order to reduce power consumption by 20 percent that would lower Opex by around 4 percent.
Baburajan Kizhakedath
