85% of global telecommunications companies have at least one generative AI use case in production today.
Generative AI is quickly becoming one of the pivotal technologies of our time, inspiring the vast majority of telecommunications providers get in the game.
Research from Altman Solon, Scaling Proofs of Concept (POCs) to Production in Telecommunication: Generative AI for the Future Enterprise, highlights common challenges faced by 104 global telecommunications executives implementing generative AI across company types, sizes and regions.
How can AI transform your business?
Let's talk
Where are you now?
Each stage of AI maturity comes with its own possibilities and pitfalls, so it’s important to know where you are today.
Wherever you are in your generative AI journey, the most important step you can take is always the next one.
7 success factors for generative AI at scale
Success factor #1: Target pockets with common ROI
Productivity and efficiency gains were the most cited areas of ROI, with cost savings and time savings as key performance indicators (KPIs) and 63% of executives aiming for ROI targets of 5–15%.
Early success with 1,200 users led to wider acceptance, decreasing reliance on call center executives by 60–80% [and our] chatbot improved response times for escalations and downtimes, with ROI realized in the first year — a milestone for the company.”
Success factor #2: Know when, where and how to deploy your use cases to maximize success
The companies that saw the most success were the ones to seize quick-win use cases early, while still investing in bigger, bolder use cases with the right foundations to achieve phased success.
Customer service implementations of generative AI had the highest success rate, with about 80% of telecommunications companies scaling these use cases to production and ~45% achieving ROI targets. Other areas of significant success were IT and development, network, and business support services like finance, procurement and supply chain functions.
We successfully deployed AIOps in our network and security operations centers to replace manual monitoring with anomaly detection, root cause analysis and preventive maintenance. AIOps improved efficiency by 30%, outperforming our existing tools.”
Success factor #3: Invest in data and AI foundations that can grow with you
The right foundation maps directly to higher adoption, quicker ROI and overall success rates. Telecommunications leaders who cited greater success with use cases scaled to production highlighted these foundational factors:
- Access to quality data assets
- Scalable data infrastructure
- Effective data integration
On top of data and AI foundations, other clear enablers driving success included business-technology alignment, defined ROI and business case, and leadership buy-in — in other words, it’s vital to build AI and data infrastructure for scale while communicating and working alongside the business every step of the way.
Strong data-driven culture was crucial in assessing the tool's value, making the business case for its broader implementation more solid.”
Success factor #4: Focus on governance when it comes to data
The biggest difference between the broad innovators and those who are catching up is proficiency in data governance and data investment. As generative AI and AI technologies work in tandem to support large-scale use cases across the enterprise, data governance and modernized data infrastructure are key to success — mapping to an average increase of almost 15%.
- Modernize data infrastructure and IT systems. Almost 70% of executives cited an inability to integrate with legacy systems as a key impediment to preparing data infrastructure.
- Centralize and consolidate data. Siloed or fragmented data systems were the second-most cited blocker to data preparation and governance.
- Prioritize data and its investment in the organization. Broad innovators’ top technical concerns were data accuracy, reliability and governance, while the catching up group’s primary concern was a lack of quality data for training.
Generative AI initiates the process, handling data and metrics writing. Then, traditional AI takes over to develop and refine models based on this framework. Currently, generative AI covers 30–40% of processes, with traditional AI handling 60–70%. This balance may shift as both technologies mature.”
Success factor #5: Keep security, privacy and control at the heart of your project
There’s no true generative AI success at scale without security. To fully implement capabilities and harness value from data, security must be included early in the design — especially in connections across data silos and sensitive enterprise, customer and stakeholder data.
By incorporating security features early in the development process, telecommunications companies can prevent disruptions and maximize ROI during the full implementation.
Currently, our implementation level is at 35%. However, we face significant challenges with data security, compliance and GDPR, which slow down our progress. Without these challenges, our implementation could potentially reach 75%.”
Success factor #6: Be mindful of costs on the road to success
Telecommunications leaders widely regard high implementation and operational costs as the top generative AI business challenge, largely driven by inefficiencies related to legacy IT systems. Although most generative AI projects are given significant allocation in IT budgets due to their strategic usage, these costs add up.
Even the broad innovators found major difficulty in deploying generative AI projects while staying within budget. Almost 80% of telecommunications executives said they were above budgeted costs as they scaled generative AI use cases into production, and about a third were 10–50% over budget.
14% average allocation for generative AI in the total IT budget
The devil is in the data. Nearly a quarter of telecommunications executives cited large language model (LLM) development and maintenance costs as top cost drivers, followed by data preparation and data and AI employee costs. As a result, lagging data infrastructure and talent gaps constrain efforts to build scalable, effective use cases.
Higher costs mainly stem from model size and data volume. For instance, our customer service uses generative AI to analyze 85–90% of call transcriptions, handling millions of calls. Input and output token costs contribute significantly. Generative AI accelerates insight gathering from customer interactions — from 3–5 months to weeks — helping us quickly update messaging and enhance customer service.”
Success factor #7: Unlock the power of partnership
The study highlighted clear differences in success rates between telecommunications leaders who engaged partners versus building in-house solutions. The more successful companies from the broad innovators and cautious and focused groups brought in partners early, often and across the deployment process to maximize ROI.
Working with transformation partners like Kyndryl early in the process drives value every step of the way. Telecommunications leaders who worked with partners reported more success in achieving ROI targets and delivering large-scale use cases within budget, with greater accuracy and while keeping security intact. With hands-on experience and a proven track record, we’ll help you clarify your business case and expected ROI — before massive investments — and define a roadmap that meets you wherever you are in your generative AI journey.
Telecommunications
+ Kyndryl in action
Productivity
A US$19B global media and communications provider worked with Kyndryl to transform employee experience by integrating generative AI to streamline onboarding, IT support, administrative tasks and knowledge management. By enhancing operational efficiency, personalizing support and unlocking a more productive workforce, the scalable solution not only improved user satisfaction but also positioned the company for future innovation.
Cost optimization
A US mobile provider worked with Kyndryl to optimize cost and asset management in call centers with a generative AI-powered solution to improve data accuracy, operational efficiency and customer insights. The AIOps solution provided dynamic recommendations and securely managed business applications, supporting enhanced cost management, valuable insights for commercial customers and scalable IT services.
Scalability
A leading mobile communications provider worked with Kyndryl to enhance in-store retail customer service with personalized recommendations and real-time data insights from a live generative AI chatbot. The solution featured high-level architecture for personalized customer experiences and cutting-edge technology for sales and support, including AI automation to accelerate response times for sales associates while enhancing the effectiveness of their answers. This scalable approach is set to expand to over 1,000 retail stores, showcasing the potential for significant growth with the right foundation.
What’s next?
Wherever you are in your generative AI journey, Kyndryl can help.
