AI simply won’t deliver on its promise unless it has sufficient compute power
There is one thing that we can count on as we head towards the end of 2025 and the beginning of 2026: Interest in “all things AI” will continue to grow. This applies to all industries, with telecom AI investments expected to rise from $3.6 billion in 2024 to $187.7 billion by 2034.
Telecom operators are positioned to play a key role in this disruptive technology. However, amid the AI explosion, it’s the compute power that is fast becoming the “new gold.” Whether on chips, in data centers, or on edge servers, AI simply won’t deliver on its promise unless it has sufficient compute power.
The large-volume data quandary
One area where compute power is needed is in processing the large volumes of data that AI-native technologies generate. Take, for example, generative AI (gen AI). If gen AI makes up only 5% of daily searches globally, that would require 20,000 servers that consume 6.5 kW on average per server to fulfill prompt requests, according to Deloitte. Goldman Sachs further predicts that, with the impetus from AI, power demand will spike by 50% in 2027 and as much as 165% by 2030.
Telecom operators have specific data needs compared to typical enterprise companies. This entails a profoundly different approach to data infrastructure, using highly efficient architecture. Operators need to capture, store, process, and analyze all network data and events. These include AI-related and non-AI-related workloads, over the top (OTT) solutions, voice calls, and more. For instance, video, audio, and imaging apps are significantly data heavy. Or IoT devices, emergency services, and autonomous vehicles connected to mobile networks, may have sensors and cameras, resulting in particular data requirements.
Telecom’s data complexity
Storing and processing the data is just one challenge. Operators are already processing billions of micro-transactions daily and this is expected to rise significantly. One hour of a live-streaming football match can use 7-10GB alone. Multiply this by millions of subscribers and it translates to significant volumes of data.
Beyond the size of the data, its complexity presents an equally significant challenge. If operators wish to leverage telecom data for use in agentic AI or generative AI solutions, it means ensuring data quality. This necessitates a solution that captures all data and events, from the radio access network (RAN) to the core network, offering a complete network overview. Furthermore, it must seamlessly integrate between real-time and historic data to maintain data integrity and provide customer-centric insights.
Do more with less
Tackling this compute-power-data-challenge leaves telecom operators struggling “to do more with less.” If a large-size operator is already managing 8,000 servers with the current data loads, adding more servers — whether they are on premise, at the edge, or in a data center — poses a dilemma: How to transition to the new AI-telco-era, while keeping costs down? McKinsey estimates that data centers alone will need to spend $6.7 trillion to keep pace with the demand for compute power by 2030.
Don’t disregard next-generation assurance
Next-generation service assurance solutions offer a data and insight foundation layer, designed to monitor tens of millions of connected devices and subscribers. This includes high speed voice, data services with encryption and non-encryption, changing conditions across all regions and more. The next-generation solutions comprise highly efficient architecture and cost-effective capacity to process massive data traffic. This means excellent service quality and network optimization even for the largest and most complex AI-driven networks. These deliver “more” on fewer servers, therefore reducing integration costs and monitoring larger amounts of voice and data traffic on the network with minimal effort. It also means that they can consolidate network monitoring platforms and other BSS/OSS solutions into one unified solution to optimize computing resources and reduce costs.
Bottom line
Telcos have the ability to transform industries from healthcare to media to manufacturing, and even mission critical services. Nevertheless, playing a key role in this AI-led era comes with significant challenges. The key, however, is taking a “work smarter, not harder” approach. Seek solutions that deliver increased value with reduced inputs. Next-generation assurance solutions deliver higher output using fewer resources — assisting operators to maximize compute power with efficient data processing, end-to-end visibility, and real-time network insights.
Table of Contents
