The evidence base

Based on a December 2025 survey of 49 telecom CxOs globally. Key finding: most telcos have failed to generate real value from fragmented AI pilots, but forward-thinking operators are showing how agentic AI can be a paradigm shift. McKinsey notes that telecom revenue grew just 1% from 2010–2023 despite 60% annual growth in mobile data traffic.

Projects that Africa's top 5 data center markets will grow from approximately 400 MW today to 1.5–2.2 GW by 2030. Identifies major bottlenecks in power, fiber, and financing. Notes that McKinsey's model anticipates AI as a significant demand accelerator for African DC capacity.

BCG argues that telco CEOs must move beyond cost-cutting AI pilots to an enterprise-wide strategy focused on value creation. BCG's 10-20-70 framework emphasizes that algorithms account for only 10% of AI success — the rest is data (20%) and organizational change (70%).

Deloitte's blueprint helps telecom organizations scale and accelerate AI transformation. Estimates $150 billion in value can be unlocked through agentic AI across network management, service delivery, and customer care. Advocates "split inference" for privacy-first on-device processing.

Analysis of how McKinsey, PwC, BCG, Deloitte, and KPMG are approaching distributed AI inference. McKinsey projects 60–70% of all AI workloads will be real-time inference by 2030. PwC advocates a three-tiered portfolio approach. Deloitte forecasts 30% of smartphones and 50% of PCs will have local AI processing by end of 2025.

Jensen Huang announced NVIDIA's partnership with Cisco, T-Mobile, and Cerberus ODC to build AI-native radio networks. He stated that annual global capex in radio networks exceeds $100 billion, and this will inevitably shift to accelerated computing with embedded AI. NVIDIA also introduced the Aerial RAN computer for AI-native 6G networks.

At GTC 2026, Huang surveyed the accelerated computing ecosystem across automotive, financial services, healthcare, industrial, media, quantum, retail, robotics, and telecom. He cited $150 billion in AI venture investment and noted computing demand has increased approximately 1 million times over recent years.

Jensen Huang at Computex 2025 declared AI infrastructure will be worth trillions of dollars. He described AI data centers as "factories" that convert energy into tokens — the building blocks of AI value. This reframes the telecom opportunity: telecoms can own these token factories.

Reports that hyperscalers are budgeting nearly $700 billion in capex for 2026. Meta plans up to $135 billion, Google up to $185 billion. NVIDIA's revenue increased 73% to $68.1 billion in Q4 2025. The question Jensen Huang is now answering: who beyond hyperscalers will drive the next wave of AI infrastructure demand?

Analysis of 250 operators globally. 75–80% of AI deployments target cost cuts; only 10–20% focus on revenue generation. Early revenue-driving use cases like GPU-as-a-Service and agentic AI are expected to grow in 2025–2026. About 60% of deployments are live. Asia Pacific leads with ~4 projects per operator.

Documents $100B+ in AI and data center commitments across the Gulf, North Africa, and select Sub-Saharan markets. Telecom-led platforms are gaining relevance relative to pure-play developers — control over fiber, landing stations, and regulated infrastructure gives telcos structural advantages.

Analysts forecast over $100 billion in emerging market data center investment by 2030. Identifies Jakarta, Nairobi, Accra, and Lagos as key growth cities. Notes that sovereign funds are co-investing in sovereign cloud infrastructure as part of national AI strategies. Over 270 million people in Indonesia alone creating urgent demand.

Indonesia's telecom market valued at approximately $13.4 billion in 2025. Data center market projected to reach $3.79 billion by 2030 at 7.99% CAGR. Three major operators — Telkomsel, Indosat, and XLSmart — dominate. Jakarta is the primary DC hub with Batam emerging as a strategic alternative.

Comprehensive review of African telecom developments: Safaricom grew 5G to 1,700 sites covering 30% of Kenya's population. MTN's Bayobab targeting 135,000 km of fibre. Vodacom acquiring 30% of Maziv for $790M. Nigerian operators unlocked over $1 billion in new infrastructure spending following tariff resets.

Safaricom committed $500 million over three years for AI-ready infrastructure across East Africa. Partnered with iXAfrica for 22.5 MW NBOX1 Nairobi Campus. Deployed Zuri AI chatbot for 30M+ M-Pesa users. Achieved 89% accuracy in fraud detection using graph neural networks with AWS. Trained 5,000 employees in AI.

Africa's largest mobile operator (300M+ subscribers) partnered with Microsoft to democratize AI access across the continent. The partnership positions MTN as an AI distribution channel — using its existing subscriber base and billing infrastructure to deliver Microsoft's AI tools to African businesses and consumers.

Analysis of how Huawei evolved from a telecoms equipment company to a full-stack AI leader — building Ascend chips, AI CloudMatrix data centers, Pangu industry-specific AI models, and Huawei Cloud. Revenues in 2024 rose 22.4% to 862 billion yuan. Jensen Huang himself called Huawei "one of the most formidable technology companies in the world."