The problem goes beyond a mere lack of data center capacity. The massive transition of B2C and B2B users to multimodal AI interfaces (real-time voice and streaming video interactions with agents) is clogging the bandwidth of cell towers. The inference of heavy LLMs requires an instant response (low latency), which current congested networks cannot provide. For macroeconomics, this is a powerful signal: the trillion-dollar capital expenditures (CAPEX) on GPU purchases are now compounded by the necessary investments in a radical modernization of the global telecommunications infrastructure. Without expanding data transmission channels, the entire AI ecosystem risks getting stuck in endless buffering.
Source: VodafoneThree / TNW
TelecomInfrastructureUKCAPEXMacroeconomics