Ambitious operators are extending their product lines into AI-esque services, such as telco-specific large language models (LLMs) and even customer-facing personal AI assistants. These are commendable diversification strategies for those that are ready to compete with hyperscalers in their digital offerings. But one thing to consider is the huge expense of the infrastructure needed to run AI systems. While many operators are coming down from peak spending on 5G network infrastructure in 2021 and 2022, there are new costs on the horizon. Operators in Japan and Korea are making huge investments in AI infrastructure. Last week, Japan’s SoftBank announced it would spend JPY 150 billion ($956 million) over the next two years to upgrade its computing infrastructure with GPUs from…