Private 5G & IoT: The Power Duet Fuelling Hong Kong's Smart City Revolution

Building a Financial AI Application Ecosystem

With its financial data centre located close to the Hong Kong Exchanges and Clearing (HKEX, 00388) data centre, NTT has long focused on serving financial clients. So revealed that NTT currently serves over 300 financial customers in Hong Kong, making it essential to prepare for the anticipated surge in financial AI applications.

“The past two years were only the beginning,” he said. “As LLM technologies mature, we believe many financial applications will be built on large language models.”

He noted that the rise of DeepSeek is accelerating the development of open‑source models, and that many LLMs will increasingly be deployed in private environments, particularly in the financial sector. Given the industry’s stringent regulatory requirements, developing financial AI applications in private environments is expected to become the prevailing trend.

In November last year, NTT was among the first to launch the APN InterLink service in Hong Kong, leveraging all‑optical network technology. Traditional networks involve optical‑electrical‑optical (O‑E‑O) conversions between transceivers, resulting in latency and energy loss. By contrast, APN connects two transceivers primarily through a single optical fibre, eliminating unnecessary conversion stages and significantly reducing latency while improving energy efficiency.

Data shows that APN connectivity can reduce latency by 30% and improve energy efficiency by 40%. So said that in the AI era, APN technology has broad applications—such as orchestrating computing resources across multiple locations via high‑speed networks, or synchronously invoking AI models deployed in different locations—offering customers greater flexibility.

Stephen Tsang, Chief Revenue Officer of NTT Com Asia, said APN technology is now fully ready for commercialisation, and with AI development driving demand for such services, he is confident customers will actively adopt financial AI applications if they deliver cost efficiency.

As an extension of its APN services, NTT plans to launch GPU as a Service in the second half of this year, driven by the sharp rise in AI computing demand—particularly for model training and inference in the financial sector. So added that, in addition to its own GPU resources, NTT will collaborate with companies that already have GPU capacity in Hong Kong.

At the application layer, NTT will build an ecosystem platform to aggregate AI solutions from ecosystem partners and deploy them in private AI environments, advancing AI adoption under a financial‑grade compliance, security, and data governance framework.