NTT develops enhanced AI performance for 6G 

Intelligence in the 6G era

NTT announced two key initiatives around the advancement of AI to meet data processing needs of 6G use cases. 

The first is a joint project between The University of Tokyo, NTT and NEC Corporation (NEC), integrating 6G and IOWN technologies to achieve high-capacity data communication and optimisation of computation processing required for AI agents.

“The successful integration affirms the establishment of infrastructure for next-generation ICT systems to enable AI agents to act autonomously, processing and transmitting massive amounts of data with low latency and high reliability. To control computational resources used, reduce end-to-end delay and maintain AI inference accuracy, this approach combines small, specialised AI distributed in a network and external information sources, streaming semantic communication technology and selective media control technology,” detailed the company, in a press statement.

The second announcement (as detailed here in full on the Computer Weekly Developer Network) surrounded a demonstration by NTT and NTT Docomo of low latency and high-speed AI video analysis using In-Network Computing (INC) Edge. 

Traditionally, AI inference processing has been controlled by applications and servers, with data transfer controlled by the network. The proposed method uses INC edge to connect commercial 5G networks and IOWN All-Photonics Networks (APN) and enables distributed GPU resources to be used as part of the 5G network, reducing communication delays.

Experts anticipate that networks in the 6G era will control communication, data processing and inference to ensure high quality of service for data-intensive applications such as AI and robotics. 

Photonics for an AI-driven society

Throughout MWC, the NTT Group booth showcased a range of initiatives that use optical technologies to support the adoption of AI and drive the transformation of datacenter and compute infrastructures. 

These exhibitions were structured around technologies that use IOWN to build energy-efficient infrastructure for an AI-driven society – AI-Resilient Infrastructure with Photonics – and real-world use cases for AI – AI-Powered Services and Solutions.

NTT also highlighted IOWN technologies, including the photonics-electronics convergence (PEC) devices spotlighted in NTT president and CEO Akira Shimada’s keynote. The company looked at how they improve energy efficiency in datacentres and optical quantum computing to enable large-scale computation with lower power consumption, lower costs and high speed. 

The company also presented recent advancements in the integration of AI within 6G mobile networks, including “Network for AI”, a concept envisioning seamless coexistence between humans, AI and robots.

Shimada san

As part of the event, CEO Shimada-san delivered a keynote speech titled “Photonics Unlocks an Intelligent Power-Optimised Future” and talked about the power of IOWN to reduce power consumption, challenging the status quo in the AI era. In his presentation, he introduced IOWN initiatives, including the commercialisation of PEC devices and the development of optical quantum computers.

“As AI adoption accelerates the rate at which power consumption is rising, IOWN has become a strategic foundation aiming to address both the explosive growth in data use and the global need for lower energy consumption. IOWN will support AI scaling through sustainable connectivity, helping to transform bold ideas into tangible impact,” said Shimada-san.

He finalised by saying that by replacing electronics with photonics, his firm hopes to give people and businesses the capacity to move beyond the limits of conventional Internet infrastructure.

Six MWC themes

The six MWC themes this year were as follows:

• Intelligent infrastructure: AI-native, cloud-based, automated networks as a foundation layer.
  
• Connected AI: aligned to revenue-generating AI deployments.
  
• AI for enterprise: spanning vertical industry use cases,
  
• AI nexus – issues related to sovereign AI, governance and the control layer.
  
• Game changers: defined as direct-to-device satellite, quantum and frontier innovation.
  
• Tech for all – Sustainability, inclusion and long-term legitimacy.

Rise of the neocloud

As demand for AI is fragmenting the landscape, giving rise to neocloud providers, sovereign clouds and specialised local services, this has spurred a rise in private fibre networks, supported by Managed Optical Fibre Networks (MOFN), these architectures historically favour hyperscalers.

As the Computer Weekly Developer Network has explained before, “A neocloud is emerging as an AI-specialised cloud offering that strikes a balance between the vast, general-purpose scale of hyperscalers (AWS, Azure, GCP) and the high control of on-premises deployments.”

Neoclouds offer an alternative to hyperscaler clouds. In some cases, they are potentially 20 times cheaper for AI workloads and even move towards the fabled zone of outcome-based pricing. In the MWC session, experts showcased why they think All-Photonics Networks (APNs) are a viable alternative as they enable new cloud providers to compete while allowing hyperscalers to extend their reach.

Smart Slicing

A final session was Private 5G in the AI-Native Era: Smart Slicing at the Enterprise Edge again with NTT’s Parm Sandhu.

Sandhu and team explained how edge AI is helping enterprises and telcos to now embed intelligence at the network edge, enabling millisecond decisioning for manufacturing, autonomous vehicles, real-time video analytics and mission-critical IoT.

Private and edge-enabled deployments are scaling fast, with suggestions of the global private mobile network customer base surpassing 1,846 customers in mid-2025, reflecting rapid adoption of campus and industrial networks. Market forecasts similarly predict a multi-billion-dollar expansion as enterprises combine private 5G, edge compute and AI to cut latency, conserve bandwidth and meet data-sovereignty needs.