Singtel and Ericsson have agreed a five-year partnership to expand 5G Advanced services and AI-powered network capabilities in Singapore, with the two companies targeting improved performance and reliability for enterprise customers.
The agreement extends a collaboration announced in March and covers the commercial deployment of network slicing and programmable connectivity for applications that need low latency and consistent performance, including industrial robotics, autonomous vehicles, smart manufacturing and logistics operations.
Programmable connectivity with contractual commitments
Singtel's plans include end-to-end network slicing backed by enforceable service-level agreements (SLAs), an initiative first set out in March that is intended to give enterprises defined performance commitments for specific applications. The operator is developing premium service tiers for enterprise, automotive, public sector and digital-native customers, and intends to expose programmable network features through commercial APIs so that supported applications can request network capabilities directly.
The companies have outlined a phased roadmap with milestones in 2026, 2028 and 2030, but did not provide launch dates for individual services.
Port deployment and market context
An existing deployment is at PSA Singapore's Tuas Port, where network slicing supports communications between automated guided vehicles, cranes and other port systems. It builds on a project announced in November 2024 to upgrade automated guided vehicles to 5G and to support real-time shipment tracking and crane operations, alongside plans for private network connectivity matched to the port's security requirements. In October 2026 the companies confirmed that slicing was carrying secure, low-latency traffic across the port's automated systems; the original project also envisaged using real-time shipment data to improve berth scheduling and cargo movement.
Ericsson's June Mobility Report counted 84 commercial offerings based on 5G standalone network slicing, up from 65 six months earlier — a measure of how many operators now sell differentiated connectivity services.
AI in network operations
Beyond connectivity, the partnership covers AI-assisted management of network operations. Singtel and Ericsson plan to extend automation across radio access and core networks so that traffic patterns can be predicted, resources allocated more efficiently and performance problems identified earlier.
A live-network trial of AI-powered radio access network capabilities produced a 19% improvement in throughput, according to the companies, which did not disclose testing conditions or corresponding figures for latency, availability or enterprise SLA compliance. The work also covers service assurance, with AI-assisted monitoring and automation intended to improve operational efficiency and reliability, and the companies are working towards Level 4 autonomous network operations under TM Forum's framework, as set out in March. That level involves predictive management and closed-loop automation in which systems adjust resources without manual intervention; Singtel has not said its network has reached it.
The companies also said the automation programme will address network resilience, cybersecurity and energy efficiency.
Singtel and Ericsson will additionally explore connectivity at Changi Airport Terminal 5, Bulim Square and upcoming JTC developments, with potential uses including autonomous mobility, remote monitoring, intelligent inspection and real-time video analytics. On compatible newer devices, the partners have introduced network enhancements supporting peak uplink speeds of up to 250 Mbps, which they say will help data-intensive applications such as high-definition video transmission and AI-driven video analytics.
"Our priority is to make these capabilities more widely accessible and commercially viable," said Ng Tian Chong, chief executive of Singtel Singapore.
The announcement sets out intentions and a roadmap rather than completed rollouts; apart from the Tuas Port deployment and the throughput trial, most elements remain planned.
