Nokia, Blaize, Datacomm Sign MoU for Hybrid AI Infrastructure in Indonesia, Targeting Telecom Edge

AI computing hardware firm Blaize announced on Tuesday, April 29, 2025, that it has signed a Memorandum of Understanding (MoU) with Nokia and Indonesian IT services provider Datacomm Diangraha to accelerate the deployment of hybrid AI inference infrastructure across Indonesia. The tripartite agreement, first reported by Developing Telecoms, marks a significant strategic move by a major network vendor to integrate specialized AI accelerators directly into telecom infrastructure, positioning the edge network as a critical platform for enterprise and industrial AI workloads.

Technical Architecture: Integrating Graph Streaming Processors into Telecom Networks

The core of this partnership involves integrating Blaize’s Graph Streaming Processor (GSP) architecture into network infrastructure provided by Nokia. Blaize’s hardware is designed for low-latency, energy-efficient AI inference at the edge, a key requirement for real-time applications like video analytics, industrial automation, and augmented reality. The “hybrid” nature of the infrastructure implies a distributed AI model where inference can occur at multiple points: on-premises at an enterprise site, at a local multi-access edge computing (MEC) node hosted by a telecom operator, or potentially within Nokia’s radio access network (RAN) equipment.

Nokia’s role will be to provide the underlying connectivity and cloud-native platform, likely leveraging its AVA AI software and Digital Automation Cloud offering. Datacomm Diangraha, with its extensive enterprise reach in Indonesia, will act as the system integrator and go-to-market channel. This model represents a shift from centralized, cloud-only AI to a telecom-centric, distributed inference model. It directly addresses bandwidth constraints and latency sensitivity, moving the compute to where the data is generated—a fundamental principle of edge computing that is now being applied to AI at scale.

Industry Impact: A New Revenue Model for Network Operators

For telecom operators, particularly in high-growth markets like Indonesia, this partnership signals a clear path to monetizing 5G and fiber investments beyond mere connectivity. By offering AI inference as a managed service from the network edge, operators can tap into the burgeoning enterprise AI market. This moves them up the value chain from being a “dumb pipe” to becoming a critical enabler of digital transformation.

The competitive landscape is also affected. Nokia’s collaboration with a pure-play AI hardware company like Blaize is a direct challenge to other network vendors who are developing in-house AI silicon or relying on generic GPUs. It suggests a best-of-breed approach where network vendors act as ecosystem orchestrators. For infrastructure investors, this underscores the growing importance of the AI-ready network, where capital expenditure will increasingly flow towards edge data centers, AI-accelerated RAN hardware, and high-capacity fiber backhaul to support the data flows between distributed inference points.

Strategic Implications for Asia-Pacific and Emerging Telecom Markets

Indonesia, with its population of over 270 million, rapid digitization, and government-driven initiatives like “Making Indonesia 4.0,” presents a prime testbed for hybrid AI infrastructure. The deployment model established here could become a template for other markets in Southeast Asia, Africa, and the Middle East where low latency and data sovereignty are major concerns. The involvement of a local powerhouse like Datacomm Diangraha is crucial for navigating regulatory environments and understanding local enterprise needs.

This initiative also highlights the strategic importance of the Asia-Pacific region in the global AI infrastructure race. As data governance laws tighten, the ability to process data within national borders becomes a competitive advantage. Telecom operators, by virtue of their distributed physical footprint, are uniquely positioned to offer localized AI processing that complies with data residency regulations. This partnership could give Indonesian operators like Telkomsel, Indosat Ooredoo Hutchison, and XL Axiata a first-mover advantage in offering sovereign AI cloud services.

Forward-Looking Analysis: The Network as an AI Inference Plane

The Blaize-Nokia-Datacomm MoU is more than a single project announcement; it is a harbinger of a structural shift in telecom architecture. We are witnessing the early stages of the network itself becoming an AI inference plane. Future network deployments will be evaluated not just on bandwidth and latency, but on their native AI processing capabilities. This has implications for RAN evolution (Open RAN with AI accelerators), core network design (AI-optimized traffic routing), and submarine cable investments (to handle aggregated inference results and model updates).

For the global telecom sector, the race is on to define the standards and commercial models for edge AI. Partnerships like this one will determine whether telecom operators capture a significant portion of the AI infrastructure value chain or remain relegated to providing transport. The success of this Indonesian venture will be closely watched by operators worldwide, as it validates—or challenges—the economic viability of distributed, telecom-hosted AI inference at scale.