Telco vendors are rushing agentic AI into OSS and BSS stacks – hard, at DTW Ignite this week. Nokia has just announced work with AWS and Databricks to build the data, cloud, and control layers for autonomous networks.
In sum – what to know:
Control layer – Telco vendors are converging on agentic AI for OSS and BSS, and making a noise at DTW Ignite; Nokia is extending its orchestration ‘fabric’ with new integrations with Databricks and AWS.
Data layer – The deal with Databricks extends its ‘fabric’ into fragmented telco data silos, introducing a unified platform and code-once workflows to reduce platform lock-in and enable cross-domain AI agents.
Cloud layer – Its expansion with AWS positions its ‘fabric’ as a cloud-hosted control layer, integrating OSS applications, AI services, and intent-based automation into a scalable execution environment.
We wrote about this yesterday a little – that while telcos may yet reinvent themselves for the AI era, and while some have fairly coherent strategies to do just that, there remains a lot of work just to get their houses in order. Which creates some tension between how the industry does the same with less to save money (which is the default AI change story) and how it does more, and differently, in order to earn more. DTW Ignite in Copenhagen has become an essential date in the telco calendar, notably as telcos have seized on its sliding-scale accreditation for network autonomy, levels one to four, and descended mob-handed on the Danish capital this week to hash out how to make AI work better in their labyrinthine operational and business support (OSS and BSS) infrastructure.
Little surprise, too, that the vendor community, selling to them, has issued a rush of announcements about their latest AI-tooled OSS and BSS shenanigans. There is not much here about telco reinvention, in the blue-sky sense; it is all about the kinds of practical nips and tucks that AI agents might stitch into telcos’ day-to-day operations. Yesterday, Nvidia announced agents to integrate telcos into the edge extension of central AI infrastructure, Nokia introduced agents to consolidate control of telco data and operations in a unified stack, and Blue Planet released agents to coordinate service workflows across multi-vendor/domain environments. Others will have announced versions of the same; and Nokia has rolled the dice again, today, with collaborations with Databricks and AWS.
The narrative is very familiar; these are not isolated partnerships for Nokia, but stacked components in the wider architecture it is assembling around its so-called Autonomous Network Fabric (ANF), which is like an operating system for telcos’ radio, core, transport, and service domains. The announcements are quite technical, and covered below, but Nokia is essentially showing a unified telco data platform with Databricks and extended cloud integration with AWS; both are about powering-up its own ‘fabric’ proposition, and helping telcos to re-organise their back-office functions to have a crack, some day, at full-scale level-four autonomy. Both put the fabric in the middle of the new AI value chain for telcos – as an AI-automation layer that consumes data, applies models, and triggers actions.
The Databricks partnership addresses the data layer, to reformat messy telco silos in a single view, and the AWS partnership deals with the AI cloud layer, to integrate rarefied telco edge workloads with AI models in data centres; in between (or on top), they plug into Nokia’s orchestration fabric, as the control plane for automated telco-wide operations. The structure says telco data is unified and standardised in the Databricks-style ‘lakehouse’, telco workloads run in AWS cloud environments, telco intelligence is generated via Nokia-made domain models and digital twins, and execution is handled in Nokia-controlled orchestration systems. And there are agents everywhere, running between. The physical network is owned and operated by the telco, as always.
Maybe when the fabric is finished, and telcos deploy it at scale, they will properly reinvent themselves in the AI value chain, as well, beyond just souped-up optical fibres and radio waves. In the meantime, or just for now, we should take a look at these functional AI-geared step-changes in familiar network operations…
Nokia and Databricks – the data layer for autonomous networks
Nokia has a new proof-of-concept with US cloud platform provider Databricks of a “unified, substrate-agnostic” data platform designed to support autonomous networks. The point is, as the industry gathers at DTW, to show how telcos can simplify and organise their fragmented data environments in order to deploy real-time analytics at scale. Nokia writes: “Telecom networks typically rely on hundreds of siloed operational and business support systems, each with its own data architecture, making it difficult to apply AI consistently across domains. To truly harness AI and multi-agent systems, operators need a common data platform that can run seamlessly across different cloud environments or on-premise infrastructure, without the need to rewrite code.”
The proof “confirms” the two can develop a joint architecture together to handle the scale and speed to feed network data to AI agents to automate cross-domain actions, they said. Engineering teams simulated analytics ingestion with an “intent to scale quickly” to match tier-one operator scale in the cloud. They are claiming a number of technical breakthroughs. These include code-once data-processing workflows for network data that can run across proprietary platforms (like Databricks) and open-source stacks (tested with Apache Flink, Kafka, and Iceberg). They also introduce vendor-neutral data transformation logic to reduce platform lock-in, achieved by separating core processing logic from platform connectors so the same workflows can be reused across different environments.
As well, a custom compiler was tested to convert Python logic, about how data should be processed, with the right connectors into the right platform formats, and automatically to adapt data workflows at deployment. Again, the point is to allow the same workflows to run in different environments without manual rewriting. The demo also showed how AI can create data products from natural language prompts, with an agent on hand to generate the workflow, request human approval, and put it to work, also reducing engineering effort. The architecture is built for agents, they said. It generates query-time data products instead of pre-storing datasets, which compute metrics and join when required, with support for zero-copy data sharing across domains.
Oguz Sunay, chief technology officer for autonomous networks at Nokia, said: “By enabling a common, flexible data platform across cloud environments, we can help operators accelerate the adoption of AI and create more efficient, resilient and sustainable networks.”
Nevash Pillay, global head of telecoms at Databricks, said: “Operators are managing increasingly complex networks and need a more consistent way to harness their data. Our collaboration with Nokia demonstrates how a unified data platform can help simplify operations and unlock the value of AI across network domains.”
Nokia and AWS – the cloud layer for autonomous networks
Meanwhile, Nokia is working with AWS on the same, a level up. The Finnish firm’s Autonomous Networks Fabric will run on AWS from later this year. “This is how telcos will compete in the AI era,” said Nokia, also signposting the path to level-four autonomy. Its OSS apps for orchestration, assurance, and inventory management are already available on AWS. Its ‘fabric’ proposition uses sundry agents and twins, plus intent-based networking and unified data management (per the work with Databricks), to deliver observability, analytics, security, and automation. The move to put the system on AWS will bring better cloud scalability, availability, and choice, it said. On the latter, Nokia referenced Amazon’s Bedrock and SageMaker tools.
It also said it is developing an optimized cloud footprint to shrink compute and storage requirements versus traditional on-prem deployments, it said.
Nokia and AWS have history, of course. They have recently showed the “industry’s first” agent for network slicing with du and Orange (in March), and the “world’s first” commercial 5G service on a cloud-hosted (SaaS) core, with Citymesh in Belgium (February). Nokia reckons its autonomous networks portfolio is delivering measurable results already, with operators achieving automation rates higher than 90 percent, service delivery times of four hours or less, and service interruption periods of one minute per year or fewer – along with up to 85 percent reduction in slice rollout time and up to 50 percent fewer customer-impacting incidents. (These stats are copied directly from the press note.)
Sunay at Nokia, said: “Autonomous networks have gone from far-off vision to business imperative. At Nokia, we move operators toward greater autonomy through the convergence of intent-based networking, agentic AI, and cloud-native architecture… We’re building a platform that scales operators’ ambitions while maintaining the control and governance they need. This is how telcos will compete in the AI era.”
Amir Rao, global director for telco solutions at AWS, said: “The shift to autonomous network operations is ultimately about speed and step-change efficiency. Speed to detect, speed to resolve, speed to monetize. Achieving step-change cost efficiency is critical for customers to unlock agentic value in the AI era. Nokia’s decision to optimize its full operational stack on AWS means operators can take advantage of elastic scalability, purpose-built AI and ML services, and the most extensive global infrastructure footprint for wherever their networks operate.”
Nokia has a good guide on its autonomous networks vision here.
