Nokia and Databricks aimed to prove that a common data platform could function identically across different cloud environments or on-premise infrastructure, enabling consistent AI deployment without code rewrites .

The Real-Time Performance Management Test

Engineering teams from both companies focused on a demanding use case: real-time performance management. They simulated analytics ingestion at a scale matching tier-1 operator requirements in the cloud. The work produced several specific technical outcomes :

• Cross-platform data pipelines without code changes: Data pipelines were written once and deployed identically on Databricks and on an open-source stack (Apache Flink, Kafka, Iceberg), supporting real-time streaming, batch processing, and query-time data products .
• Vendor-neutral data logic: Nokia engineers built transformation logic in Python using platform-independent expressions, keeping the core logic separate from platform-specific connectors so the same workflows could be reused across environments .
• Automated deployment via a custom compiler: The teams validated a custom compiler that automatically adapts workflows at deployment time, translating abstract logic into native formats (e.g., Delta Live Tables for Databricks, Flink SQL for open-source) and adding platform-specific connectors without manual rework .
• Data fabric for the agentic world: The platform supports query-time data products that compute derived metrics on read instead of duplicating data; zero-copy sharing for lightweight cross-domain data consumption; and a mechanism to selectively feed upper temporal layers in the cloud where AI agents run retrospective tasks like root-cause analysis .

The Intelligent Data Fabric Agent

Beyond the infrastructure layer, the PoC demonstrated an AI-powered intelligent data fabric agent that streamlines operations. Using simple natural language prompts, the agent can generate new data products, request human validation, and deploy the pipeline automatically — accelerating innovation with less manual effort .

In an agentic world, the same mechanism enables agent-to-agent communication: other AI agents can request dynamic data products on demand by interacting directly with the data fabric agent .

Why This Matters: Nokia's Broader Autonomous Networks Push

The Databricks PoC is one part of a larger strategy. On the same day, Nokia announced an expanded collaboration with AWS to run its Autonomous Networks Fabric on AWS, with availability expected later in 2026 .

Key elements of that expansion:

• Level 4 network autonomy: Operators gain access to advanced AI and cloud services needed for highly autonomous network operations .
• Four core capabilities unified: The Fabric integrates Unified Data Management across domains, Agentic AI for service operations, Digital Twin simulations for proactive impact assessment, and Intent-Based Networking for automated closed-loop actions .
• AI services underpinning it all: The solutions will leverage AWS AI/ML services including Amazon Bedrock and Amazon SageMaker for elastic scalability and broad model choice .
• Customer results already reported: Operators using Nokia's autonomous portfolio have reported automation rates exceeding 90%, service delivery times of four hours or less, up to 85% reduction in slice rollout time, and up to 50% fewer customer-impacting incidents .

Nokia Also Upgraded Its Agentic AI Portfolio

At DTW Ignite, Nokia unveiled its Autonomous Networks Agent Library — a suite of pre-built AI agents for security, assurance, and service operations. The library covers zero-day attack identification, anomaly reasoning, event triage, service-order troubleshooting, and multi-agent coordination. Nokia reports productivity gains ranging from 60% to 80% compared to traditional security operations .

Other agentic upgrades announced include a refresh of the Autonomous Networks Suite with on-premise deployment options, VoLTE service quality optimization, and enhanced RAN subscriber experience; updates to the MantaRay service management and orchestration platform with non-real-time RAN Intelligent Controller functionality; and an authentication framework added to WaveSuite for optical networks .

What Comes Next

Nokia and Databricks plan to continue their collaboration around enhancing autonomous network capabilities, helping operators transition to a future where AI applications increasingly access, correlate, and act on large-scale network data in real time .

Nokia and AWS are committed to a collaborative innovation agenda that combines Nokia's telecom-trained AI models and domain expertise with AWS's AI services to deliver increasingly autonomous network operations, and they plan to go-to-market together to help operators evolve their operational stacks and unlock new revenue streams .

Nokia and Databricks aimed to prove that a common data platform could function identically across different cloud environments or on-premise infrastructure, enabling consistent AI deployment without code rewrites .

The Real-Time Performance Management Test

Engineering teams from both companies focused on a demanding use case: real-time performance management. They simulated analytics ingestion at a scale matching tier-1 operator requirements in the cloud. The work produced several specific technical outcomes :

• Cross-platform data pipelines without code changes: Data pipelines were written once and deployed identically on Databricks and on an open-source stack (Apache Flink, Kafka, Iceberg), supporting real-time streaming, batch processing, and query-time data products .
• Vendor-neutral data logic: Nokia engineers built transformation logic in Python using platform-independent expressions, keeping the core logic separate from platform-specific connectors so the same workflows could be reused across environments .
• Automated deployment via a custom compiler: The teams validated a custom compiler that automatically adapts workflows at deployment time, translating abstract logic into native formats (e.g., Delta Live Tables for Databricks, Flink SQL for open-source) and adding platform-specific connectors without manual rework .
• Data fabric for the agentic world: The platform supports query-time data products that compute derived metrics on read instead of duplicating data; zero-copy sharing for lightweight cross-domain data consumption; and a mechanism to selectively feed upper temporal layers in the cloud where AI agents run retrospective tasks like root-cause analysis .

The Intelligent Data Fabric Agent

Beyond the infrastructure layer, the PoC demonstrated an AI-powered intelligent data fabric agent that streamlines operations. Using simple natural language prompts, the agent can generate new data products, request human validation, and deploy the pipeline automatically — accelerating innovation with less manual effort .

In an agentic world, the same mechanism enables agent-to-agent communication: other AI agents can request dynamic data products on demand by interacting directly with the data fabric agent .

Why This Matters: Nokia's Broader Autonomous Networks Push

The Databricks PoC is one part of a larger strategy. On the same day, Nokia announced an expanded collaboration with AWS to run its Autonomous Networks Fabric on AWS, with availability expected later in 2026 .

Key elements of that expansion:

• Level 4 network autonomy: Operators gain access to advanced AI and cloud services needed for highly autonomous network operations .
• Four core capabilities unified: The Fabric integrates Unified Data Management across domains, Agentic AI for service operations, Digital Twin simulations for proactive impact assessment, and Intent-Based Networking for automated closed-loop actions .
• AI services underpinning it all: The solutions will leverage AWS AI/ML services including Amazon Bedrock and Amazon SageMaker for elastic scalability and broad model choice .
• Customer results already reported: Operators using Nokia's autonomous portfolio have reported automation rates exceeding 90%, service delivery times of four hours or less, up to 85% reduction in slice rollout time, and up to 50% fewer customer-impacting incidents .

Nokia Also Upgraded Its Agentic AI Portfolio

At DTW Ignite, Nokia unveiled its Autonomous Networks Agent Library — a suite of pre-built AI agents for security, assurance, and service operations. The library covers zero-day attack identification, anomaly reasoning, event triage, service-order troubleshooting, and multi-agent coordination. Nokia reports productivity gains ranging from 60% to 80% compared to traditional security operations .

Other agentic upgrades announced include a refresh of the Autonomous Networks Suite with on-premise deployment options, VoLTE service quality optimization, and enhanced RAN subscriber experience; updates to the MantaRay service management and orchestration platform with non-real-time RAN Intelligent Controller functionality; and an authentication framework added to WaveSuite for optical networks .

What Comes Next

Nokia and Databricks plan to continue their collaboration around enhancing autonomous network capabilities, helping operators transition to a future where AI applications increasingly access, correlate, and act on large-scale network data in real time .

Nokia and AWS are committed to a collaborative innovation agenda that combines Nokia's telecom-trained AI models and domain expertise with AWS's AI services to deliver increasingly autonomous network operations, and they plan to go-to-market together to help operators evolve their operational stacks and unlock new revenue streams .