Nokia and AWS expand deal to run autonomous network fabric in cloud
Nokia and Amazon Web Services have announced an expanded collaboration to host Nokia's Autonomous Network Fabric on AWS infrastructure, aiming to give telecommunications operators a fully cloud-based operational stack capable of what the partners describe as Level 4 autonomy. Availability is planned for later in 2026, building on an existing set of Nokia digital operations applications already running on the platform.
The Autonomous Network Fabric combines intent-based service orchestration, AI-powered observability, digital twin simulation and unified data management into a single platform. Nokia says the system can handle multi-domain, multi-vendor environments through closed-loop automation, translating business intent into network actions without manual intervention. Reported operator results attributed to the wider Nokia autonomous networks portfolio include automation rates above 90%, service delivery within four hours, and service interruption periods of under one minute per year, along with up to 85% reductions in network-slice rollout time. Nokia did not name the operators behind these figures.
The deal
Running on AWS, the Fabric gains access to Amazon Bedrock and Amazon SageMaker for model serving and ML workflow management, alongside the hyperscaler's global availability footprint. Nokia says it has engineered the deployment to reduce compute and storage requirements compared with traditional on-premises installations, though no specific efficiency ratios were disclosed.
Oguz Sunay, CTO for AI and Autonomous Networks at Nokia, said the company was converging intent-based networking, agentic AI, and cloud-native architecture to give operators both scale and governance. Amir Rao, global director for Telco Solutions at AWS, framed the partnership as compressing years of operational transformation into months, citing speed of fault detection, resolution, and service monetisation as the primary value drivers.
The announcement extends a string of joint milestones. At MWC Barcelona in March 2026, Nokia and AWS demonstrated what they described as the first agentic AI-powered network slicing in collaboration with du and Orange. In February, Belgian operator Citymesh launched what the companies called the world's first commercial mobile service on a 5G Core SaaS platform powered by the same partnership. The consistency of that "world's first" framing across multiple announcements warrants editorial scrutiny; independent verification has not been confirmed.
Market context
Autonomous network operations sit at the intersection of two large spending cycles: telecoms operators' multi-year effort to reduce operational expenditure and the hyperscaler push to capture workloads from purpose-built telecom infrastructure. Nokia competes in this space with Ericsson, which has its own cloud-native OSS portfolio and a partnership with Google Cloud, and with Amdocs and IBM, both of which offer operator-facing AI and orchestration platforms with hyperscaler integrations.
The Level 4 autonomy target referenced in the release is drawn from the TM Forum's Autonomous Networks framework, which defines a five-level maturity scale from fully manual (Level 0) to fully self-optimising (Level 5). Most commercial deployments today sit at Level 2 or Level 3. The credibility of a Level 4 claim will rest on independently audited closed-loop performance data, which neither Nokia nor AWS has published to date.
Regulatory dynamics also bear watching. The EU's European Electronic Communications Code and the UK's new telecoms security framework under the Telecommunications (Security) Act both impose obligations on operators around network resilience and supply-chain oversight. Deploying critical network management software on a US hyperscaler raises data-sovereignty questions that operators in regulated markets will need to resolve before full migration, particularly given ongoing EU-US data-transfer discussions following the current adequacy framework.