Nokia Shifts 6G Architecture to Nvidia GPUs, Abandoning Custom Hardware Strategy
Nokia's GPU-centric architecture locks Nvidia into 6G RAN procurement during the standard-setting phase (3GPP Release 19+), which means U.S. semiconductor export controls now carry direct telecom infrastructure consequences — any future restrictions on Nvidia's data-center GPU sales to non-allied carriers would force Nokia to redesign RAN stacks mid-development cycle.
This also creates a supply-chain vulnerability: if Nvidia cannot meet telecom-grade availability or reliability SLAs, Nokia's 6G roadmap stalls, and allied operators lose a non-Ericsson RAN option precisely when supply diversification matters most for NATO interoperability.
This move signals Nokia's acceptance that commodity GPU acceleration now meets telecom performance requirements, reducing the economic moat of custom silicon and shifting capital intensity downstream to software optimization.
For Nvidia, it locks a tier-1 RAN vendor into its ecosystem during the 6G standardization window — critical because RAN architecture decisions made now cascade into 2028-2032 deployments across allied operators. For Nokia's competitors (Ericsson, Samsung), the choice creates pressure to either match Nvidia integration or differentiate via proprietary silicon — a costly fork.
The real constraint is whether Nvidia's software stack (CUDA optimization for telecom workloads) matures fast enough; if not, Nokia's hardware abstraction becomes a liability during 6G rollout.
The article does not specify which Nvidia GPU families Nokia will use, the timeline for production deployments, or whether this applies to Nokia's entire RAN portfolio or specific use cases (e.g., edge compute vs. centralized units).
Fault Lines
Strategic intelligence, synthesized daily — with a public track record. Every call graded against what actually happened.