Nokia and NVIDIA Sign Nearly 20 Telcos to AI-RAN Commercial Deployment Push

By Hector Herrera | October 26, 2026 | vertical: telecom | type: company-news

Nokia and NVIDIA announced this week that nearly 20 telecom operators — including T-Mobile, BT, Deutsche Telekom, Elisa, and Orange — are running or actively piloting AI-RAN: artificial intelligence embedded directly into radio access network hardware. It is the broadest coordinated commercial deployment of AI in wireless network infrastructure announced to date.

What AI-RAN Actually Does

Traditional RAN hardware — the base stations and antennas that connect mobile devices to the network — runs on purpose-built chipsets optimized for fixed signal-processing tasks. AI-RAN replaces or augments that with NVIDIA GPU-based compute that can dynamically optimize signal processing, spectrum allocation, and energy use in real time, adjusting to traffic patterns and interference conditions faster than static algorithms allow.

In controlled deployments, AI-RAN has demonstrated 30–40% energy savings and measurable improvements in network performance during peak congestion. According to the joint announcement from Nokia and NVIDIA on October 21, operators are now moving beyond pilots into commercial deployments — meaning real customer traffic is running on AI-optimized infrastructure, not test environments.

The Sovereign AI Play

The second announcement in the same press release may be more strategically significant than the AI-RAN deployment numbers. Nokia launched a sovereign agentic AI platform for telco autonomy — infrastructure that allows operators to run AI inference on-premise without routing data through a hyperscaler cloud.

This addresses a specific and growing regulatory concern: European and Asian operators face data sovereignty requirements that prevent them from sending network traffic metadata through US-based AI APIs. Nokia's on-premise platform lets an operator in Germany, France, or Japan deploy AI-driven network management and customer intelligence without requiring a data transfer agreement or cloud dependency. The platform supports agentic AI workflows — meaning it can take automated operational actions, not just generate recommendations.

Why Nearly 20 Operators Matters

One or two pilots can be attributed to vendor relationships. Nearly 20 operators across multiple continents in commercial deployments is a market signal. The AI-in-networking market is projected to grow from $8.4 billion in 2026 to $21.5 billion by 2032, according to figures cited in the announcement.

For context: the global RAN market is roughly $25 billion annually. If AI-RAN achieves meaningful penetration of new builds and upgrade cycles over the next five years, it rewrites the economics of every major network infrastructure vendor and creates a new recurring revenue stream tied to compute, not hardware sales.

Competitive Implications

Ericsson and Samsung — Nokia's primary RAN competitors — are each developing their own AI-RAN variants. Ericsson has a partnership with Qualcomm for AI-enabled RAN processing. Samsung has a deeper partnership with T-Mobile in the US on Open RAN architecture.

The Nokia-NVIDIA partnership positions Nokia differently from both: NVIDIA's GPU-based inference advantage is the primary selling point, not Nokia's baseband software alone. For operators willing to bet on NVIDIA's AI processing roadmap, the Nokia partnership delivers both the network expertise and the compute platform in a single agreement.

For NVIDIA, telecom represents a growing enterprise segment it has been building toward since its SoftBank partnership in 2022. Each additional operator running NVIDIA compute inside its RAN infrastructure creates a recurring inference workload that scales with network traffic — exactly the kind of sticky enterprise deployment that supports NVIDIA's data center revenue projections.

What to Watch

Whether Ericsson and Samsung announce competitive AI-RAN commercial deployment numbers before Mobile World Congress in February 2027, and whether US operators accelerate AI-RAN adoption to meet FCC energy efficiency targets announced earlier this year.