Nokia and NVIDIA target doubled spectrum capacity with GPU-based AI-RAN
Nokia has announced a commercial AI Radio Access Network (RAN) platform powered by NVIDIA's GPU architecture, targeting a 100% increase in spectral efficiency by 2028. The solution competes with Ericsson’s existing AI-based RAN implementations and is designed to support 4G, 5G, and 6G workloads via subscription-based software updates.
Key Takeaways
- Targets 50% spectral efficiency gains by 2027 and a 100% increase by 2028, effectively doubling traffic capacity on existing spectrum.
- Utilizes NVIDIA GPUs to run deep-learning-based receivers, a shift from the ASIC-based AI approach recently deployed by rival Ericsson.
- Introduces a new value-based software subscription model, delivering continuous algorithm updates without requiring physical hardware refreshes.
- T-Mobile US is the lead deployment partner, with field trials scheduled for 2026 and a commercial start targeted for 2027.
Why It Matters
This marks a strategic shift from purpose-built hardware toward software-defined, GPU-accelerated infrastructure for high-bandwidth video traffic. By offloading complex signal processing to programmable GPUs, Nokia aims to bypass the physical constraints of licensed spectrum, offering a path to serve twice the user load without new frequency auctions. For the streaming ecosystem, this promises more reliable delivery in dense urban environments where congestion currently limits peak bitrates. The primary signal to watch is the 2026 T-Mobile field trial, which will verify if Nokia's high-intensity GPU approach can deliver the promised 100% gain over Ericsson's lighter, ASIC-based AI implementation.
Additional Context
The launch solidifies a deep strategic tie-up between Nokia and NVIDIA that includes significant financial stakes. Per reports from Semicone and The Next Web in July 2026, NVIDIA acquired a 2.9% stake in Nokia through a $1 billion investment in October 2025, with shares priced at $6.01. This partnership aims to capture a portion of the AI-RAN market that research firm Omdia projects will exceed $200 billion by 2030, as operators move away from traditional hardware upgrade cycles toward continuous software-driven performance improvements. While Nokia targets aggressive long-term gains, Ericsson has already achieved a first-mover advantage with its own AI-native RAN software. Per reporting from Fierce Network in May 2026, Ericsson and T-Mobile successfully demonstrated a 10% increase in spectral efficiency and a 15% boost in downlink throughput using AI-native scheduling on a live 5G Advanced network. Unlike Nokia’s GPU-reliant architecture, Ericsson’s solution runs on existing baseband silicon, allowing for mass deployment across 15 global operators as of June 2026. T-Mobile US remains the critical testing ground for both architectures, having integrated 5G Advanced nationwide in 2025. Per official statements from May 2026, the carrier is utilizing neural networks to predict changing radio conditions in real time, aiming to improve reliability for bandwidth-intensive applications like cloud gaming and high-resolution video streaming. The competition between Nokia’s computationally intensive GPU model and Ericsson’s energy-efficient ASIC approach will likely define the 6G infrastructure roadmap over the next three years.
Read full article at techtimes.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source