Nokia launched its AI-RAN platform on July 15, 2026, built on its anyRAN software and NVIDIA's Aerial system. The company calls it the industry's first GPU-accelerated AI-RAN platform. The move is meant to revive Nokia's radio business, which CEO Justin Hotard recently described as not delivering acceptable returns. But the platform's reliance on NVIDIA GPUs creates a vendor lock-in that rivals like Ericsson are already exploiting with silicon-independent alternatives.
Why Nokia went all-in on NVIDIA
Nokia's platform offers three deployment options: a GPU-powered plug-in card for its AirScale sites, a standalone AI-RAN node, and a cloud-server build through partners. Nokia claims the platform has shown more than 20% spectral efficiency gains so far, with targets of 50% by 2027 and more than 100% by 2028. The company plans pilots at the end of 2026 and commercial availability in 2027.
📊 Market Data Snapshot
The deep ties to NVIDIA are no accident. NVIDIA invested $1 billion in Nokia in October 2025 for roughly a 3% stake. Nokia's CTO Pallavi Mahajan acknowledged that at least some Layer 1 software is bound to NVIDIA hardware. That means operators buying into Nokia's vision are also buying into NVIDIA's GPU roadmap.
Ericsson's counterpunch
Ericsson began selling a commercial AI-in-RAN software subscription in June 2026. It claims up to 20% higher downlink throughput and up to 10% better spectral efficiency across more than 15 live deployments. Crucially, Ericsson's AI runs on existing baseband silicon — no GPU required. That makes it silicon-independent, a direct contrast to Nokia's approach.
Omdia analyst Rémy Pascal, quoted in Nokia's announcement, puts the cumulative AI-RAN opportunity above $200 billion by 2030. But the question is which architecture captures that value.
The GPU lock-in debate
Nokia's GPU dependency mirrors the early days of crypto mining, when GPUs dominated before ASICs took over by offering better efficiency and independence. Ericsson's silicon-independent AI-in-RAN could follow a similar path. If operators prefer flexibility over proprietary acceleration, Ericsson's approach may win the long game.
For crypto traders, the immediate effect is indirect. Nokia's platform will increase demand for NVIDIA's H100 and B200 GPUs, potentially tightening supply for both AI compute and crypto mining. That could boost GPU-based tokens like RNDR and AKT as their underlying hardware becomes scarcer. But Ericsson's alternative could limit that demand if it gains operator adoption.
The platform also validates software-defined radio and virtualized RAN — core technologies behind decentralized wireless projects like Helium and Pollen Mobile. Nokia's centralized approach could either compete with or legitimize those networks. Either way, the mainstream telecom industry is moving toward the same architecture.
Nokia plans to start pilots by the end of 2026, with commercial availability in 2027. Ericsson's AI-in-RAN is already live in more than 15 deployments. The race between GPU-dependent and silicon-independent AI-RAN is just beginning — and crypto investors should watch which standard wins operator buy-in.
