Ericsson vs Nokia RAN Vision: Both Cannot Be Right
The Compute Architecture That Will Decide Margins, Market Share, and 6G Leadership
AI has forced a fundamental question inside the Radio Access Network: what kind of compute should power 6G? For decades, the answer was obvious: Purpose-built silicon optimized for deterministic signal processing wins on power, cost, and reliability. Now that assumption is under direct scrutiny. NVIDIA class-accelerated compute is entering the baseband conversation, and two European vendors with more than a century of survival behind them are taking opposite paths.
Ericsson is doubling down on custom ASIC and energy-efficient Layer 1 acceleration. Nokia is aligning its future baseband roadmap with Nvidia accelerated platforms and positioning RAN as a distributed AI compute fabric. This is not a minor branding disagreement; RAN accounts for roughly 60% of Ericsson's revenue and about 40% of Nokia's. A wrong architectural bet in the 6G cycle could be fatal. But also, the final RAN Vision will also shape margins, market share, and long-term relevance.