Nokia Bets Hardware-First AI-RAN Can Beat Huawei

TL;DR

• Nokia unveiled Doksuri Remote Radio Heads on March 1, 2026, targeting energy-efficient AI-ready 5G infrastructure at Mobile World Congress.
• The new AirScale portfolio additions work alongside existing Habrok Massive MIMO products to handle AI-driven network demands across TDD and FDD spectrum.
• A new Marketplace for Service Management and Orchestration (SMO) launches alongside hardware, promising cognitive radio systems that optimize operations and slash energy consumption.
• Directly challenges Huawei‘s AI-Native O&M platforms with hardware-level AI-RAN optimizations — not just software promises.

Nokia Bets on Doksuri Hardware to Win the AI-RAN Infrastructure Race

Nokia announced expansions to its AirScale portfolio on March 1, 2026, introducing Doksuri Remote Radio Heads designed to deliver energy efficiency and capacity gains for AI-ready 5G networks. The Finnish telecom giant positioned the launch as a direct response to surging data demands from AI workloads — the kind that choke networks built for yesterday’s traffic patterns.

The Doksuri RRHs complement Nokia’s existing Habrok Massive MIMO products, creating a unified hardware stack that spans both TDD and FDD spectrum bands. This matters because operators can’t just throw more radios at the AI problem — they need infrastructure that handles compute-intensive applications without torching their power budgets or requiring truck rolls every quarter.

Nokia also launched a Marketplace for Service Management and Orchestration (SMO) alongside the hardware. The company is demoing cognitive radio systems at MWC26 in Hall 3B20 that reportedly optimize network operations and energy consumption in real time — the kind of software-defined intelligence that separates modern infrastructure from legacy gear.

According to Nokia’s announcement, “Artificial Intelligence can have a definite and concrete impact when applied to Radio Access Networks, and this kind of radio innovation allows the radio hardware to support future enhancements to capacity and reliability performance in the network.” Translation: the hardware ships ready for AI features that don’t exist yet, betting operators will want headroom as models get hungrier.

Why Doksuri Signals Nokia’s Play for AI Workload Dominance

Here’s what Nokia is actually doing — it’s building infrastructure that treats AI traffic as the baseline, not the exception. And that’s a fundamental architectural shift.

Traditional radio heads optimize for mobile broadband and voice. Doksuri optimizes for the kind of bursty, asymmetric, latency-sensitive traffic that AI inference and training generate. That means operators deploying this gear can support edge AI applications — think autonomous vehicles, real-time video analytics, industrial automation — without retrofitting their RAN six months later when capacity chokes.

The energy efficiency angle isn’t marketing fluff. Telecom networks reportedly consume around 2-3% of global electricity, and that percentage climbs as data traffic explodes. If Doksuri can squeeze more bits per watt out of the same cell sites, operators save operating costs and hit sustainability targets simultaneously. That’s the kind of dual incentive that accelerates deployment.

But here’s the competitive wrinkle — Huawei has been pushing AI-Native O&M and DTN platforms hard, promising intelligent operations through software. Nokia is countering with hardware-level optimizations baked into the radios themselves. It’s the difference between a smart thermostat and a house designed for passive cooling. One adapts to conditions; the other changes the conditions.

I think Nokia’s betting that operators trust silicon over promises. Huawei’s software-first approach requires operators to rip and replace management layers, integrate new APIs, and trust black-box algorithms. Doksuri ships as a radio that slots into existing AirScale deployments — evolutionary, not revolutionary. For risk-averse telcos, that’s the easier sell.

The SMO Marketplace is where things get interesting long-term. Nokia is essentially creating an app store for network intelligence — third-party developers can build cognitive radio applications that run on top of the Doksuri hardware. Think of it like the iPhone’s App Store, but for spectrum optimization algorithms. If Nokia can cultivate an ecosystem of AI-RAN applications, it locks operators into its hardware platform the same way iOS locks users into Apple silicon.

Does this actually move the needle on AI adoption in telecom? Maybe. The bottleneck isn’t hardware capability — it’s operator willingness to deploy unproven technology at scale. Nokia is making the hardware argument easy (drop-in compatibility, energy savings, future-proof capacity). The harder sell is convincing CIOs that AI workloads will generate enough revenue to justify the refresh cycle.

How Doksuri Fits Nokia’s Broader AirScale Evolution

Nokia’s AirScale portfolio has been evolving to meet AI-era network demands for the past few product cycles. The Habrok Massive MIMO products launched earlier to handle high-capacity TDD deployments — think dense urban environments where spectrum is expensive and capacity is king.

Doksuri fills the gap for operators who need AI-ready infrastructure across both TDD and FDD bands. FDD spectrum is where a lot of legacy LTE and 5G coverage lives, especially in rural and suburban markets. If you’re a Tier 1 operator, you can’t just upgrade your urban core and ignore the 70% of your network running FDD — Doksuri gives you a unified AI-RAN strategy across the entire footprint.

The shift to software-defined cognitive radios is the subtext here. MWC26 demos showcase radios that adjust parameters — power levels, beamforming patterns, modulation schemes — based on real-time traffic patterns and energy costs. That’s a fundamentally different architecture than static radios configured once during deployment and left alone until the next maintenance window.

This also positions Nokia for the next phase of network disaggregation. As operators adopt Open RAN principles, the radio units need to be smarter and more autonomous because they’re no longer tightly coupled to proprietary baseband processing. Doksuri’s cognitive capabilities make it a better fit for multi-vendor Open RAN deployments than dumb radios that require constant babysitting from the orchestration layer.

What Telcos Should Monitor as Doksuri Deployments Scale

First, watch how quickly Tier 1 operators announce commercial Doksuri deployments. Nokia can demo cognitive radios in a booth all day, but the real test is whether Verizon, Deutsche Telekom, or China Mobile commit capex to large-scale rollouts. If we don’t see commercial announcements by mid-2026, that signals operators are still skeptical about AI-RAN ROI.

Second, track energy consumption benchmarks in production networks. Nokia is making efficiency claims, but the proof lives in real-world deployments under variable load conditions. Operators will publish case studies if Doksuri actually slashes their power bills — silence on this front six months out means the savings didn’t materialize at scale.

Third, monitor the SMO Marketplace adoption. If third-party developers start building cognitive radio applications on Nokia’s platform, that creates ecosystem lock-in and validates the app store model for telecom infrastructure. If the marketplace stays empty or dominated by Nokia’s own apps, the strategy flopped. Ecosystem health is the leading indicator for whether this becomes a platform or just another product.

FAQ