Huawei AUTINOps: 20+ AI Agents Target Ericsson, Nokia

TL;DR

• Huawei unveiled AUTINOps at MWC 2026 — an AI-native network operations platform with over 20 autonomous agents that hit 80% risk identification accuracy and 90% diagnosis accuracy.
• The system runs predictive T-1 prevention and T-0 active response across multi-vendor networks using End-to-End Digital Twin architecture and the EDNS 2.0 agentic model.
• Includes Agent Studio toolchain for custom development and specialized agents like MBB Fault Agent for autonomous fault resolution.
• Targets AN L4 intelligent operations — shifting telcos from reactive firefighting to predictive ‘expert + digital employees’ model.

Huawei Ships AI Agents for Telco Operations

Huawei launched AUTINOps at Mobile World Congress 2026 in Barcelona, pitching it as the industry’s first AI-native intelligent operations solution. The platform deploys over 20 specialized AI agents across network operations — including an MBB Fault Agent designed to resolve faults autonomously without human intervention. This isn’t monitoring software with a chatbot bolted on. It’s agentic infrastructure.

The system centers on an End-to-End Digital Twin Network and the EDNS 2.0 agentic model. Those twins mirror live network behavior, letting agents run simulations before touching production systems. The EDNS 2.0 model powers decision-making across the agent fleet — pattern recognition, fault prediction, remediation planning.

Huawei claims the platform hits over 80% accuracy identifying risks before they cascade and over 90% accuracy diagnosing faults after they surface. Those numbers matter in telco operations, where a single misconfigured router can black out a metro area. The platform operates in two temporal modes: T-1 predictive prevention catches problems before they ignite, and T-0 active response handles fires the moment they start.

Lucas Lu, Vice President of Huawei GTS, stated: “A new era of intelligent operations has arrived. The upgraded AUTINOps solution will help operators address challenges, create greater value, redefine the operations paradigm, and accelerate progress toward AN L4.” AN L4 refers to Level 4 autonomous networking — networks that self-optimize and self-heal with minimal human oversight.

Huawei also shipped Agent Studio, a toolchain for building custom agents. Telcos can train new agents on proprietary network topologies or vendor-specific quirks. That flexibility targets multi-vendor environments — the messy reality where Nokia radios talk to Ericsson core switches through Huawei transport layers.

Why AUTINOps Threatens Ericsson and Nokia’s Operations Stack

This launch puts Ericsson and Nokia in an awkward spot. Both companies sell network management platforms — Ericsson’s Intelligent Automation Platform and Nokia’s NetGuard Cybersecurity Dome — but neither has shipped an agentic architecture at this scale. Huawei just leapfrogged them with a platform that doesn’t just monitor networks, it operates them.

The 80% risk identification and 90% diagnosis accuracy benchmarks set a new performance bar. If those numbers hold in production deployments, operators will pressure other vendors to match them. And if Ericsson and Nokia can’t deliver comparable agent-driven automation within the next 18 months, they risk losing operations contracts even in markets where they supply the radios.

The multi-vendor interoperability angle is a direct shot at proprietary lock-in. Ericsson and Nokia both prefer selling end-to-end stacks — radios, core, transport, management — because integration friction keeps customers sticky. Huawei’s Agent Studio flips that model. Build agents that work across any vendor’s gear, and suddenly the telco controls the intelligence layer while commoditizing the hardware beneath it.

I think this is Huawei playing the long game on 6G. If telcos adopt AUTINOps now for 5G operations, they’ll default to Huawei’s agentic architecture when 6G rollouts accelerate in 2028-2030. That’s a land grab disguised as an operations upgrade.

The platform also shifts where value concentrates in the telco stack. Traditionally, equipment vendors captured margin by selling proprietary hardware and charging maintenance contracts. But if AI agents handle the maintenance — predicting failures, rerouting traffic, patching vulnerabilities — then the operations software becomes more valuable than the boxes it manages. It’s like watching the network become a commodity while the brain running it becomes the product.

Think of it like this: AUTINOps is to network operations what autopilot is to driving. You still need the car — the radios, the fiber, the switches — but the system deciding when to brake, when to accelerate, when to reroute around traffic? That’s where the intelligence lives. And whoever controls that layer controls the economics of running the network.

There’s a talent angle here too. If agents genuinely hit 90% diagnosis accuracy and resolve faults autonomously, telcos won’t need armies of network engineers babysitting dashboards at 3 a.m. They’ll need smaller teams orchestrating agents — writing policies, training models, auditing decisions. That’s a different skill set. Less CCNA certification, more prompt engineering and reinforcement learning.

The T-1 predictive prevention mode is the real prize. Reactive operations — waiting for alarms, then scrambling to fix them — burns money and customer goodwill. Predictive operations — catching the misconfiguration before it crashes the cell tower — saves both. If Huawei’s agents can reliably operate one step ahead of failures, that’s a structural cost advantage for any operator running the platform.

Huawei’s AI Push Amid Western Market Lockout

Huawei unveiled AUTINOps at the 4th Intelligent Operations Forum during MWC 2026, building on years of AI-focused telco tools. The company has been locked out of most Western markets since 2019, when U.S. sanctions blocked American firms from selling components to Huawei and pressured allies to rip out its equipment. That geopolitical squeeze forced Huawei to double down on software and AI — areas where it doesn’t need Western chips or supply chains.

The multi-vendor interoperability focus is partly a survival strategy. Huawei can’t sell radios in London or New York, but it can sell software that manages Ericsson and Nokia radios. If AUTINOps works across any vendor’s gear, it slips through the hardware ban and competes on operational efficiency instead.

This also positions Huawei against Western cloud giants like Microsoft and AWS, both of which are pushing AI-driven infrastructure management tools. Microsoft’s Copilot for Azure and AWS’s DevOps Guru target enterprise IT operations — not telco networks specifically, but the competitive dynamic is similar. Whoever builds the best agentic operations layer captures recurring revenue as infrastructure scales.

The AN L4 target — Level 4 autonomous networking — signals where Huawei thinks the industry is heading. L4 means networks that self-configure, self-optimize, and self-heal without human intervention except in edge cases. It’s the same autonomy ladder that self-driving cars climb, applied to telecom infrastructure. And Huawei is betting that operators will pay a premium for platforms that get them there faster.

What Telcos Will Watch in AUTINOps Deployments

The first thing operators will test is whether those accuracy numbers — 80% risk identification, 90% diagnosis — hold outside controlled demos. Trade show benchmarks often crumble when they hit production networks with legacy gear, unpredictable traffic patterns, and a decade of technical debt. If AUTINOps can sustain those metrics across messy multi-vendor deployments, it’s a genuine breakthrough. If the numbers drop to 60% and 70% in the field, it’s vaporware with good marketing.

Agent Studio’s custom development toolchain will determine adoption speed. If telcos can train new agents in weeks without vendor support, the platform becomes a competitive weapon. If custom agent development requires Huawei consultants and six-month integration projects, it’s just another expensive middleware layer. The gap between those two outcomes decides whether AUTINOps becomes infrastructure or shelfware.

The geopolitical response matters too. Western regulators could block AUTINOps on national security grounds — arguing that AI agents managing critical infrastructure create backdoor risks. That playbook worked to freeze Huawei’s hardware sales. Whether it works against software that runs on competitors’ equipment is an open question. Operators in Europe, Asia, and Latin America will calculate whether the operational gains outweigh the regulatory risk.

Source: Huawei