Verizon Bets on Ericsson’s Open Controller to Turn AI-RAN Into a National Reality

Here’s a question worth sitting with for a moment: when the network gets smarter, who gets to write the intelligence?

That sounds like a philosophical riddle, but for telecom operators it is a very practical, very expensive question. Do you buy your AI-powered network features the way you buy radios and antennas — as a bundled product from a single vendor? Or do you insist on a platform where your own engineers can build the clever bits themselves, and swap in someone else’s clever bits when they’re better?

On November 3, Verizon answered that question in public. The US carrier announced an expanded partnership with Ericsson that will see the Swedish vendor’s Intelligent Controller — the EIC, in industry shorthand — rolled out across the United States. The headline isn’t just that AI is coming to Verizon’s radio network. It’s that Verizon intends to write some of that AI itself, on an open platform that Ericsson says is built to work with equipment from multiple vendors.

Let’s unpack why that matters, what was actually demonstrated, and what it tells us about where the whole industry is heading.

What Verizon and Ericsson actually announced

The core of the news is the nationwide deployment of the Ericsson Intelligent Controller, which Verizon plans to begin rolling out this year. The EIC is a key piece of something called the Ericsson Intelligent Automation Platform — essentially Ericsson’s open network management and automation layer, designed to work in multivendor, multitechnology environments. It supports Ericsson’s 4G and 5G radio portfolio as well as its core networks.

If you’re new to this world, here’s the simplest way to think about it. Modern radio networks are increasingly run by small software applications — called rApps — that sit on top of a control layer and make decisions: steer traffic here, save energy there, flag this cell as behaving oddly. The controller is the stage. The rApps are the actors. Whoever controls the stage decides which plays get performed.

Verizon’s stated goal is to use the EIC as an open, standards-based platform on which it can deploy rApps it designs itself, alongside rApps from Ericsson. According to the announcement, the two companies have already run this in a live environment, demonstrating AI-driven “cell shaping” of Massive MIMO antennas — a fancy way of saying the network dynamically adjusts how its radio beams are formed — plus traffic steering, with what they describe as significant gains in throughput and efficiency.

There was also a nicely concrete engineering win. Verizon implemented a connected-mode handover method from LTE to standalone 5G for Fixed Wireless Access devices, which eliminated the need for manual intervention when handovers failed. That’s the unglamorous plumbing work that makes a network feel reliable, and it’s exactly the kind of thing automation is supposed to fix.

Matt Butz, Verizon’s Vice President of Network Planning, framed it around performance across mobility and Fixed Wireless Access, and around preparation for 6G. The company also plans to test and deploy new rApps — including ones named Cell Anomaly Detector, Root Cause Explainer and Anomaly General Optimizer — to improve performance and operational efficiency.

Notice the naming. These aren’t marketing features. They’re the network equivalent of a doctor’s diagnostic toolkit: spot the anomaly, explain the root cause, fix it automatically. That’s the daily grind of running a large radio network, and it’s where AI can quietly save real money.

Why the controller, and not just the radio, is the story

For years, the AI-in-networks conversation has centered on the radio itself — putting smarter algorithms into base stations so they can adapt in real time. That’s important, and it’s happening. But there’s a second, less visible battle happening one level up, at the management and automation layer.

That’s where the controller lives. And that’s where the question of openness gets decided.

Here’s the tension in plain terms. A vendor that supplies both the radios and the controller, and also writes all the rApps, offers a tidy, well-integrated package. It will probably work smoothly. But the operator becomes dependent on that vendor for every future improvement, and the pace of innovation is set by the vendor’s roadmap, not the operator’s needs.

An open controller changes the math. If the platform genuinely supports rApps from multiple sources — including the operator’s own software teams — then the operator can prioritize what matters to it, bring in specialist startups, and keep vendors honest on price and speed.

Verizon’s language leans hard into this. The announcement describes the EIC as an “open standards-based platform” and emphasizes Verizon’s ability to self-develop rApps. That’s a deliberate signal. It says: we want the intelligence layer to be programmable by us.

Whether the reality matches the rhetoric is the fair question every operator should ask. Openness exists on a spectrum, and “open” platforms often come with practical constraints. But the direction of travel is unmistakable, and Verizon is putting its flagship network behind it.

The market numbers behind the move

This announcement didn’t land in a vacuum. It arrived in the same 48-hour window as a wave of research and news pointing the same direction.

SNS Telecom & IT published findings projecting that annual spending on AI-enabled RAN automation and optimisation software and services will grow at roughly 9% a year, reaching about US$1.1 billion by the end of 2030. That figure covers the deeper integration of AI and machine learning into radio automation — what the firm calls “AI-for-RAN.”

But here’s the more striking number. In a separate piece, SNS projected that global spending specifically on SMO, RIC and d/x/rApps — the open automation components — will grow at a compound annual rate of approximately 40% between 2026 and 2030, with the Open RAN automation market representing more than US$660 million in annual investment by 2030.

That gap between 9% and 40% is the whole story in two numbers. The overall automation market grows modestly. The open, standards-based slice of it grows explosively. Operators, in other words, aren’t just buying more automation — they’re buying it in a different shape.

SNS also noted why. Legacy self-organizing network technology — the older, proprietary approach to automatic network tuning — has limitations around multi-vendor interoperability, scalability and latency. And operators are chasing something called Level 4 autonomy for high-impact scenarios like energy savings, traffic optimization, fault management and service assurance by 2030. Level 5, fully autonomous, remains a distant goal, but Level 4 — systems that can handle complex tasks without human intervention in defined situations — is the near-term prize.

That’s precisely what Verizon is building toward.

The quieter subplot: where does the intelligence live?

There’s an architectural debate running underneath all of this, and it’s worth understanding because it will shape the next five years.

The first camp favors putting AI directly into the radio access nodes — deep in the baseband, close to the antenna, where decisions happen in microseconds. Some vendors are pursuing proprietary, real-time optimisation delivered as subscription software upgrades to existing radio platforms. It’s fast, it’s integrated, and it keeps the intelligence close to the physics.

SNS’s research highlights growing interest in a newly proposed interface called E3, which would give applications a standardized way to talk directly to Open RAN baseband nodes for closed-loop control beyond the 10-millisecond latency limits of the near-real-time controller. That would open the door to lower-layer performance optimization, geolocation, integrated sensing, spectrum sharing and interference cancellation in 5G and future 6G networks.

The second camp favors the controller layer — the rApp ecosystem — where intelligence is more portable, more multi-vendor, and more open to third-party innovation. That’s the Verizon-Ericsson bet.

The honest answer is that both will matter. Real-time radio tricks need to live in the radio. Network-wide optimization needs a broader view. The interesting question is where the boundary settles, and who controls each side of it. Verizon’s move is a vote for keeping a meaningful chunk of that boundary on an open platform.

What this means for Open RAN

Open RAN, at its heart, is a simple idea: break the radio access network into pieces connected by standardized interfaces, so operators can mix equipment from different suppliers rather than buying a single-vendor bundle.

For a long time, the critique was that Open RAN was a nice idea waiting for a business case. The automation layer is increasingly that business case. If the value of the network shifts from the hardware to the software that runs it, then whoever owns an open, programmable software layer owns the future.

That’s why the Verizon announcement matters beyond one carrier. It’s a large, high-profile operator saying: we will run AI-driven automation on an open controller, and we will write our own apps for it. If that works at national scale, it becomes a template.

It also fits a broader pattern visible in the same news cycle. In the UK, Virgin Media O2 ran what it described as the country’s first live trial of AI-RAN with Nokia at a busy London site — Paddington station, which sees on average around 191,499 passenger journeys a day — and reported a 20% boost in spectral efficiency, meaning more data carried over the same spectrum. O2’s chief technology officer, Jeanie York, framed it around smarter use of existing spectrum and infrastructure. Nokia’s AI-RAN efforts have accelerated since Nvidia invested $1 billion in the company in October.

Different operator, different vendor, same underlying thesis: the intelligence layer is where the next round of gains comes from.

The geopolitical backdrop

One more thread worth pulling. Also this week, the Quad nations — India, the US, Japan and Australia — pledged over $85 million to expand trusted AI across the Indo-Pacific, and explicitly named Open RAN as an area of cooperation. The ministers agreed to identify priority countries for cooperation in advanced communications, including 5G, 6G and Open Radio Access Networks, and to explore AI-enabled RAN.

That’s a reminder that “open” in Open RAN is not only a technical property. It’s a supply-chain and security argument. Interoperable, multi-vendor networks are seen by governments as more resilient and less dependent on any single supplier. Verizon’s decision to build on an open automation platform sits comfortably inside that larger story.

The skeptic’s checklist

It would be easy to over-read a press release. So let’s be disciplined about what we actually know and what we don’t.

What we know: Verizon plans a nationwide EIC rollout beginning this year. It has demonstrated AI cell shaping and traffic steering in a live environment with reported throughput and efficiency gains. It has built its own rApp and intends to build more. It named three specific rApps it plans to test and deploy.

What we don’t know: the size of those gains in absolute terms, how much of the nationwide rollout is committed versus planned, how many third-party (non-Ericsson, non-Verizon) rApps will actually run on the platform, and how the economics are structured. “Significant gains” is a phrase, not a number.

The fair reading is that this is a serious, concrete step — a live demonstration plus a national deployment plan, not a lab experiment — while the commercial details remain to be proven. That’s typical of where this industry is right now: past the proof-of-concept stage, into the harder work of scale and economics.

The bigger picture

The RAN market itself has been through a rough patch. Reporting cited by industry observers notes that spending on radio products fell from around $45 billion in 2022 to roughly$35 billion in 2024, which explains years of layoffs at the major vendors. Against that backdrop, software and automation are the growth story everyone is chasing.

So the real contest isn’t whether AI comes to the radio network. It will. The contest is over the shape of the layer that runs it — open and programmable, or bundled and proprietary. Verizon just placed a very public bet on the former, and it did so with a vendor that has every commercial incentive to keep the platform somewhat closed.

That tension is the story to watch. If Verizon’s engineers genuinely ship their own rApps at national scale, and if third-party developers can join them, the open automation thesis gets its strongest proof point yet. If the platform turns out to be open in name more than practice, operators will learn that lesson too — and the next round of vendor negotiations will reflect it.

Either way, the question from the top of this piece now has a live answer in the field: when the network gets smarter, Verizon intends to write at least some of the intelligence itself. The industry will be watching closely to see whether the platform lets it.

Sources

• Verizon and Ericsson partner to transform AI-native Autonomous Networks (PR Newswire)
• Verizon and Ericsson team up to integrate AI into RAN automation (MarketScreener)
• SMO/RIC-based RAN automation gains momentum (RCR Wireless)
• AI-enabled RAN automation and optimisation a US$1.1 billion market, says SNS Telecom & IT (IoT Now)
• The week in AI-native telco (24-30 Oct 2026), TelecomTV
• What’s up with… SpaceX, Ericsson, Virgin Media O2, The Future of RAN (TelecomTV)
• Quad Nations Pledge Over $85 Million to Expand Trusted AI Across Indo-Pacific (Bold News)
• Ericsson vs Nokia: AI Sales +123% vs 7% Growth (Tech-Insider)