Huawei’s Lobbying Win and Open RAN’s Nascent Reality: What the Connect Europe Report Really Says

Every so often a document lands that is more useful for what it admits than for what it argues. The Assembly Research report commissioned by Connect Europe — dated July but only surfacing publicly this week — is one of those. It was written to support Europe’s telcos in their long-running argument that they cannot afford to remove Huawei, and in the process it delivers the most candid assessment of open RAN’s commercial failure in the EU that we have seen from an industry-funded source. For engineers who have spent the last eight years building, testing, and integrating disaggregated RAN, it is worth reading carefully, because the technical community’s response to it will shape procurement decisions for the rest of the decade.

What the Report Actually Claims

The headline finding, as reported by Light Reading, is blunt: “In the EU, the open RAN market remains at a nascent stage, with most operators relying primarily on an end-to-end solution from a single vendor.” That sentence is doing a lot of work. It is not a claim about technology immaturity — it is a claim about deployment reality. The O-RAN Alliance has been publishing specifications for more than eight years. The interfaces exist. The testbeds exist. The rApps and xApps exist. What does not exist, according to the report, is a European market where operators are buying multi-vendor disaggregated RAN at scale.

The supporting data point is the market share picture from Omdia, a Light Reading sister company. In the first half of 2026, Huawei, Ericsson, Nokia, ZTE, and Samsung collectively served roughly 95% of the global RAN market. That is the number that should sit at the top of any open RAN strategy deck. The five incumbent vendors have not been displaced. If anything, the concentration is intact, and the open RAN ecosystem’s promise of a sixth, seventh, and eighth supplier has not materialized into meaningful share.

Light Reading’s own framing was harsher than the report’s. Where Assembly Research wrote “nascent,” the article suggested “stillborn would be a more accurate description.” That is editorializing, but it captures the mood. The report is being read in Brussels and in operator boardrooms as evidence that the vendor-diversity argument for open RAN has not delivered, and therefore that Huawei’s continued presence in European networks is a practical necessity rather than a policy choice.

The Architecture Question Nobody Wants to Answer

From an engineering standpoint, the interesting thing about the report is the gap between what open RAN specifies and what operators actually deploy. The O-RAN Alliance’s architecture disaggregates the base station into a Central Unit (CU), Distributed Unit (DU), and Radio Unit (RU), connected by open fronthaul — typically the 7.2x split — with the RIC providing near-real-time and non-real-time control loops for xApps and rApps. In theory, this lets an operator mix an RU from Vendor A with a DU from Vendor B and a CU from Vendor C, and run third-party intelligence on top.

In practice, the integration burden is enormous. Fronthaul interoperability at the 7.2x split is a well-known pain point: the CUS-plane and U-plane message formats are specified, but timing, jitter tolerance, and beamforming control implementations vary enough between vendors that multi-vendor fronthaul still requires significant engineering per pair. The RIC ecosystem, meanwhile, has matured faster on the non-real-time side, where rApps can run on standard cloud infrastructure, than on the near-real-time side, where xApps need deterministic sub-10ms control loops and tight coupling to the DU scheduler. That coupling is exactly where vendors guard their differentiation.

The result is that “open RAN” in many European deployments means something narrower than the full disaggregated vision: often a single vendor’s cloud-native RAN software running on commodity hardware, with open interfaces enabled for specific functions but not for the full multi-vendor stack. That is a real architectural improvement over purpose-built baseband — it enables software upgrades, cloud economics, and some degree of vendor flexibility — but it does not deliver the competitive dynamic that open RAN’s advocates promised. The report’s “end-to-end solution from a single vendor” description is consistent with that narrower reality.

Why This Matters Now: The AI-RAN Pivot

The timing of the report’s publication is not accidental. It lands in the middle of the industry’s pivot from open RAN as a disaggregation story to AI-RAN as an intelligence story. The same week, Nokia was making headlines with its commercial AI-RAN platform — built on its anyRAN software and NVIDIA’s Aerial AI-RAN platform — claiming more than 100% spectral efficiency gains by 2028 and up to 30% power efficiency improvements. ZTE was presenting its AIR MAX architecture at its Global Summit in Kuala Lumpur, with AIR RAN, AIR Net, and AIR Core components and a claimed 98%+ service identification accuracy. Radisys launched a V.AI ecosystem for AI-powered communication services. VNPT selected RADCOM for AI-native, cloud-native service assurance in Vietnam.

The strategic logic is clear. If open RAN’s multi-vendor promise has stalled, AI-RAN offers a different value proposition: not cheaper radios through competition, but better performance through intelligence. GPU-accelerated baseband, AI-driven scheduling, predictive optimization, and energy savings are benefits an operator can capture within a single-vendor stack. They do not require the multi-vendor integration that has proven so difficult. In that sense, AI-RAN is partly a response to open RAN’s commercial disappointment — a way to keep the cloud-native, software-defined architecture while deferring the disaggregation problem.

But there is a tension the report exposes. AI-RAN’s most ambitious use cases — cross-vendor RIC control, third-party xApps optimizing another vendor’s scheduler, AI-driven spectrum sharing across heterogeneous radios — depend on the very openness that has not scaled. If the market consolidates around a handful of AI-RAN platforms from Nokia, Ericsson, Samsung, and ZTE, each with its own accelerated computing stack and its own AI framework, operators may find themselves with better performance but less leverage than before. The lock-in risk does not disappear; it moves up the stack from silicon to AI models and data pipelines.

The Geopolitical Layer

The report’s immediate political purpose is to give European operators ammunition in the debate over Huawei. Deutsche Telekom, Telefónica, and others have argued for years that removing Huawei from their networks is technically and financially painful, and that there are no viable alternatives at the scale required. The Assembly Research findings support that position: if no credible open RAN challengers have emerged in eight years, then the “rip and replace” argument loses its technological foundation and becomes purely a security question.

That is a significant shift. The original case for open RAN in Europe was partly industrial policy — build a European alternative to Huawei and ZTE, reduce dependency on Chinese vendors, and create a competitive supplier ecosystem. If that ecosystem has not emerged, then the policy has failed on its own terms, and the security argument has to stand alone. The report’s language — “nascent” rather than “failed” — is a diplomatic way of acknowledging this without conceding the policy defeat outright.

For operators, the practical consequence is that the Huawei decision becomes a cost-benefit calculation rather than a strategic bet on open RAN. If multi-vendor disaggregation is not a realistic near-term option, then the cost of removing Huawei is the cost of replacing it with Ericsson, Nokia, or Samsung — a single-vendor swap, not a market transformation. That is a much more concrete and defensible number, and it is why the report is being read as a lobbying win for Huawei’s defenders.

What the Engineering Community Should Take From This

The temptation for open RAN’s technical community is to dismiss the report as vendor lobbying dressed up as research. That would be a mistake. The report’s central factual claim — that most EU operators rely on single-vendor end-to-end solutions — is consistent with what anyone who has worked on real deployments knows. The multi-vendor vision has been demonstrated in labs, trials, and a handful of commercial sites, but it has not become the default procurement model.

The more useful response is to ask why, and to be specific about the technical and commercial barriers. Fronthaul interoperability remains hard. The near-real-time RIC’s performance requirements make third-party xApps difficult to certify across vendors. The integration and testing cost of multi-vendor deployments often exceeds the savings from competitive procurement, especially for operators with limited engineering capacity. And the business case for open RAN has always depended on scale — a vibrant supplier ecosystem needs enough volume to sustain investment, and that volume has not materialized outside a few markets.

AI-RAN changes the calculus in ways that could help or hurt. On one hand, AI workloads are inherently more portable than baseband signal processing: a trained model can run on different accelerators, and the data pipelines that feed it can be standardized. If the industry converges on common AI-RAN interfaces — and the AI-RAN Alliance is working on exactly that — then the intelligence layer could become genuinely multi-vendor even if the radio layer remains consolidated. On the other hand, the tight coupling between AI inference and real-time scheduling means that the most valuable AI-RAN features will be those that vendors control end-to-end, which pulls in the opposite direction.

The report does not resolve that tension, and neither will any single document. But it does something valuable: it forces the industry to be honest about where open RAN actually stands. Eight years in, with 95% of the market still held by five vendors, the disaggregation story has not delivered. The question for the next eight years is whether AI-RAN can deliver something different — and whether operators will have the architectural leverage to capture its benefits rather than simply trading one form of lock-in for another.

Sources

• Huawei gets lobbying boost from European pals after open RAN fail
• Nokia Oyj (HLSE:NOKIA) - Stock Analysis - Simply Wall St
• ZTE Global Summit 2026 Opens in Kuala Lumpur
• Radisys Launches V.AI Ecosystem to Accelerate Telecom AI Service Innovation
• VNPT Selects RADCOM for AI-Native, Cloud-Native Service Assurance in Vietnam
• Nokia alert to risks as it puts AI agents into mobile core
• AI-RAN Market Size to Reach USD 46.69 Billion by 2035 as Open RAN