So, Is Open RAN Dead? An Update on AT&T’s Open RAN Rollout and the State of Open RAN

It has now been over 15 months since the US$14 billion multi-year commitment to deploy Open Radio Access Network (RAN) and have 70% of its wireless traffic going through these sites by 2026 (for a deep dive into this announcement, see ABI Insight “How AT&T’s US$14 Billion Contract with Ericsson Impacts Nokia and the Entire Open RAN Market”). Initially, only Fujitsu was named as a third-party vendor that would support this deployment, with no clear explanation for what role Fujitsu would play. Fast forward to December 2024, AT&T updated the industry with an announcement that it had signed new agreements with Fujitsu and Mavenir to develop radios specifically for crowded urban areas. This contract will involve the deployment of Open RAN small cells that can be attached to existing utility and light poles (4T4R radios).

Aside from the open interfaces that Open RAN promises, AT&T is also working with Ericsson to utilize a single RAN automation platform, for its entire network. This is the Ericsson Intelligent Automation Platform (EIAP), a Service Management & Orchestration (SMO) and Non-Real Time Radio Intelligent Controller (Non-RT RIC) solution for which the ecosystem is becoming much more mature, with Ericsson having already established an rApp directory and developer portal, and multiple third-party applications approved for its solution.

While the above is theoretically Open RAN, the validation and integration process is entirely handled by Ericsson when selecting third-party vendors to ensure that the hardware can be managed by EIAP. AT&T’s entire network will now be predominantly Ericsson hardware, as the third parties still only supply small cell radios, with all Massive Multiple Input, Multiple Output (mMIMO) radios coming via Ericsson. Furthermore, vendor diversity is limited to the radio, with Ericsson supplying all the baseband units. While both CommScope’s ONECELL radio solution and Corning’s Enterprise RAN (ERAN) have been integrated into EIAP via the O1 interface, announced at the end of January 2025, both solutions are for small cells specifically.

When looking at the wider landscape of Open RAN, no major Tier One vendors have announced Open RAN plans that come anywhere near the scale of the AT&T contract, and those that have been made have also been predominantly single-vendor approaches. The Deutsche Telekom Open RAN announcement only includes Nokia and Fujitsu, and there is evidence that only Nokia will supply the mMIMO radios, with Fujitsu supplying mid-band remote radio head products. This all indicates that the multi-vendor vision of the O-RAN Alliance, to put it lightly, is not proceeding as hoped. Many challenger Open RAN radio vendors have struggled, gone bankrupt, or even filed for Chapter 11 to avoid that fate.

The other aspects to deploying Open RAN networks have always been the SMO and RIC, which do not look any better than the hardware element. Ericsson, Samsung, and Nokia’s automation platforms dominate network activities. Juniper Networks has yet to announce a second RIC deal, and its (approval pending) merger with HPE will mean that the two entities will consolidate their respective SMO and RIC portfolios. Broadcom (formerly VMWare) recently announced that it will be shutting down its RIC and Telco Cloud Platform RAN products. Furthermore, AT&T is committed to adopting a single pane of glass approach for its RAN automation platform, so the EIAP will manage and orchestrate the entire network.

While AT&T’s single automation platform approach simplifies the management of the network for the operator, it immediately reduces the number of future third parties it will be able to add to its network, given that it is reliant on Ericsson integrating its solutions onto the platform. In a worst-case scenario, AT&T will find itself more locked into Ericsson, given that it has taken such a single-vendor approach than it has in the past. What happens when AT&T wishes to change automation partners? The new partner will have to integrate Ericsson’s entire RAN portfolio onto its own, which a Nokia or Samsung may not wish to do for competitive reasons and vice versa. That would likely be an impossible task, but AT&T would likely have no other alternative because only Ericsson and Nokia are large enough to cater to one of the largest Mobile Network Operators (MNOs) in the world.

The big issue for Open RAN is that its deployment is in the mid-cycle of a challenging cellular generation in 5G, with operators in all markets reporting stagnating or even falling revenue. Simply offering open interfaces with no additional revenue drivers is not a strong enough business case for wide-scale commercial deployment. Furthermore, the O-RAN Alliance excluded the incumbent infrastructure vendors when it was founded in 2018 in order to give new challengers an opportunity to displace them. However, the incumbents have adapted very quickly and updated their portfolios to ensure Open RAN compliance. And given their vast budgets for Research and Development (R&D) and experience in deployment scenarios, the market has begun consolidating in both the hardware and software spheres.

Open RAN is indeed being deployed in greenfield environments such as by Boost Mobile (formerly DISH). However, a greenfield environment is a much easier environment to perform a rip-and-replace in than a brownfield environment, even at the platform level. Furthermore, the opportunity for third-party rApps running on an SMO and RIC, and for operators to develop their own on top of a vendor’s platform, offers the opportunity for innovative applications to be developed by all stakeholders, which remains unique to Open RAN compared to traditional deployments.

This is not the first time that the telecommunications industry has gradually “closed” an open ecosystem. The Common Public Radio Interface (CPRI) was initially an open standard, but was quickly customized by major vendors, and the Advanced Telecommunications Computing Architecture (ATCA) aimed to create standardized telco servers, but proprietary modifications were eventually made. The same took place for the Open Base Station Architecture Initiative (OBSAI). All three initiatives were designed to open the base station market, but only managed to further lock it in.

What’s different here, however, is the continued support and commitment by both the operators, such as AT&T, and the vendors for the open interfaces for Open RAN. The initial vision of a truly open and vendor-neutral ecosystem is fading, and was perhaps unrealistic for the industry to believe would materialize. Governments in Western markets will continue to support the Open RAN initiative as they continue to prioritize supply chain diversification, but the vast majority of commercial deployments, especially for brownfield networks, will continue to be led by incumbent vendors. The approach taken by AT&T is likely to be followed very closely by other operators in terms of both the hardware element and the SMO element.

A company that has been able to position itself to capitalize long term is Fujitsu, having ingratiated itself and its radios (whether non-mMIMO or small cell) into both the Nokia and Ericsson ecosystems. Long term, should operators wish to continue diversifying their Open RAN supply chain, the first step will inevitably be reducing the reliance upon a single vendor’s portfolio and increasing the share of the network supplied by third-party vendors that were selected. One company that has benefited from the Open RAN initiative is Rakuten, which began expanding its presence outside of Japan with deals in Ukraine and partnership with the NTIA in the United States to support local RAN vendors.

This all means that while Open RAN is not dead in the literal sense, we are now experiencing a reality check for the opportunity and capabilities of Open RAN, and coming to terms that the vendor-neutral vision will not materialize. Long term, the industry focus among both operators and vendors will continue shifting toward the Artificial Intelligence (AI)-RAN concept, which does not rely on open interfaces for implementation and aims to address the need for better monetizing network assets at the edge (for more information, see ABI Insight “Exciting Results from SoftBank and NVIDIA's AI-RAN Trial Could Ignite a New Era in the Telecoms Industry”). The RIC is expected to be a part of this concept; however, an AI orchestrator will be necessary to manage both the RAN and the various AI workloads running on the infrastructure. This may necessitate an evolution of the SMO’s role, or a new orchestration platform that sits on top of the SMO in the stack. More news in this space is expected during MWC Barcelona 2025.