Ericsson & Google Launch True 5G Public Cloud

The ambitious journey of telecommunication companies (telcos) to migrate their 5G core networks to the public cloud has been fraught with challenges. While the control plane acts as the “brain” of a 5G core, meticulously directing traffic and making critical decisions, the user plane (UP) serves as the “brawn,” handling the immense volume of data traffic.

 

It is this user plane function (UPF) that has historically presented a significant hurdle for telcos attempting to leverage the public cloud’s vast but geographically dispersed data centers. A notable example is Telefónica, which encountered scalability issues when hosting Nokia’s UPF on an Amazon Web Services (AWS) facility in Germany. Their service struggled to support even a small fraction of its customer base, ultimately requiring an AWS Outposts server – essentially an on-premises deployment by AWS – to resolve the issue.

 

This scenario has led many critics to question whether such deployments truly represent a “public cloud” solution, often likening them more to on-premises arrangements managed by a cloud provider. Even Microsoft has openly labeled its telco offering as a “hybrid” cloud, acknowledging that a pure public cloud approach isn’t always feasible for core network functions. Among the “big three” hyperscalers, Google has traditionally maintained the lowest profile in the burgeoning market for telco workloads.

 

However, a significant shift has just occurred. Google’s newly announced 5G partnership with Ericsson marks a pivotal moment, promising what could be the most genuinely public cloud offer seen in the telecom sector to date. This collaboration signals a bold strategic move for both companies, potentially reshaping the future of 5G network deployment.

 

A Distinct Cloud Strategy: Ericsson’s Approach

The new alliance between Ericsson and Google holds profound implications for both companies, particularly highlighting Ericsson’s unique strategic path compared to its main rival, Nokia.

 

Divergent Paths: Ericsson vs. Nokia

Ericsson and Nokia, two of the world’s leading telecommunications equipment vendors, have pursued markedly different strategies when it comes to cloud infrastructure and core networks. Nokia, for instance, has publicly declared its decision to abandon work on its own cloud infrastructure platforms. Instead, the Finnish vendor has forged a strong partnership with IBM’s Red Hat, naming it as its chief collaborator for cloud-native deployments.

 

Nokia often points to its existing tie-up with AWS for Telefónica Germany as evidence of its “agnosticism,” suggesting it can deploy its core network applications across various cloud environments. This approach emphasizes flexibility and compatibility with multiple third-party infrastructures.

 

Ericsson, by contrast, has historically maintained a different stance. The Swedish vendor has remained firmly committed to its in-house Cloud-Native Infrastructure Solution (CNIS) platform. This proprietary platform forms the backbone of Ericsson’s core network deployments. While Ericsson has, on occasion, integrated its 5G core network applications with third-party infrastructure – for instance, the UK’s BT runs the Ericsson core within its own network cloud – there has been a notable absence of any prior accommodation or direct partnership with a public cloud provider for core network functions.

 

This new collaboration with Google therefore represents a significant strategic pivot for Ericsson, marking its first deep dive into the public cloud for its essential 5G core functionalities. It signals a willingness to extend its core network capabilities beyond its traditional on-premises or private cloud deployments.

 

“Ericsson On-Demand”: A Native Public Cloud Solution

The essence of this new partnership with Google is captured in Ericsson’s offering, dubbed “Ericsson On-Demand.” According to Eric Parsons, Ericsson’s Vice President of Emerging Segments in Core Networks, this collaboration necessitated Ericsson going completely “native” within the Google environment. This “native” approach means a deep integration and optimization of Ericsson’s 5G core applications specifically for Google’s cloud infrastructure, moving beyond a simple “lift-and-shift” migration.

 

Parsons emphasized that this native integration is the key differentiator. It allows the critical user plane function (UPF) of the 5G core to run directly at Google’s hyperscaler facilities. This design aims to circumvent the scalability and performance problems that previous attempts, such as the AWS and Nokia collaboration, encountered when trying to host the UPF in a public cloud. Parsons explained, “We have departed from a lift-and-shift approach and really gone network-native in the Google environment.”

 

He further clarified the technical depth of this integration: “It requires a mating between the networking layers, between the type of workload that we have and the underlying networking capabilities of the infrastructure.” This statement suggests a custom-engineered solution that leverages Google’s specific cloud architecture and networking fabric, allowing Ericsson’s core network components to operate with maximal efficiency and performance.

 

The Trade-off: Native vs. Agnostic

However, Parsons also explicitly stated a crucial implication of this “native” approach: it does not mean agnostic. Unlike a truly agnostic solution that could theoretically be redeployed across different cloud providers with minimal effort, the Ericsson On-Demand solution, as currently designed with Google, could not be seamlessly migrated or deployed with AWS or Microsoft.

 

This dedicated integration means that while Ericsson gains optimized performance within Google’s cloud, it sacrifices the flexibility to easily switch hyperscalers without significant additional investment. This strategic choice underscores Ericsson’s belief that deep, native integration is necessary to achieve the desired performance and scalability for 5G core functions in a public cloud environment, even if it means a degree of vendor lock-in.

 

Performance, Economics, and Industry Skepticism

Despite the technical promises, the lack of commercial deployment of Ericsson On-Demand has naturally led to some industry skepticism regarding its real-world performance and long-term economic viability.

 

Addressing Performance Concerns

The performance of the UPF in a public cloud environment has been a historical sticking point. Critics wonder if a truly public cloud deployment can measure up to the demanding low-latency and high-throughput requirements of 5G user traffic. In one possible scenario envisioned by Ericsson, the UPF could even be hosted at a Google facility located outside the country where the telco operates. This raises questions about data sovereignty and latency for real-time services.

 

However, Eric Parsons strongly insists that this approach involves “no big compromises” in performance. He points to Google’s robust global infrastructure as a key enabler. “It is very clear, with the 2 million miles of fiber that Google has to run networks across the globe, that they’re extremely reliable now,” he affirmed. This extensive, high-speed global network is critical for ensuring reliable and low-latency connectivity, even across international boundaries.

 

Parsons further asserted, “We have the ability to run the types of network capacities that operators need to be able to light up in order to provide services to their mobile network consumers.” This confidence stems from the deep, “network-native” integration that optimizes how Ericsson’s software interacts with Google’s underlying network capabilities, aiming to deliver carrier-grade performance even in a public cloud setting.

 

Economic Attractions for Operators

Ericsson On-Demand is expected to hold significant attractions for telcos, particularly those keen on minimizing capital expenditure (CapEx) and reducing the complexities of infrastructure ownership and management. Traditionally, deploying Ericsson’s 5G core technology on-premises would necessitate substantial upfront investment in hardware, including servers, storage, and networking equipment, followed by ongoing costs for maintenance, power, and cooling.

 

By contrast, relying on Google’s extensive network of data centers allows an operator to avoid this significant CapEx. Instead of purchasing and maintaining physical infrastructure, telcos can effectively outsource the management effort to Google. This translates into an operational expenditure (OpEx) model, where operators pay only for the resources they consume, akin to a utility model.

 

This “pay-as-you-go” structure offers financial flexibility, reducing the need for large lump-sum investments and potentially freeing up capital for other strategic initiatives. For smaller operators or those looking to rapidly deploy new services without the burden of infrastructure build-out, this model can be particularly appealing.

 

Potential Downsides and Sovereignty Concerns

Despite the clear economic advantages, the software-as-a-service (SaaS) model inherent in Ericsson On-Demand comes with its own set of potential downsides.

 

The most obvious is the loss of direct control over the underlying infrastructure. While management is outsourced, critical decisions regarding hardware refresh, specific security configurations, or highly customized environments might be constrained by the hyperscaler’s standardized offerings. Not everyone in the industry is convinced that the SaaS model will necessarily translate into lower costs in the long run, especially for very large-scale, long-term deployments, where the cumulative OpEx might eventually exceed a CapEx investment.

 

Furthermore, a significant concern, particularly for governments and larger telcos, is the increasing demand for “sovereign cloud” solutions. Many nations are becoming more rigorous about requiring data to be stored within national borders and, in some cases, even managed by a local player. This mandate aims to enhance data security, ensure compliance with national regulations, and protect critical infrastructure from foreign influence.

 

A purely public cloud model, especially one where UPF traffic could theoretically traverse facilities outside national boundaries, might conflict with these emerging sovereign cloud requirements. This regulatory landscape presents a complex challenge for widespread adoption of such a globalized public cloud model for core network functions.

 

Target Market and Strategic Outlook

Ericsson acknowledges that its “On-Demand” offering will not be a universal solution, instead targeting specific segments of the telco market.

 

Tailoring the Offer: Smaller Operators and Secondary Use Cases

Eric Parsons is realistic about the market fit for Ericsson On-Demand. He recognizes that the offer will not suit every telco. The immediate and primary target market is likely to be smaller operators. These operators often lack the extensive resources, technical expertise, or capital to build and manage large-scale, custom on-premises network clouds.

 

For them, Ericsson On-Demand offers a more “streamlined,” off-the-shelf product that requires minimal customization. “Just to be clear about that, it’s not customizable in the same way that the on-prem solution is,” Parsons clarified. This pre-configured, standardized approach appeals to operators looking for quick deployment and reduced operational complexity.

 

For larger telcos, which typically have significant existing infrastructure and a strong preference for highly customizable solutions, Ericsson On-Demand is more likely to be utilized for what Parsons calls “secondary use-case interests.” This could include niche services such as a regional Internet-of-Things (IoT) service, temporary capacity boosts for specific events, or the rapid deployment of new enterprise-focused applications that don’t require core, mission-critical network functions to reside on proprietary infrastructure. This dual targeting allows Ericsson to maximize the appeal of its new offering without cannibalizing its traditional on-premises market.

 

Complementary Portfolios, Unchanged Core Strategy

Despite this significant move into the public cloud, there is certainly no talk of Ericsson On-Demand replacing the more traditional CNIS product. This means Ericsson still has no plans to follow Nokia’s lead and completely exit the cloud infrastructure market for telcos. Parsons emphasized this point: “We have complementary portfolios leveraging the same battle-proven technology.”

 

This indicates that Ericsson views the public cloud offering as an extension, rather than a replacement, for its existing core network solutions. “The complementary deployment now allows it to run in the public cloud,” he added, reinforcing the idea of an expanded ecosystem. He further reiterated, “There is, though, ‘no change in strategy in terms of what we’re doing for on-prem-type offerings.'” This confirms Ericsson’s continued commitment to providing solutions for telcos that prefer to build and manage their own private network clouds, thereby maintaining a diversified product offering that caters to various operator preferences and regulatory requirements.

 

The Agnosticism Debate: A Fundamental Choice

The Ericsson-Google partnership rekindles a long-standing debate in the telco cloud domain: the trade-off between deeply integrated “native” solutions and flexible, platform-agnostic deployments.

 

Call for Separation and Standardization

Senior executives within some telcos have been vocal proponents of greater separation of network applications from underlying infrastructure platforms. Their vision advocates for increased standardization that would enable them to combine the same network application (e.g., a 5G core component) with a variety of infrastructure platforms, whether they be from different hyperscalers, private clouds, or hybrid environments. This approach promises greater vendor choice, reduced lock-in, and potentially lower costs through increased competition and portability.

 

However, a customer of Ericsson On-Demand would, by design, not be able to seamlessly move the 5G core application to AWS, Microsoft, or any other alternative cloud provider.

 

Having chosen “nativeness” over “agnosticism” for this specific offering, Ericsson would be required to make significant new investments and engineering efforts to support its service with a different hyperscaler. This fundamentally limits the customer’s portability once they commit to the Ericsson-Google integrated solution, a point that directly contradicts the desires of many “agnostic” advocates in the industry.

 

The Economic Case for Nativeness

Despite this limitation, Eric Parsons firmly believes that the “native” approach makes greater economic sense for Ericsson. “In order to make the economics work, you have to go native, and that’s what we’ve done,” he asserted. He explained that achieving optimal performance and cost efficiency requires a highly tailored integration: “We have load balancers that are designed for the type of workload we’re talking about and that interface with our engines.”

 

Bringing these specialized components together necessitates a joint investment between Ericsson and its chosen hyperscaler partner, Google. Parsons acknowledges that “Should we at some point decide to partner elsewhere, yes, we will have to invest. And we would expect there to be investments required wherever we go.” This candid admission highlights that true agnosticism, while appealing in theory, may be prohibitively expensive or technically complex to achieve with optimal performance for demanding 5G core workloads.

 

The underlying insinuation from Parsons is that a truly agnostic, “lift-and-shift” approach for core network functions simply “is not practical” if operators demand peak efficiency and performance. “The efficiency is far from ideal,” he said, indicating that generic cloud deployments without deep optimization cannot meet the stringent requirements of 5G.

 

He concluded with a strong endorsement: “We have to solve this problem in a native fashion together with a partner, and the partner that has really met the challenge is Google.” This statement is a significant testament to Google’s capabilities, especially given its previously lower profile in telco workloads. However, this commitment to a deep, native integration, while potentially delivering superior performance, will undoubtedly fuel further debate among those who advocate for a more open and agnostic approach to telco cloud infrastructure.