Rethinking the Role of Hyperscalers in the Telecom Industry

Dish recently announced a major collaboration with Amazon Web Services (AWS) to build its cloud-native nationwide 5G network. Unlike Rakuten’s cloud-native network architecture, Dish will be using the hyperscaler’s cloud computing infrastructure to host its 5G core and Radio Access Network (RAN) for network capacity and coverage. Leveraged by the public cloud resources in AWS Regions, AWS Local Zones, and AWS Outposts, the company will use the infrastructure to accelerate the pace of its network deployment and offer end-users more flexible and agile network services. Beyond this, the company also expects to tap into the software developer ecosystem of AWS to create new service opportunities for efficient network management and cost reduction.  

Apart from the AWS partnership, Dish also works with other software vendors to bring multi-vendor interoperable solutions to improve its network sustainability and drive innovations. These companies include Allot, Nokia, and Palo Alto Networks for security management, Mavenir and Altiostar for virtualized RAN functionality, Netcracker for automation and orchestration, Amdocs for billing, VMware for the cloud-native virtualization platform, and Oracle for 5G core Service-Based Architecture (SBA). Dish proposes a responsibility assignment model for supporting the qualified end-to-end service delivery. Each segment of the network will have a responsible vendor for the segment performance, and Dish will act as a system integrator for the entire end-to-end service assurance.  

Following Dish’s deployment example, telecom network architecture has started moving from vertically integrated solution to a modular-based approach. By disaggregating software from hardware and introducing open interface and reference designs, network operators target unlocking the traditional telco supply chain, which was dominated by a handful of incumbent infrastructure vendors. Moreover, they also work with different stakeholders, e.g., software and hardware vendors, hyperscalers, system integrators, and application developers, to build more flexible and agile network architectures and meet specific connectivity needs.

The open virtualized and cloudified network architecture offers network operators great opportunities to create new business models for monetization. However, the dynamic connectivity requirements give rise to an even complicated network management and further increased the Total Cost of Ownership (TCO). When considering simplifying network management and making the most efficient use of limited radio resources, network automation and RAN Intelligent Controller (RIC) implementation are required. In this case, the network deployment focus should move from distributed RAN architecture to centralized ones, such as Cloud RAN.

Leveraged by analytics and data-driven approaches, it is envisioned that the future network will allow a much more intensive interaction between network infrastructure and end-users. With such consideration, Artificial Intelligence (AI)/Machine Learning (ML) technology will play a pivotal role in processing the collected data and giving network operators more precise and valuable network management insights. To access the transformative benefits of AI/ML technologies, the high computing capability of a network is a critical requirement, which is where hyperscalers are well-positioned to leverage their cloud computing resources in this domain.

There has always been debate about whether hyperscalers, such as  Amazon, Google, Microsoft, and Facebook, will replace network operators and/or telecom vendors and dominate the telco market, especially when network virtualization/cloudification becomes fully mature and enterprise use cases reach large scale. From ABI Research’s perspective, that will not likely happen, at least in the next five to ten years. The key considerations are summarized as below:

• Hyperscalsers resort to their cloud-native platforms to attract more developers and benefit from software innovations. However, deploying and managing a telecom network, especially in the public domain, is not as simple as operating an application store on a smartphone or managing a network through redundancy. Apart from the advanced radio hardware development challenges, there will be enormous complexity in terms of network reliability, security, and performance guarantee.
• With the advent of various 5G services demonstrating stringent latency and reliability requirements, there is no surprise that major carriers will have their edge computing deployment relationship with hyperscalsers. However, managing these resources for Service-Level Agreement (SLA) assurance and improved user Quality of Experience (QoE) will be challenging and may need an experienced team for the end-to-end network management.

Despite the above consideration, ABI Research sees that hyperscalers will become the platform of choice for future networks, considering their cloud computing resources and broad software developer ecosystems. Network virtualization and cloudification have catalyzed the partnership between network operators and various cloud service providers for multi-layer network orchestration and strong network sustainability. To avoid losing the market share, key stakeholders, including network operators and infrastructure vendors, should be open and proactive for innovations. The standard interfaces and open reference designs can help foster a better ecosystem and lower the threshold for smaller and new entrants, but the development progress of these specifications is slow. Front runners in the telecom market should not wait for their maturity and start testing solutions now to seize the first-mover advantage.