Decoding Hyperscalers’ New Egress Fee Waivers: Limited Cost Advantages

By Nelson Englert-Yang | 2Q 2024 | IN-7308

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.

All three big public cloud providers-Google, Microsoft, and Amazon Web Services (AWS)-have reduced transfer and egress fees, complying with Data Act regulations in the United Kingdom and Europe by instituting a new company-wide fee policy. This was done by Google Cloud Platform (GCP) in January, and then by AWS and Microsoft Azure in March. This sequence is significant as GCP initiates egress-elimination trends to promote business transfer from the two more popular cloud services; Microsoft had the most momentum toward fees, combining both egress and fees for Microsoft-licensed software outside of the cloud.

Fees for removing data from cloud services have long constrained mobile operators' use of the cloud. Indeed, ABI Research has found that among most operators, the technological challenges of running telco-grade network functions on General Purpose (GP) cloud infrastructure have given way to cost challenges of mounting Operational Expenditure (OPEX), especially amid hidden transfer and egress fees. As evidence of these challenges, an entire segment of Financial Operations (FinOps) dedicated to cloud cost monitoring has emerged.

While originating from U.K. and European regulation, changes to fee policies occur at the company-level and will impact all service users. Yet, cloud providers' stipulations over who receives free-egress privileges reduces the impact. Both Google and Microsoft stipulate that a client must be terminating cloud services within 60 days. AWS stipulates that a client must remove data within 60 days without requiring full-service termination. All three require customers to submit an application for reimbursement of fees. In effect, this "elimination" of egress fees only provides relief for large service changes, not for repeat operations.

This means there is no impact on how the public cloud is used. No relief is provided for hybrid architectures with network functions split between on- and off-premises, or multi-cloud architectures with lateral data exchanges among cloud providers. This is unfortunate. as egress fees will continue to skew planning away from these architectures, which may otherwise provide optimal distributions of network function in the cloud.

Nevertheless, fee changes may impact whether a public cloud service is used at all: by allowing free retrieval of data assets after deployments, cloud providers lower barriers to both service experimentation and planned service exits. First, operators still shopping for public-cloud services are now at an advantage, as they may compare across services, or public/private architectures, without the worries of post-trial egress costs in reclaiming assets. Second, operators that only require the use of a public cloud for a major service project after which they plan exit are also positioned to benefit from these changes. For example, an operator may use the public cloud's scale for training a Large Language Model (LLM), but remove the LLM for on-premises inferencing.

Within limits, the fee changes reduce the costs of experimenting with public cloud services and risks of reverting from service. This will lead to only a modest increase in cloud adoption and competition among hyperscalers, as there remain challenges associated with a permanent exit. Waivers also introduce yet another layer of strategy surrounding partnerships with hyperscalers, especially for temporary uses of the cloud.

As a result of these changes, operators may take the following steps to increase the likelihood of capitalizing on egress waiver benefits in both the present and the future:

• Understand the More Liberal Fee Positioning of AWS: It is unclear howstringent AWS will be in granting free egress for service changes. However, the combination of being highly price-competitive, offering the standard free 100 Gigabytes (GB) per month of egress, and waiving egress without account termination assures that it offers clients the most freedom among the three providers. For operators that are planning a cloud deployment of new services, the benefits of AWS and its moderate positioning among competitors may be worth considering.
• Chart Scenarios of Service Experimentation or Planned Service Exists from Which You May Benefit: Planning temporary use of a public cloud for network services is uncommon because migrating network services out of the cloud takes significant effort and operational planning. Rather than exit, operators often opt for optimization changes and implementing FinOps. However, planned service exists may occur when: 1) migrating data out is relatively easier than for core network functions like Operations Support Systems (OSS) workloads or Artificial Intelligence (AI) model training, and 2) operators benefit from public cloud elasticity for a set duration, after which it is advantageous to deploy on-premises. If an operator is temporarily deploying a workload in the cloud, and that workload is sufficiently detached from existing cloud workloads, it may be cost-effective to uncork a new cloud service for the egress waiver.
• Closely Monitor for Regulatory and Fee Changes: Regulators' objective is to increase market competition among public cloud providers. While these reforms are a step in the right direction, their impact on competitiveness is mitigated by the cloud providers' complete oversight of conditions for eligibility and the waiver process. Regulators will continue pressuring hyperscalers for additional reforms. While egress fees will not be wholly eliminated, we expect that this is the beginning of dynamic restructuring of cloud service fees that will eventually impact how cloud services are used. Operators should be vigilant in following these dynamics to further reduce egress costs.