EU Data Act - Focus on Cloud Services: switching charges - who should shoulder the cost?

Europe

In the third Law-Now in our cloud services series on the EU Data Act (the “Act”), we focus on the proposed measures to reduce and remove switching charges. The Act aims to facilitate customers switching between cloud service providers and other data processing services. In pursuit of this aim, the Act will regulate how cloud service providers can charge for switching and what customers can be expected to pay to switch to another provider. It’s likely that switching charges will be a topic of significant negotiation during the trialogue, as evidenced by the various amendments to the relevant sections in recent proposals issued by the Council and Parliament.

In this article we discuss some of the challenges the legislature will face in seeking to achieve an appropriate balance in delineating responsibility for switching costs, and as a consequence what cloud service providers will be entitled to charge their customers for.

For the purposes of this article, we have used the latest available draft, the fourth compromise text (published 24th January 2023) as our reference point. It should be noted that the various amendments proposed by members of the European Parliament differ slightly from this.

Practical considerations

As discussed in our previous article in this series, customers select the cloud services they want, based on their functionality and the other services and tools the cloud service provider (“CSP”) offers that can be leveraged. Each customer designs its own solution, choosing the features and functionality it wants when leveraging the cloud service to build its desired IT environment functionality. The built environment will inevitably evolve, likely growing in capacity and complexity over time. The complexity and efficiency of the customer’s solution is a key determining factor in how easy and cost effective it is to switch to, or interface with, an alternative or additional solution. 

When a customer first purchases a cloud service, the CSP will most likely not - and probably cannot - know what the customer’s solution will look like at the time the customer seeks to exit in several months or years’ time.  Nor can the original CSP (or, likely, the customer) know at commencement of its services what nature of solution the customer will seek to build in an unidentified future successor CSP’s cloud environment.

This raises some obvious questions in the context of allocating responsibility for switching charges, and how to achieve the best customer outcomes. Scope certainty (what is being priced) and competitive pressure are generally required to achieve the best value pricing. So who is best placed to price the switching costs, and when will they be best placed to do so? To the extent any professional services are required (e.g. data analytics, data mapping etc.), do you ask the original CSP to price the unknown and build its best assessment of the cost into its base service cost against an assumption of a minimum service term? Or, instead of asking the original CSP to price an unknown future cost, should the customer, at the point of exit and with knowledge of both its final solution parameters at that time and its desired new solution choose to do the work itself or to go into the market and seek tenders for the support it requires – whether from the original or successor CSP or a specialist service provider?  If the CSP is required to ‘price the risk’, and the customer chooses to exit early, is the CSP to be allowed to levy a termination charge to recover any shortfall in the recovery of this ‘hidden’ cost?

What are ‘switching charges’, and what can customers be charged for?

The latest draft of the Act differentiates between: (i) switching charges; and (ii) data egress charges. This in contrast to the Commission’s original proposal, which only defined switching charges. Data egress charges appeared as a concept in the second Czech Presidency compromise text and was then added to Article 25 in the third compromise text.

Data egress charges’ are defined as charges imposed on a customer by a data processing provider for transferring data to another provider or to its own data processing infrastructure (known as ‘on-premise’). ‘Switching charges’ comprise any charge, other than a data egress charge, imposed on the customer by a data processing provider for switching its systems (as provided for under the Act) to another provider. There’s a degree of legal uncertainty around the intended scope of these definitions, one interpretation is that data egress charges will relate to the direct costs charged for transit of data, whereas switching charges will cover a potentially much broader range of costs incurred by the data service provider in allocating or providing additional personnel or other resources to facilitate the transfer of customer data.  Its arguable that, as currently drafted, switching charges could be construed broadly enough to include remunerating third party professional service providers involved in the switching process. If the legislature does not intend this, it will be important for them to carefully delineate the professional services which are intended to be covered by each definition.

The Act allows CSPs to levy ‘reduced’ switching charges and data egress charges during a transition period of three years following the Act’s entry into force (the “Effective Date”). While it is clear that there is to be a transition period, there remains significant uncertainty around the situation following this. It is possible that, for example, CSPs are not allowed to levy either data egress or switching charges. During the transition, while both sets of costs can be charged, they must not exceed the costs incurred by the provider for the data transfer and switching process. The Commission will have the power to adopt delegated acts to introduce a monitoring mechanism so it can police compliance with these requirements.

Implications for CSPs

Prior to the Effective Date it appears that CSPs will be able to recoup a proportion of the costs they incur in facilitating customers switching to other providers or on-premise infrastructure.  The extent of what they are permitted to charge customers is unclear, but appears capped at the aggregate amount of the costs “directly linked” to the data transfer or the switching process itself. This could be interpreted as meaning that while the reduced ‘run’ costs for effecting the data transfer may be charged, the ‘build’ costs of developing the data transfer mechanism itself cannot be.

Following the Effective Date CSPs will not be permitted to charge anything for the switching assistance that the Act requires them to provide. 

In practice it appears inevitable that customers will still, ultimately, have to pay for these switching costs.  If CSPs have to incur costs, but aren’t permitted to levy any additional stand-alone charges for them, then the most likely outcome appears to be that the CSPs’ standard service pricing will need to increase to absorb this as a ‘hidden’ cost. 

While it is understandable why the Commission would seek to reduce and then remove switching charges for customers as part of a drive to incentivise switching, the Act as drafted arguably doesn’t reflect the technical and practical realities of the switching process.

Customers choose to switch and diversify cloud services for varying reasons, which will not be necessarily visible to the original CSP prior to the request. In addition, it is highly likely that in most (if not all) scenarios, CSPs won’t know the purpose of the customer’s data transfer or whether they are seeking to switch to another provider or to an on-premise solution. From the CSPs perspective, there is no difference, as both require the use of the same network resources.   Similarly, the tools, other resources and the work involved and as a consequence the costs, of a customer wishing to switch from one complex cloud solution to a differently architected one with a different CSP are likely to considerably exceed those of a customer simply replicating its data for disaster recovery purposes.

Cloud services are commonly provided on ‘self-help’, ‘pay as you go’ basis. The CSP provides an environment and tools which the customer leverages to build its desired solution. The same principle applies through the service lifecycle, including on termination. As such, any professional services that may be needed, including at termination (or to assist diversification), are generally self-provided or, if sourced from the outgoing or incoming CSP or a third party, payable in addition (or in the case of the incoming CSP, factored into its ongoing service pricing).  

Recital 74 states that the Act ‘does not require data processing service providers to develop new categories of services within or on the basis of the IT-infrastructure of different data processing service providers to guarantee functional equivalence in an environment other than their own systems’. But it is not clear that this extends to not having to provide any professional services at all in connection with switching. Clarity in this regard is needed so that customers and CSPs alike know who is responsible for what.

If the costs associated with the switching process that CSPs are required to absorb go well beyond mere data egress costs – and so are potentially quite significant - the effects on customers could be quite detrimental.  While competition in the industry should drive CSPs to competitively ‘price the risk’ of what the average unknown switching costs they have to bear might be, it may take some time for that pricing to settle and in the interim customers may face increased charges.  In these economically challenging times that could in turn have an adverse impact on the cloud industry by slowing - or even reversing - the transition from on-premise to cloud services.  

What can be done?

Instead of limiting and then abolishing charges, thereby driving them to become ‘hidden’ costs, it would arguably be better for the Act to require transparency on costs so that customers have a clear view of what is being paid for and how this varies depending on the work involved in a switch. The separation of ‘data egress costs’ from ‘switching costs’ seems to be a logical first step towards delineating costs, but the draft proposals currently prohibit both, failing to acknowledge that switching charges can cover a range of optional, but potentially resource heavy, services that CSPs offer.

Clarity is also required as to the extent of the additional services, and hence costs, that are in scope.  This is important to avoid negative outcomes for both CSPs and their customers. As drafted, it is unclear as to the extent that additional services (and particularly professional services) are caught by the definition of switching charges. Given the costs involved, all parties would benefit from the position in this regard being clearly set out.

Finally, data transfers for purposes other than switching to another provider (e.g. offering data for download to end users or streaming content) should be clearly differentiated from transfers for the purposes of changing or duplicating CSP so that originating CSPs are not providing services for free for purposes which are not intended, nor specifically provided for, by the Act.

Conclusion

As the draft Act moves into the trialogue phase, we expect negotiations on switching costs and charges more generally, including their scope and recoverability, to continue to be a central feature of the discussions.

In our next Law-Now, we will focus on the interoperability requirements for data processing services.

For more information on the Act or if you have any questions on how it could affect your business, please reach out to your usual CMS contact.