Technology Blogs by SAP
Learn how to extend and personalize SAP applications. Follow the SAP technology blog for insights into SAP BTP, ABAP, SAP Analytics Cloud, SAP HANA, and more.
Showing results for 
Search instead for 
Did you mean: 


This blog is intended to give you detailed market insights on a trend visible since approximately 2017: The large-scale exodus of enterprise customers away from hosting providers to hyperscaler data centers.

In this first part you will find a collection of explanations and data points that help understand the logic behind this. The second part will focus on the impact of this dramatic shift on the hosting provider business model and – consequently – the remaining customer base.



This first part describes the market transition from hosting to hyperscalers and tries to explain, why these changes have taken place. Let us start with some definitions, so we are on the same page.

The IT service category application hosting, short hosting, has been around since the late Seventies. Hosting combines the technical management of data centers, server infrastructure, operating systems, databases and applications running on the managed systems. Often, the service is named after the application in scope, e. g. SAP hosting, or the managed application class, for example archive hosting. It belongs to the overarching IT service categories of managed services and IT outsourcing, which are often used as synonyms for hosting.
Sometimes called private cloud, hosting is rather associated with an on-premises deployment and together they form the “Any-Premise” and non-cloud categories.

In contrast, the term cloud belongs to the “true” cloud alternatives, Infrastructure-as-a-Service (IaaS) and Software-as-a-Service (SaaS).
Core to the hyperscaler offerings, the provisioning of workloads is named after their main delivery components, “storage and compute”. Apart from the obvious server and storage systems, hyperscalers always need to provide data center, network, backup monitoring subsystems etc. to make storage and compute usable for the enterprise customer.

The predominant hyperscaler market players covering more than 70% of the market share are:


Shifting Perception

During the last five years, I have had many dialogues with IT managers about the relevance of hosting in the presence of an ever-expanding offering of diverse cloud services. What I am seeing since a while now, is a shifting mindset within large and mid-sized customers alike. Let me explain this briefly.

While before the Corona pandemic, I would regularly receive push-back when expressing that mid-term, hosting would be an endangered service species. The hyperscalers being the comet that doomed the dinosaurs of hosting. Objections were around security, US data protection regulations etc. Maybe a feeling of disbelief was added subconsciously. How could a service category, that has been around for so long, be endangered?

Nowadays, it is only few customers that actively consider hosting as the future mode of operation. Interestingly, Dave Cappuccio from Gartner already knew in 2018, that “this is not an overnight shift, but an evolutionary change in thinking”, when his famous blog post “The Data Center Is Dead” was published.

What has happened?


Paradigm Change

There are several drivers that fuelled this change in perception of IT decision makers. In the following paragraphs we will investigate three prevalent factors driving this paradigm change:

  • Cloud First

  • Transformation to SaaS

  • Big Data & Analytics


Cloud First

Firstly, the rise of the hyperscalers in the 2010s has led to more acceptance of the term cloud, which has further increased the rise of the hyperscalers. A virtuous circle.

Infrastructure-as-a-Service providers cannibalized the infrastructure services offered as part of hosting. From the hyperscalers point of view, their focus on infrastructure helped generate even stronger economies of scale, which further increased with growth opportunities. “While we need six years to build a new data center, from planning phase to go-live, Microsoft builds it in six months”, a manager from a hosting provider told me in 2019.

While we need six years to build a new data center, from planning phase to go-live, Microsoft builds it in six months” – manager of a hosting provider (2019)

Today, in most cases you talk to IT decision makers, the credo “Cloud First” is common ground. And what that means is a hyperscaler-based cloud deployment, either infrastructure-centric as Infrastructure-as-a-Service or combined with software, so Software-as-a-Service.


Transformation to SaaS

Secondly, the influence of smartphone apps on enterprise application landscapes should be considered. Around 2008, independent from IT, business users started adding smartphone apps to their solution landscapes. Thus, they enriched or sometimes replaced existing client-server applications with mobile point solutions located anywhere, preferably at a hyperscaler data center.

The mobile app boom began with the introduction of Apple’s iPhone and is one of the defining motions of IT landscapes in the 2010s. Typical examples for app functionalities replacing business solutions: travel claims and contact management. With this mobile app boom, the category of Software-as-a-Service became a super star.

For existing software vendors, it was obvious that the demand for cloud and Software-as-a-Service created market opportunities and the risk of falling behind at the same time. Most software companies started to migrate their solutions to the cloud, some exclusively, some as an alternative offering. The pressure on their commercial model was high, but the only chance was to adapt. For the hosting providers every SaaS customer meant losing workloads.


Big Data & Analytics

Lastly, there was the early Big Data & Analytics market. With the growing relevance of data-driven business models, the demand for data lakes, data warehouses, business intelligence solutions, predictive apps etc. increased significantly. Scalability and raw computing power were key for Big Data, as large amounts of data needed to be received, transformed, aggregated etc. A difficult task for a hosting environment for two reasons.

First, the hosting providers had to overcome the lack of expert resources and hardware manufacturer lead times for the ever-increasing amounts of data. Second, hosting customers needed to overcome stiff commercial models tied to financial depreciation and contract signatures.

To overcome this, existing data warehouses etc. were migrated to the cloud. For new Big Data solutions, in most cases a hosted solution would not even be considered.

Figure 1: hyperscalers cannibalizing hosting

Beyond Cannibalism

All of this enabled a massive shift of hosting data center workloads to hyperscaler storage and compute services, see figure 1. This happened over the last 5-10 years and continues, just as Dave Cappuccio predicted.

The initial hosting exodus can be explained by the exemplary drivers stated above – Cloud First, transformation to SaaS and Big Data. On top of this, there was an ever-increasing acceleration of hyperscaler demand. This demand never reached a hosting provider but went directly to the cloud. This led to further economies of scale for the hyperscalers and took away market opportunities for the hosting provider landscape.

We will now examine some additional influencing factors, that complete the picture:

  • hyperscaler commoditization

  • digitization boom

  • new SaaS offerings


Hyperscaler Commoditization

On the one hand, hyperscaler migrations became more and more commodity. The availability of hyperscaler reference customers as well as both cloud migration and operations experts increased significantly. At the same time, customers collected knowledge about hyperscaler workloads and grew more and more confident with the cloud approach.

On the other hand, hosting was beginning to be labelled as a non-strategic operations model. In the late 2010s, it became common sense even for business executives to talk about cloud and the benefits that come with it. Thus, opening the market for hyperscaler deployments even further, growth rates started exceeding the 30% mark across the board.



Over the last two decades, the number of digitization projects has continuously increased. During the Corona pandemic, this trend found its peak, leaving not a single corporate or government client unconvinced of digitization necessities. Agility of business models, simplification etc. were on most executives’ agenda since many years already.

To enable digitization, the placement of workloads came more and more into focus:
Talking about digitization automatically involved moving to the cloud.


New SaaS offerings

Over the course of the 2010s, there were a lot of forces changing business and IT alike, as discussed above. The result was a tectonic shift to the cloud, being Infrastructure-as-a-Service or Software-as-a-Service. The overall mindset was now different.

New software developments projects would automatically be started at a hyperscaler. And for sure, the development would be agile and realized as DevOps, yet another disruptive change that took place contemporarily to the cloud trend. The deployment model would undoubtedly be SaaS, just as the enterprise market requested. As a result, the SaaS market on average grew by 25% annually during the 2010s.

Combining all of the above, Cloud First, existing and new SaaS offerings, Big Data, the digitization boom and the overall commoditization of hyperscaler workloads sealed the fate of hosting during the second half of the 2010s.


Facts & Figures

Now let us examine some figures.

Looking at below graph of IT spend over the 2019-2026 period, we can see the persistent gap between cloud and non-cloud budgets. While overall spend growth averages at approximately 7%, non-cloud spend is circa at 3% and cloud at almost 21% (at constant currencies).

In the current “Black Book” from analyst IDC you can find some detail figures for 2023. The decline of cloud and non-cloud spend alike to pre-pandemic growth rates is obvious. Non-cloud growth in 2023 shrinks by -1%, cloud continues to grow slower than before at 19%.

Diving deeper into cloud figures, we find SaaS growing at 19%, just as overall cloud, but outperforming the combined on-premises and cloud software which is stuck at 11% growth. The IaaS market is growing at 26%, even faster than overall cloud.


Figure 2: worldwide IT spending by deplyoment type, source: IDC Black Book (April 2023)


According to the current “Infrastructure-as-a-Service (IaaS) Market” report from Fortune Business Insights, the IaaS market will continue to grow at this pace until at least 2030: “The global infrastructure as a service market size is projected to grow from $130.08 billion in 2023 to $531.84 billion by 2030, at a CAGR of 22.3%”.


Summary and Outlook

We have taken some quick looks at the reasons for the paradigm change driving the perception of IT executives – and consequently workloads – away from hosting providers to hyperscalers. This happened over the last 10 years and is a continued trend.

Looking at current market figures, we see this to persist: A decline for non-cloud spend and 22-26% growth for hyperscaler workloads is a clear message.

But what will we deduct from this? Building on the foundation of this analysis, part II of this blog post will focus on the consequences for the hosting providers, their employees and customers. There are substantial risks building up and we need to talk about this.


More information on hosting and cloud deployments can be found here:


Take the opportunity to maximize the value of your digitization project.
Check this resource regularly to find more interesting blog posts on the topic of digitization.

Follow me, so you do not miss any of my future blog posts!