The Premiumisation of Cloud

The next 18 months will see more changes in Cloud economics than the prior 5 years combined. Enterprises will need to adapt their strategy & practices to deliver best value against this backdrop of change.

We will see the premiumisation of hyperscale Cloud, with exponentially greater value delivered through higher cost services on public Cloud, with the potential for the increasing costs of higher premium consumption offset by increasingly (relatively) competitive non-hyperscale infrastructure for lower value workloads.

This blog explores:

• What is the premiumisation of Cloud?
• What is driving changes in cloud economics
• Why KPMG sees premiumisation
• The impact on Cloud services for enterprises
• How enterprises need to adapt

Firstly, what do we mean by the premiumisation of Cloud? This is a phenomenon we see in public Cloud whereby premium services become the dominant part of the offering, attracting a correspondingly premium price. Let’s break that down:

• The cloud is the primary means by which most enterprises get the latest digital technology
• It has always included premium services such as the latest analytics tooling, but generative AI in particular sees a significant gulf between mid-market capabilities on non-public Cloud services, and the premium capabilities on public Cloud
• These services are costly to develop, therefore necessitate high prices
• And they have a significantly higher value to customers of public Cloud who will pay this premium

Secondly, what is driving these changes in cloud economics? Fundamentally, this is because the availability of innovation especially around generative AI is combining with four powerful forces to disrupt the cloud economic model which has prevailed in the market for nearly a decade:

So, thirdly, why does KPMG think this will result in premiumisation? Hyperscale computing costs are increasing (even without generative AI) for the same reasons inflation has affected many sectors – energy, labour and supply chain costs. This means the cost of cloud computing overall will increase and that’s especially the case because generative AI is both compute intensive, but also demands newer and more sophisticated chips.

The customers of hyperscalers are using the sophisticated capabilities of their platforms to transform their own enterprises – to reduce cost, manage disruption or to serve their customers or citizens better. It makes little sense to slow down the potential of this innovation and switch to a cost play. And if the supply of their compute and storage risks being constrained, it’s logical they would price to incentivise consumption towards the highest value, higher cost capabilities.

But for their competitors (and let’s remember non-hyperscale infrastructure providers have also seen healthy growth), the technical barriers to interoperability have reduced and regulators such as Ofcom are applying pressure for this to continue (e.g. the removal of egress costs). For them a cost play to secure lower value, less mobile commodity compute and storage does make sense – it’s their core business.

So what? This is our fourth topic – what does this mean for the Cloud services enterprises consume? In simplified terms:

• Use of public cloud concentrates on higher value, premium services like generative AI which enable more complex and intelligent solutions for data analysis, automation, and decision-making
• Simple storage, compute goes to the lowest cost platform, and non-hyperscale providers provide economic, basic services plus tailored packages of capabilities for specific products and industries

These trends won’t be absolute (i.e. there will never be a clean dividing line between premium and commodity), but even if directionally right there are a number of implications.

Finally, these implications change the ways enterprises seek value from Cloud, and so they change the way they need to architect and organise for cloud. Public cloud is where most value will be created, so enterprise need to be able to exploit it effectively. But it’s not enough to say hyperscale Cloud is always the best solution to create value: being flexible, portable and value oriented is most important.

• More routinely modelling and reviewing the cost of your cloud workloads and what value they generate – to avoid being caught off-guard by price increases or missing opportunities in the market. KPMG’s Transformation Impact Modeller is helping clients do this for over £5bn of spend
• Managing value effectively from premium workloads, including understanding whether your speed to deliver is competitive, investing in initiatives such as KPMG’s Cloud Value Lab to properly explore and define value from breakthrough capabilities
• Architecting for premiumisation – by ensuring your architecture supports future portability and minimises lock-in within your Landing Zones, automations, controls and major architectural choices such as containerisation
• Resetting your toolset choices to consider portability and the ability to manage high value, high velocity workloads andlow churn, commodity workloads in different platforms and at different cost points
• Assessing, designing and engineering the Data Architecture to be most effective in a hybrid cloud environment – for example, what you store in the Cloud versus what you expose to the Cloud when needed
• Rethinking the skills which are needed – and the extent to which your skills create dependencies on individual platforms which impede being dynamic as the relative value of platforms changes

CSPs are pushing new Premium services to create more value for their clients. The value and cost of innovation is combining with supply-side constraints on energy, chips and other hardware components, moving Cloud services towards higher-cost, higher-value premium models. This isn’t just about the cost on the price tag, it’s also about changes in the products on the shelf.

This will incentivise enterprise to review their enterprise strategy, working out new ways to create value from premium services and take efficiencies from commodity services. This will change enterprise Cloud strategies and the way they are executed – or at least, it will for those who stay ahead of the curve.