Paying for under-utilized infrastructure? Wasn’t cloud supposed to solve that problem?

Back in the mists of time when it was still considered risqué for enterprises to migrate to public cloud infrastructure, one of the key selling points to encourage IT buyers to make the move was that they would only pay for the compute power they used. This, of course, is still a selling point, but the rules of engagement between enterprise buyers and suppliers are changing as use of cloud infrastructure has become mainstream. And nowadays the under-utilized infrastructure is in the server farms of the cloud service purveyors, rather than in the enterprise’s own data center real estate. But under-utilized cloud infrastructure is beginning to prove costly to enterprises as recent research from Infosys shows.

Storm clouds gather over enterprise compute utilization

The dramatic cost-savings that cloud infrastructure initially offered enterprises as they shifted spend on compute power from CapEx to OpEx, by closing down private data centers and moving to the cloud has largely been realized, yet spend on cloud services continues to rise in order to enable digital business initiatives.

However, Infosys Cloud Radar 2023 found in a global survey of 2,523 IT executives that companies have on average utilized only 47% of all the cloud they have committed to, and the 12 cloud and cloud-based software providers contacted reported that more than $300 billion in corporate cloud commitments have not been used.

For example, according to the study, in 2022 Google Cloud reported a $64.3 billion revenue backlog primarily related to Google Cloud and unused capacity from customers, while Amazon’s AWS noted $110.4 billion of unused commitments, and Microsoft reported $53.8 billion in unearned revenue, mainly from contracted cloud services not yet in use. This matters, because what enterprises do not use by the end of their contracts, must either be paid for (the balance cleared) or contract terms and pricing must be renegotiated for the next contract on the basis of the enterprise’s failed initial commitment.

The challenge is that enterprises favor multi-cloud strategies in order to use the best cloud services for specific applications, avoid single vendor lock-in and improve business resilience. But multi-cloud usage has expanded in a decentralized and, frequently, siloed way to quickly enable new departmental and company initiatives. Consequently, adoption of cloud is beginning to replicate some of the bewildering and cumbersome features of legacy infrastructure.  

Managing cloud differently as it becomes more complex and pervasive

Infosys argues that there are three ways to reclaim the enterprise management of cloud infrastructure to avoid under-utilization: 

1. Master monitoring and prediction

It is no secret that technology and business leaders need to work together closely in order to properly monitor and predict cloud costs, and, this requirement becomes even more urgent as enterprises rush to take advantage of generative AI, IoT and data analytics. Each enterprise needs governance around who can buy, use and secure cloud services. But, of course, getting the balance between enabling or stifling innovation via a governance process requires the skills of a trapeze artist. And not many corporates have one of those.

2. Embed the business case into cloud

Only around a fifth of those surveyed by Infosys link cloud deployments to an approved business case. Using cloud should never just be an IT project, and much internal education needs to be undertaken to get cloud teams to understand that the one-off credit card payment for compute cycles for their project is only the tip of the iceberg of cloud cost for an enterprise.

3. Adopt a value-centric cloud operating model

Hmm, so what is a value-centric cloud operating model then? Infosys suggests that instead of setting up cloud teams around products, they should be organized around customer journeys, client experiences or value streams. According to Satsang Randhelia, associate partner, Infosys Consulting: 

You must change the way you work to maximize the value you can get from cloud. To do that you have to ask different questions: Who are the users? Do the products you offer across the stack serve their needs? How do you organize dev and ops into one team focused on value?’

Ultimately the Infosys conclusion is that to regain control of their cloud infrastructure enterprises must channel their inner hyper-scaler, via a new operating model focused on value rather than products.

My take

It is a proven adage that you cannot manage what you cannot measure and there are many tools on the market to help enterprises address their management of cloud services, from a range of vendors, including Apptio, Flexera, HashiCorp, IBM, Microsoft, Morpheus Data, NetApp, Nutanix, ServiceNow, SUSE and VMware. However, the bigger point being made by Infosys, which is a services company after all, is that what is really needed is for enterprises to rethink the culture of what they are doing with cloud. Consultants and advisors have been explaining that cloud deployment needs to be closely aligned with value for over a decade now but while cloud infrastructure offered those quick, profitable wins, not many buyers were listening. Now, however, as organizations continue to tighten their belts and find that as they renew contracts with their cloud providers, the terms may not be what they were expecting because of under-utilization, they may be more attentive.