It will come as little surprise to readers of The Stack that our latest State of Java Survey and Report has confirmed there is continued underutilisation of Cloud compute capacity, writes James Johnston, VP of EMEA, Azul.
In the global study, 71% of say that more than 20% of their cloud compute capacity is untouched. At a time when organisations continue to face significant strain on IT budgets, while the IT department is still expected to support innovation, over allocating cloud resources is costly.
Working with large enterprises processing millions of transactions every day we witness firsthand the importance of sweating your IT assets. Optimising IT systems is key to performance, because it is no longer feasible to keep adding server capacity; even those with the largest IT budgets are being pragmatic with IT spend, not just to be cautious, but because we’re just entering a new era of innovation that may continue to strain cloud budgets by an order of magnitude.
Yes, I’m talking about AI. Everywhere you look, the major cloud service providers, chip manufacturers and the largest AI vendors are scrambling to secure more capacity to feed the ever more compute hungry beast that is artificial intelligence. 72% of our survey respondents acknowledge that their compute consumption will have to grow to support Java applications with AI functionality. The challenge is that not every company can afford to invest heavily in more processing capacity, so they must find alternative ways to be more efficient.
See also: The Big (Easy) Java Migration
So, what is the solution? According to our survey there is no one answer. The top response across all the countries we surveyed (except France) are establishing internal rules for cloud asset management. In France the emphasis is on renegotiating cloud contracts. Globally, the second choice is to focus on newer, more efficient compute instances and processors, which is interesting as it suggests the likes of AWS Graviton and ARM processors are becoming a credible alternative hardware solution for efficiency and speed. The UK and France want to work more closely with cloud providers to better understand their tools and best practices, while Germany wants to explore FinOps techniques such as dynamic and schedule-based autoscaling. The UK also wants to look at using a high-performance JDK to improve performance.
Other costs related to Java
Java is a significant proportion of the application and infrastructures of large enterprises, leading organizations to not just look for ways to optimize their cloud spend but their overall cost of ownership.
For example, Oracle began charging for Java nearly five years ago creating an unanticipated cost for many organizations. While Oracle offers a free period of use for newer version of Java, eventually licensing obligations kick in. This occurred recently for Java 17 this past September which is a considerable concern for the 34% of organizations that are using Java 17 according to the State of Java Survey & Report. There are also still customers on Java 6 (10%) and Java 7 (13%), legacy versions which are no longer supported by Oracle altogether creating huge risks for organizations that are not partnered with an Oracle alternative solution provider.
Juggling multiple versions of the Java Development Kit (JDK) can be a distraction for your engineering team, which is underlined by the admission that 62% of our survey participants say dead or unused code affects DevOps productivity. 33% of participants also say that more than 50% of their DevOps teams’ time is wasted on Java-related security vulnerability false positives. Java will have been around for 30 years this year (May 2025), so it is easy to forget how much Java code has been built up in that time. We should not be surprised that there is dead code floating about.
But now imagine the time it consumes for your developers if they are distracted by dead code or have to worry about security false positives, rather than concentrating on how best to optimise applications and migrate them to the cloud. It is no wonder that a fifth of cloud compute capacity is going unused.
Add to this, 41% of users expressed concerns over encountering critical production security issues within their Java ecosystem on a weekly or daily basis. These are all unwelcome distractions that prevent systems engineers and IT operations from improving the efficiency of their Cloud environments.
