Rocket Software: mainframes are modern critical infrastructure

Rocket Software is building a new pipeline of Australian mainframe engineers while expanding AI across its application modernisation portfolio, as banks, government agencies and other large organisations continue to depend on decades-old core systems.

The company's first Australian NextGen Academy cohort graduated in Perth this month after completing a six-month, full-time programme. All 10 graduates - eight software engineers and two technical project managers - have accepted permanent roles with Rocket Software across infrastructure, data modernisation and Asia-Pacific customer projects.

Core systems

"Globally, 40 of the largest 50 banks in the world run on mainframes. So yes, 90% of all credit card transactions run on the mainframe. Banks, financial institutions, essential government services and telecommunications," said Karla Bester, Country Manager Engineering, Rocket Software.

Mainframes are increasingly being paired with newer hardware and software capabilities rather than treated solely as legacy technology. IBM's current generation includes the Z17, while vendors such as Rocket are developing tools to extend existing applications and connect them with cloud services, APIs, DevOps workflows and AI systems.

"It is an IBM Z box. It looks like a black fridge. The latest model is IBM Z17. Every three or four years they bring out a new model, and the chips are getting smaller and faster. The new Z17 now has in-built AI for detection on the chip. We have pervasive encryption at rest. We have in-built cryptography on the hardware, and we also have quantum safety capabilities, so it's evolving," added Bester.

Rocket argues that organisations do not necessarily need to replace these systems wholesale to adopt newer technologies. Its modernisation products support COBOL and PL/I applications across mainframe and distributed environments, including approaches that retain workloads on existing infrastructure, replatform applications or refactor code for other environments.

"This is the future, and it's not the past. These core systems have been running for decades, and they are so secure. It is very, very difficult to hack into a mainframe. The data is secure. The application is secure. It runs at scale, which is incredible. For example, one box can run 282,000 credit card transactions a second while doing AI fraud detection at the same time on-chip. So it's not going anywhere because the workloads of the financial institutions are growing, so the demand is getting bigger and bigger. The footprint of the mainframe globally is growing and not decreasing at all," added Bester.

Skills gap

A more immediate challenge is finding people with the skills to maintain and modernise these systems as experienced mainframe specialists approach retirement.

"In Australia, mainframes are still essential in running critical operations in both the corporate and public sector, and will continue to play a key role for the foreseeable future. However, a significant proportion of today's mainframe workforce is approaching retirement age, so initiatives like our NextGen Academy are bridging this emerging skills gap. As organisations increasingly adopt AI and other new technologies, it's important we don't forget about the infrastructure that supports them. We need people in the workforce who are familiar with these new technologies but also understand how that critical underlying infrastructure works," added Bester.

Australia is the fifth location for Rocket's academy programme, following the US, UK, Lithuania and India. The Perth curriculum combines mainframe foundations with hands-on software engineering and exposure to AI-assisted development. Training is delivered by industry instructors and experienced Rocket engineers.

The shortage has developed over decades. Large organisations once commonly trained mainframe staff internally, but many reduced or stopped those programmes as expectations grew that workloads would eventually migrate away from the platform.

Rocket's first Perth intake has now moved into several parts of the business. Four engineers joined its infrastructure modernisation operation, four joined the data modernisation product team, and two technical project managers moved into customer projects across Asia-Pacific. All 10 roles are based in Perth.

Entry-level training is also available outside company-run academies. IBM offers free online learning that can give university students exposure to mainframe technologies and an opportunity to demonstrate foundational skills to prospective employers.

Modernisation paths

Customers are taking different approaches to their core systems, ranging from continued on-premises investment to hybrid deployments and full migration.

"That's an interesting perspective. I have a different perspective. I see customers running their core workloads on a platform that they highly appreciate. It's investing in the future of their own systems. For some customers, modernising their existing stack is much simpler than migrating everything onto the cloud or an enterprise server environment. Yet some customers do that, and that's where Rocket plays really well. We work with customers wherever they are on their journey and whatever their strategy is. The first thing we would do is understand where they are and where they want to go. Most customers, I would say, choose to modernise on-prem, and we work with them on what that means, and we have the products and tools. Other customers choose to migrate perhaps some of their core workloads onto the cloud, and then they have a hybrid solution, and we have the tools and products that help with that as well. Some customers choose to migrate, and we help with that as well, so it really depends on the customer itself," added Bester.

That range of options has become a central part of Rocket's product strategy. Earlier in August, Gartner named the company a Challenger in its 2026 Magic Quadrant for AI-Augmented Code Modernization Tools. Rocket's portfolio includes AI-assisted code analysis, business-rule extraction, dependency analysis, automated testing and validation, alongside tools for replatforming and code refactoring.

Gartner states that its publications represent the opinions of its research organisation and do not constitute endorsements of companies, vendors, products or services. It also advises technology users not to select vendors solely on the basis of ratings or designations.

AI coding

Rocket has also introduced additional AI capabilities across IBM i, IBM Z, MultiValue and Uniface environments. They include AI-powered code analysis, AI-generated unit testing, AIOps functions and AI-based replication performance optimisation. The tools are designed to operate within existing development processes and business logic rather than as separate consumer-style AI systems.

The additions include code analysis and testing functions that integrate with development toolchains including Git and Azure Pipelines, alongside AI capabilities for MultiValue and Uniface. Rocket also operates an internal AI Governance Board to oversee its policies for the technology.

For engineers working with languages such as COBOL and assembler, generative AI plays a different role from the "vibe coding" workflows increasingly associated with mainstream languages. General-purpose models can be less capable with enterprise languages, although Bester said their performance is improving.

"Yes. I'm going to reframe and not use obscure languages, but use beautiful enterprise languages. That is a really good point. The models are much better with the high-level languages. But it is improving. We're seeing a lot of improvement in that space. Our engineers will not be replaced. They're not using models to do all the coding for them. They use them to interpret the existing code base to accelerate the learning and give them some ideas," added Bester.