Commonwealth Bank of Australia (CBA) has completed the migration of its entire data platform and core banking system to Amazon Web Services (AWS), marking one of the largest financial cloud transitions undertaken in the Southern Hemisphere. The move positions the bank to scale its data and AI operations while modernising its infrastructure for faster development and improved resilience.
According to the bank, the migration of its enterprise data platform began in mid-2024 and involved transferring more than 61,000 data pipelines to AWS in partnership with HCLTech. The project, described as one of CBA’s most significant technology transformations, was executed without compromising data integrity or performance, according to the participants.
Terri Sutherland, CBA’s lead for data platforms, said the migration underpins the bank’s ability to deliver highly personalised services using AI. “We’re now making around 55 million AI-driven decisions every day across more than 2,000 models built on 157 billion data points,” she said. “The cloud platform gives us the flexibility and scale needed to continue that growth.”
Dynamic scaling
The bank said the move simplifies its technology stack, reduces on-premises infrastructure, and enables dynamic scaling in response to demand. CBA also cited greater alignment with AWS’s standards for resilience, security, and compliance – a key consideration given the bank’s regulated operating environment.
In a recent interview with The Australian Financial Review, CBA’s group executive for technology, Gavin Munroe, confirmed that the bank’s core banking system – covering the ledgers for deposits, loans and transaction processing – was moved into AWS-managed data centres late last month. The 18-month project, which replaced CBA’s mainframe infrastructure, makes it the first of Australia’s major retail banks to fully migrate its core to the public cloud.
Munroe described the transition as a “cornerstone for our enterprise transformation,” adding that “innovation and velocity are faster on the cloud.” He said the migration should deliver cost outcomes that are “neutral or slightly better” than the bank’s legacy infrastructure, while improving speed, safety and product development agility.
CBA worked closely with the Australian Prudential Regulation Authority (APRA) to ensure the migration met local regulatory requirements, including keeping banking data in AWS data centres located within Australia. AWS operates multiple domestic facilities connected to different power grids and networks, providing geographic redundancy and eliminating single points of failure.
The shift also brings CBA’s data lake, previously hosted on Snowflake within AWS, into closer proximity with its core systems, which Munroe said will improve AI performance and efficiency. “When you have an agentic framework and expectations to interact with the bank in almost real-time, this requires a lot of performance tuning and architectural discipline,” he told the AFR. CBA plans to migrate its final legacy system, which manages credit card operations, to AWS within the next year.
Project Coral and AI
Beyond the cloud migration, CBA has been committing to embedding AI in its operations. In August it published further details of its internal AI strategy, highlighting a dedicated “AI-Powered Engineering” effort and a strategic partnership with OpenAI. It also disclosed its work on agentic AI under the auspices of Project Coral, an agentic AI framework designed to act on tech debt – code that slows developers down and makes future development more difficult.
The system integrates with existing monitoring tools to detect issues, analyse code, suggest improvements and even generate pull requests with proposed fixes – though human review remains mandatory. The architecture is modular and agnostic to specific AI toolchains, allowing flexibility in deployment across the engineering organisation.
Brendan Hopper, the bank’s CIO for technology, and distinguished engineer, stated that the goal is to offload repetitive, manual tasks from engineers so they can focus on high-value work.
This engineering AI initiative complements broader generative AI ambitions. The bank and OpenAI declared a multi-year collaboration to explore generative AI in areas like fraud and scam detection, and to personalise customer experiences. Employees will progressively gain access to OpenAI’s enterprise AI tools, including ChatGPT Enterprise, supported by internal training and upskilling programs to embed responsible AI practices broadly.
The bank also announced an expanded strategic partnership and investment in AI company Anthropic in March this year.
By placing core banking operations and data proximity in the cloud, CBA reckons it is positioning its AI models – both agentic and generative – to operate with minimal friction. The alignment reduces latency, eases integration, and centralises governance of compute and data assets.
That said, the transition is not without challenges. Agentic AI frameworks require strong safeguards, audit trails, model provenance and human oversight. The migration also raises pressure on AWS and CBA teams to maintain consistent performance under live, unpredictable workloads.