CBA flags surging AI expenses as tasks grow complex and warns against low-value “work slop” infiltrating the banking system.
The honeymoon phase of simple generative AI implementations in corporate finance is giving way to a harsher economic reality. On Tuesday, the head of the Commonwealth Bank of Australia (CBA), the nation’s largest lender, issued a sharp warning to the corporate world: as artificial intelligence is deployed for increasingly complex, multi-layered operations, the financial costs of running these systems will rise in highly volatile and unpredictable ways.
Speaking at a technology briefing covered by Reuters, CBA Chief Executive Officer Matt Comyn identified managing the compounding expenses of next-generation AI as one of the definitive corporate management challenges of the decade. Concurrently, Comyn targets a growing cultural and operational hazard born from unchecked automation, heavily criticizing the rise of automated “work slop” within corporate workflows.
Moving Past Basic Prompts to Agentic AI
Over the past few years, global financial institutions have leaned heavily into basic generative AI tools to assist with text summarization, basic customer messaging, and code drafting. However, the industry is transitioning rapidly toward “agentic AI”, autonomous software agents capable of executing multi-step, reasoning-heavy financial tasks over extended periods without direct human intervention.
While these autonomous systems offer immense potential to reshape banking economics, they require significantly more processing power. Comyn stressed that while the initial licensing fees for enterprise AI models are relatively static, the actual computational consumption costs explode when systems are tasked with deep, complex analysis. This computational intensity means a single advanced query or an autonomous agent loop can cost exponentially more to run than standard algorithmic processes, cloud storage, or legacy databases.
The warnings come at a time when CBA has already positioned itself as an aggressive technological first-mover. The bank recently appointed world-renowned academic Mary-Anne Williams as its new Chief AI Scientist, signaling a deeper institutional push into proprietary artificial intelligence architectures.
The Rising Threat of “Work Slop”
Beyond the hard computing costs, Comyn pointed out a secondary, more insidious consequence of the AI boom: a degradation in the quality of internal communications and corporate documentation, which he labeled “work slop.”
The term refers to an overwhelming influx of low-effort, AI-generated content, such as over-verbose reports, bloated emails, and poorly audited summaries, that employees generate with a single click and pass on to colleagues. According to Comyn, instead of increasing productivity, this phenomenon creates an inefficient feedback loop where humans use AI to write massive documents, and other humans are forced to use AI just to condense them back down. CBA leadership made it clear that the bank will enforce stricter guardrails to ensure that AI serves to sharpen analytical accuracy rather than muddying operational waters with automated filler.
The Broader Economic Ripple Effects
CBA’s public reckoning with AI infrastructure expenses mirrors broader macroeconomic trends taking hold in Australia. According to data published in a recent CommBank Global Economic Outlook Report, the domestic pipeline for data center construction and computing capacity has swelled to roughly six gigawatts, representing an estimated $150 billion in capital investments.
This infrastructure gold rush is large enough to directly impact national accounts and GDP. However, economists warn that this continuous expenditure is adding significant demand to the economy at a time when central banks are actively battling inflation, potentially putting upward pressure on long-term neutral interest rates.
As financial giants like CBA continue to navigate this multi-year tech transition, the focus is rapidly shifting from what AI can do to how much it costs to run effectively. For the broader banking sector, Comyn’s remarks serve as a critical reminder that true technological maturity requires balancing the allure of automated scale with rigorous cost controls and a rejection of low-value digital output.
