Why is Commonwealth Bank of Australia (ASX:CBA) warning about artificial intelligence costs?

CBA Chief Executive Matt Comyn warned that enterprise AI expenses are scaling exponentially rather than linearly as businesses transition to highly complex tasks. Corporate AI users pay using "token-based" pricing models, meaning costs compound heavily when models are utilized for deep reasoning, tool integration, and processing massive context windows.

How do corporate AI expenses differ from retail consumer AI costs?

Retail consumers typically access AI tools via free tiers or fixed-cost monthly subscriptions. In contrast, enterprise users face variable, volume-based pricing determined by the number of tokens (units of text or data) processed, which makes corporate AI budgets far more volatile and unpredictable as workloads scale.

What structural constraints is corporate Australia facing with AI rollouts?

Beyond direct software and token costs, Australian companies are navigating significant structural bottlenecks. These include managing widespread workforce disruption, alongside mitigating the immense energy and water resource demands that modern data centres place on local Australian infrastructure.

What is CBA's current corporate strategy regarding AI adoption?

Despite flagging cost concerns, CBA is aggressively positioning itself as a leading AI adopter in the banking sector. The bank recently hosted a dedicated AI summit featuring OpenAI CEO Sam Altman and hired Australia's first chief AI scientist at a commercial bank to spearhead its technological rollout.

What does Matt Comyn mean by the term "work slop" in corporate AI?

"Work slop" refers to the massive proliferation of low-value, automated outputs such as generic AI-generated slideshows, emails, and Word documents. Comyn suggests that skyrocketing token costs could actually benefit corporate productivity by forcing companies to filter out this low-quality digital clutter.

Image for illustrative purposes only. Not a real photo.

CBA warns of surging, unpredictable AI costs

Written by Liezl GambePublished Jun 2, 2026, 6:15 AM

The Commonwealth Bank of Australia (ASX:CBA) issued a stark warning over the soaring and unpredictable costs of artificial intelligence as businesses transition to highly complex tasks.

Speaking at an Australian Financial Review conference, CBA CEO Matt Comyn labelled AI expenditure a critical emerging management challenge for corporate Australia, predicting a global tightening of AI budgets throughout 2026.

Unlike retail consumers utilising fixed-cost software, enterprise users face compounding financial pressures due to "token-based" pricing models, where costs scale exponentially rather than linearly.

Comyn noted that while early, simplified adoptions kept token costs relatively modest, the latest generation of advanced AI models demands heavy computational power for deep reasoning, tool integration, and processing vast amounts of context.

This fiscal hurdle arrives alongside broader structural constraints on local rollouts, including intense workforce disruption and the severe energy and water burdens that data centres place on Australian infrastructure.

The nation's largest lender remains positioned in the sector, recently hosting a high-profile summit with OpenAI's Sam Altman and appointing the bank's first chief AI scientist.

Comyn suggested that rising financial barriers might offer a silver lining by naturally curbing the proliferation of low-value, AI-generated output—a phenomenon he termed "work slop".