AustralianSuper weighs AI model options as usage costs rise

AustralianSuper weighs AI model options as usage costs rise
AustralianSuper eyes new AI

Australia’s largest pension fund is reviewing alternative artificial intelligence tools after a sharp increase in usage costs strained its internal resources. The assessment includes open weight models such as China-linked DeepSeek and Moonshot AI’s Kimi, with any eventual deployment likely to stay separated from member data and production systems.

Highlights

  • AustralianSuper is evaluating open weight AI models after exceeding its coding token allowance in three days, aiming to reduce escalating usage costs.
  • Head of AI Sarah Carney confirms the review includes Chinese models like DeepSeek and Moonshot AI’s Kimi, with plans for use in internal chat and cyber defense testing.
  • The strategy reflects a sector trend of isolating experimental AI from member data to manage efficiency, costs, and regulatory risk for pension funds and financial institutions.

AI model review amid rising internal demand

As reported by Bloomberg, AustralianSuper is assessing a range of open weight AI models as it looks to contain costs after running out of coding tokens within three days last month.

Sarah Carney, the fund’s head of artificial intelligence, says the review includes Chinese models such as DeepSeek and Moonshot AI’s Kimi. She says any model ultimately selected would likely be ring-fenced from member data or production systems and used for internal chat tools as well as cyber defense testing.

Operational and sector implications

The approach shows how large financial institutions are balancing the potential efficiency gains of AI tools against the expense of heavy usage and the risks tied to sensitive data environments.

For pension funds and other regulated investors, keeping experimental or lower-cost AI models isolated from core systems can help expand testing while limiting exposure to member information and critical operations.

In a previous article, we explained how the US AI data center boom faces financing, construction, and power-grid constraints.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.