BNY introduces AI spending controls to limit unnecessary costs
The bank’s finance chief says matching tasks to suitable models will become more important as costs increasingly factor into AI use.
BNY is monitoring its artificial intelligence spending with tools that track token use and match tasks with suitable models, chief financial officer Dermot McDonogh told Bloomberg Television. He said the bank does not use costly models for low-cost queries, as it seeks to prevent overspending.
The comments represent a shift from his remarks to Fortune in July. At that time, McDonogh said the bank focused on results and did not track token consumption. He now expects the choice of model for each task to matter more as costs become a consideration.
About 400 AI agents operate at BNY, where artificial intelligence makes up only a portion of the annual US$4 billion technology budget. McDonogh said the bank uses models offered by major cloud providers rather than relying on one supplier. BNY maintains that approach despite chief executive Robin Vince having joined OpenAI’s board in July.
McDonogh also cautioned against allowing cost controls to restrict AI use. He said BNY frames its adoption around building employees’ skills and improving their capabilities, rather than efficiency or job cuts.
