BNY Mellon has developed internal systems to monitor its artificial intelligence token expenditure and intelligently distribute workloads across different models, according to chief financial officer Dermot McDonogh in remarks to Bloomberg Television. Operating roughly 400 AI agents within its infrastructure, the institution allocates $4 billion annually to its technology operations. The financial sector faces mounting pressure as expenses tied to AI deployment continue climbing.
The bank's approach centers on preventing unnecessary overspending as AI becomes increasingly central to operations. McDonogh explained the bank's methodology: "We've become quite disciplined and developed some sophisticated telemetry so we can track how we're spending and how we're using our tokens," he told Bloomberg Television.
A key component of BNY's strategy involves matching computational tasks with appropriate models based on complexity and cost. "We don't use expensive models for cheap queries," McDonogh stated. He emphasized that this alignment between task requirements and model selection will grow increasingly important "over time as costs come into play."
Rather than committing to a single vendor, BNY draws on models from major cloud infrastructure providers, maintaining flexibility across its AI infrastructure. This independence persists despite BNY chief executive Robin Vince joining OpenAI's board in July. The AI expenditure represents only a portion of the bank's broader technology budget.
Across the financial sector, institutions confront a counterintuitive challenge: while the cost per individual token has declined, overall AI bills continue rising because AI systems consume substantially larger volumes of tokens. Commonwealth Bank of Australia's leadership flagged escalating expenses in June, while Atlassian implemented monthly spending caps for its engineering teams in July.
McDonogh's recent comments represent a notable shift in messaging from BNY. During a July conversation with Fortune, he indicated the bank did not monitor token consumption, stating "It's not something we spend any time talking about." He maintained that cost management should not constrain innovation. "We don't talk about it in relation to efficiency or job cuts," McDonogh said. "We talk about it in relation to upskilling and trying to do better things."
Other financial institutions have adopted more transparent stances regarding AI's workforce implications. Citi's chief executive Jane Fraser disclosed in July that the bank has already begun experiencing job reductions attributable to AI implementation.
Source: The Next Web
