The French banking group leverages Google Cloud and Gemini AI to streamline operations and drive returns.
French banking giant BNP Paribas SA is tying its artificial intelligence investments to well-defined financial targets, most recently projecting $853 million in value creation through its newly announced partnership with Google Cloud.
The assigned value is not always fully declarative, said Marc Camus, group chief information officer of BNP Paribas, during a Sept. 24 press conference.
“There is a finance committee which will use different submissions from the different entities and functions. There is a check and challenge by the finance function on these value creation figures. And when I say value creation, it does not necessarily mean cost savings. It can be revenue generation, for instance, through pricing optimization techniques. It can indeed be pure cost savings by optimizing processes.”
“The bank views AI as a P&L,” said Charles Holive, managing director and chief AI officer for BNP Paribas’s corporate and institutional banking business. ”Every program we decide on, especially the big transformation ones, is tied to specific financial targets. Those could be revenue, could be cost savings or risk avoidance.”
Google Cloud
Under the terms of the five-year deal, the bank will have access to Google Cloud—the cloud computing division of Google parent Alphabet Inc.—and Gemini Enterprise’s library of more than 200 open-source AI models, which BNP Paribas can develop into purpose-built AI agents.
“This technology pairing is going to be supported by a group-wide AI enablement program, and that’s covering everything from foundational AI literacy right through to specialized agent development,” said Georgina Bulkeley, managing director and global industry lead of financial services at Google. “It’s a really important cultural aspect of these kinds of partnerships.”
Early deployments target BNP Paribas’s CIB business, which has extended AI access to its 65,000 employees, focusing on middle- and back-office processes like credit memo generation and improving the know-your-client processes.
The announcement comes four months after BNP Paribas inked a three-year extension to a 2024 groupwide agreement with Mistral AI to design and develop generative AI offerings based on the vendor’s large language models that meet the bank’s operational and regulatory requirements.
Camus said that this was clearly in line with what BNP Paribas has been doing in the past few years. “We don’t necessarily stick with just one partner,” he explained. “Depending on the use case we are working on and depending on the business requirements, we will go for some on-premises solutions and some public cloud solutions.”
Reallocating Human Capital
Camus said that the agreement may lead to some change in headcount within the bank, but that has been an ongoing result from implementing process automation for many years. The bank, however, looks to bulk up its workforce with more AI and data experts, he added.
“We need to invest in the workforce in these areas,” said Camus. “There is also a need for change management in a number of teams. For example, IT developers need to get used to working with AI agents.”
Rob Daly covers economics and fintech. Contact him at rdayly@gfmag.com.