BNP Paribas and Google Cloud have agreed to a new five-year partnership that will expand the French banking group’s access to cloud infrastructure and Google’s Gemini AI technologies. The agreement is designed to support broader deployment of agentic AI across the bank while retaining its multi-cloud, multi-model architecture and existing controls for security and data governance.

BNP Paribas is expanding its use of cloud and artificial intelligence as part of a five-year partnership with Google Cloud, with the bank planning to use Gemini Enterprise and Gemini models to develop and deploy AI agents across selected banking workflows.

The agreement comes as large financial institutions move beyond generative AI assistants toward agentic systems capable of handling multi-step business processes. For BNP Paribas, the technology rollout is being positioned within an existing multi-cloud and multi-model strategy rather than as a move toward a single AI or cloud provider.

The bank said Google Cloud’s infrastructure and AI capabilities will support its broader technology environment while allowing individual business units to select the model, infrastructure and level of control appropriate to specific use cases.

That approach is significant for financial institutions because AI deployment involves more than model performance. Banks also need to manage sensitive customer and financial data, regulatory requirements, access controls and operational risk. BNP Paribas said certain categories of data will therefore not be stored in public-cloud environments.

The initial focus will be BNP Paribas Corporate & Institutional Banking (CIB). The bank plans to use Gemini Enterprise to build and deploy purpose-specific AI agents for selected workflows, including preparing corporate credit memos.

Additional applications are planned across sales, trading, research and structuring activities. These functions involve large volumes of financial information and documentation, making them potential candidates for AI-assisted research and workflow automation.

Credit memo preparation is particularly relevant because it typically involves gathering financial information, reviewing documentation and structuring findings for credit professionals. An AI agent could automate portions of that process, but the final credit decision remains a fundamentally different task from generating or organizing information.

BNP Paribas is also integrating Gemini models into LLM@CIB, its generative AI assistant currently available to more than 65,000 employees, according to the bank.

The partnership extends beyond the corporate and institutional banking operation. Google Cloud and Gemini technologies have already been deployed at Nickel, BNP Paribas’ payments-account business within its Commercial, Personal Banking & Services division.

At Nickel, Gemini serves as the foundation for Nickel Assist, which helps approximately 200 customer advisers find answers to customer requests using Nickel’s procedures and operating guidelines.

The deployment provides an example of a more controlled generative AI model in financial services: rather than allowing an AI system unrestricted access to corporate data, the assistant is designed around defined procedures and information sources relevant to customer support.

The next phase, involving agentic AI, introduces additional considerations. BNP Paribas said each AI agent will be authenticated and given access only to resources necessary for its assigned task. Interactions between agents and the bank’s information systems will also be monitored and controlled.

This emphasis on identity and authorization reflects a key difference between conventional generative AI applications and agentic systems. An AI model that generates a response presents one set of risks; an agent that can retrieve information, interact with enterprise systems or initiate workflow actions introduces another layer of operational exposure.

For the broader banking technology and financial technology market, BNP Paribas’ approach highlights the emerging importance of AI orchestration and governance alongside model selection. Banks increasingly have access to multiple foundation models, cloud platforms and specialized AI tools, but deploying them at scale requires a consistent framework for permissions, data handling, monitoring and oversight.

Google Cloud’s role is therefore broader than supplying a model. The partnership gives BNP Paribas access to cloud infrastructure, Gemini Enterprise and Gemini models while fitting those technologies into the bank’s existing architecture.

The strategy also reflects a competitive shift in enterprise AI. Microsoft, Amazon Web Services and Google Cloud are all competing to provide financial institutions with combinations of cloud infrastructure, foundation models, AI development tools and enterprise controls. Banks, meanwhile, are increasingly avoiding one-size-fits-all AI strategies in favor of architectures that can accommodate different models and workloads.

BNP Paribas said the partnership will contribute to its group-wide AI adoption program, including training ranging from basic AI literacy to specialized development of AI agents.

That workforce component may prove as important as the technology itself. Moving from AI experimentation to production requires employees who understand how to use AI systems, validate their outputs and work within defined governance controls.

For BNP Paribas, the five-year agreement represents an effort to scale AI while preserving flexibility over models, infrastructure and data. The challenge will be turning that technical flexibility into measurable improvements in banking workflows without weakening the controls required for a highly regulated financial institution.

Market Landscape

Banks are increasingly shifting from generative AI copilots toward agentic AI, with potential applications spanning customer service, research, risk management, compliance, software development and financial analysis.

BNP Paribas’ multi-cloud and multi-model approach reflects a broader industry preference for avoiding excessive dependence on one model provider. At the same time, agentic systems require stronger identity, access, monitoring and governance mechanisms because they can interact with enterprise systems rather than simply generate content.

The development is particularly relevant to Banking Technology Innovation, AI infrastructure, Digital Payments Platforms and Financial Technology, as financial institutions seek to integrate AI into existing operating environments without compromising security or regulatory controls.

Top Insights

• BNP Paribas will use Google Cloud’s Gemini technologies to expand agentic AI across selected banking workflows under a five-year partnership.
• Initial CIB applications include AI-assisted corporate credit memo preparation, with potential expansion across sales, trading, research and structuring.
• Gemini models will also support LLM@CIB, already available to more than 65,000 BNP Paribas employees.
• Nickel Assist uses Gemini to help approximately 200 customer advisers answer requests using established procedures and operating guidelines.
• BNP Paribas will retain a multi-cloud, multi-model strategy with authentication, restricted permissions and monitoring for AI agents.

Get in touch with our fintech expert