Six months into a deal with OpenAI, Spanish bank BBVA says staff created 2,900+ GPTs and reported productivity gains, but questions ChatGPT's scalability
Isabelle Bousquette / Wall Street Journal :
Context & Ripple Effects
BBVA is an early enterprise test of OpenAI’s corporate offering, following OpenAI’s initial roster of 260 enterprise customers and broader evidence that large companies were adopting its tools.
The bank’s experience adds operational detail to OpenAI’s reported reach across major businesses: employee-led creation can produce many task-specific tools, while deployment at scale remains a separate test.
First-order effects
- BBVA employees gain a large internal library of GPTs and report productivity improvements from using them.
- BBVA’s scalability concerns make the value of its OpenAI arrangement contingent on whether those tools can be deployed and maintained beyond early use cases.
Second-order effects
- OpenAI faces pressure to convert broad enterprise access into repeatable, scalable use across a customer’s workforce rather than isolated employee experimentation.
- Enterprise buyers are likely to evaluate AI tools on dependable workflow rollout and ongoing usefulness, not simply on the volume of custom GPTs created.
Third-order effects
- If this pattern persists, enterprise generative AI will be shaped less by initial adoption counts than by whether organizations can operationalize employee-built tools across workflows.
- The balance of leverage may shift toward large model customers that can demonstrate productivity gains but demand scalability before expanding commitments.
The trend: Enterprise AI is moving from broad employee experimentation toward proof that customized tools can scale into durable workflows.