Skin In The Game: Why Banks Are Taking Stakes In AI Labs

Rogo, a US-based financial-services AI startup, secured $30 million in funding recently. Notably, some of the world's largest banks, such as Barclays, BNP Paribas, Citi, MUFG, and Société Générale, participated in this investment, even though Rogo also received venture capital backing. Now, nine global banks collectively hold nearly $20 trillion in combined assets on Rogo's cap table.

However, this scenario is not unique; large banks and their venture investment arms are increasingly making substantial investments in AI companies. According to Evident AI's analysis of the 50 biggest global banks, such investments have grown at a compounded annual rate of 21% since 2023, with Wells Fargo, Citi, and Goldman Sachs being the most active US bank investors.

Deals around the world and in India continue to support this trend. Citigroup and Spain's Santander have invested in Tokyo-based Sakana AI, which develops foundational models and custom AI for financial services. In India, HDFC Bank and Canara Bank have invested in CoRover, while IDFC Bank has partnered with Sarvam to establish an R&D lab aimed at creating the world's first self-improving bank.

Three factors appear to be driving banks away from traditional vendor relationships. Firstly, deep integration into regulated, high-stakes workflows necessitates greater trust and customization than a simple licensing agreement can offer. Secondly, owning a stake or jointly developing the research and development (R&D) provides banks with influence over the roadmap they have become increasingly dependent upon.

Lastly, for BFSI institutions possessing vast proprietary data and extensive distribution, contributing that data to co-developed AI systems can be more valuable than a mere services contract. According to Mukesh Pandey, founder of consultancy firm Rupyaa Paisa, "License provides a means to access technology, while investment or co-development may result in a more profound collaboration between a bank and a startup.

Cocreation gives the bank an ability to affect product architecture and workflows instead of simply fitting the bank's processes into a ready-made solution." It remains to be seen whether this will become the standard model for bank-AI relationships or if it will remain concentrated among the largest, most AI-forward institutions.

The outcome for smaller AI startups negotiating with a bank that is simultaneously their client, investor, and co-developer remains uncertain. While banks without any bank partnerships or strategic investments may struggle to access the same level of data, distribution, and trust, the banking sector is increasingly becoming more than just a customer; it is also becoming a financial stakeholder in the companies building the AI technology.