Venture capital is increasing its investments to fintech companies worldwide, with artificial intelligence the dominant narrative.
AI-centric fintechs captured $32 billion of global fintech funding the fourth quarter of 2024, according to KPMG. That figure is overshadowed by huge deals in companies like OpenAI and Databricks, but there are also companies raising funds for applying generative AI to labor-intensive processes like fraud detection.
Some firms are using genAI for front- or middle-office functions. These include UBS Group investing in banking assistants to personalize service, Betterment (the US robo-advisor) for optimizing portfolios, or Shopify using it for embedded payments solutions – according to recent reports by FT Partners and Silicon Valley Bank.
But the bulk of funding is going to compliance, processing, and fraud detection. And most of it is going to the US: the Americas accounted for $78.7 billion in VC investment in Q4, says KPMG, while Asia’s VC funding hit a decade low of $12.9 billion.
The future of work
Against this backdrop, Dušan Stojanović, founding partner and director of True Global Ventures, says the biggest investment theme now is AI for the back office, be it in financial services or other industries.
“The overall theme is the future of work,” he told DigFin, noting certain industries or processes are rapidly changing due to the advent of AI, starting with machine learning and neural networks, and more recently genAI interfaces.
He is also finding most of his portfolio companies in the US, but says GTV is backing Tookitaki, the Singapore-based fintech focused on compliance and money-laundering protections. He’s also backing AI companies helping insurance agents streamline their administration.
From blockchain to AI
Stojanović says the most recent fund raised by TGV, from 2022, was originally meant as a growth-equity strategy that would take stakes in the most successful investees of earlier funds. And that meant blockchain-related plays, such as metaverse The Sandbox, NFT purveyor Animoca and hardware company Ledger.
But last year the firm rebranded that fund, the $167 million TGV-5, as an ‘opportunity fund’, which means AI.
“Over the past decade we have invested in blockchain and AI, but we were more successful on the blockchain side,” Stojanović said. “We used the two positions to hedge each other. But that changed at the end of last year, when both themes began working well.”
That’s a vote of confidence in both tech stories, each of which is a prominent driver of fintech business models.
Where AI works
But Stojanović says he sees AI limited to several verticals, led by back-office and compliance functions. This may not be a sexy narrative, but he notes that after the 2008 financial crisis, banks went on vast hiring sprees in compliance, including KYC, AML, and fraud detection. These teams tended to avoid the axe in subsequent bouts of downsizing, even as compliance costs rose.
Those teams now look to be the most likely to be overhauled with the help of AI tools.
TGV is also investing in AI startups focused on front-office support, educational AI (especially around new ways of coding), coding itself, and sustainability.
Its final theme is the intersection of AI with blockchain. “Sandbox has data, but there are no analytical tools to help it value land in its metaverse,” he said. “Web3 is all siloed and lacks the tools that Web2 companies have.”
But again, the focus is likely to be on compliance, operations, and fraud detection – not on market-type activities such as tokenomics or tokenizing real-world assets.
“AI is a governance tool for Web3,” Stojanović said. “It’s good at flagging things. It makes for a better scout than humans.”
Software versus data
But what differentiates one AI startup from another? Stojanović says TGV spent time last year mapping the AI universe and found 130 companies of interest. These are companies that have proprietary data that they can leverage with AI, or they have federated AI capabilities, letting them leverage third parties’ data.
There are other criteria – a good niche, a standout team – but the data is the most valuable.
DigFin asked if “AI” was really about finding companies with a chokehold on useful data, rather than the software itself.
Stojanović acknowledges that both the data and the software are important.
“There is expertise with data power,” he said. “For example, proprietary data that are contributed from financial institutions for new fraud patterns taken in by Al in a federative learning approach.”
This is what Tookitaki relies on to spot fraudulent activity. Stojanović also cited Jus Mundi, a French startup that has a database of case law in international arbitration that enables lawyers to be more efficient at research.
Ownership or access to these specialized pools of data also reduces the risk of hallucination by large-language AI models.
But it’s not just the data, Stojanović said: “There is also meaningful expertise in the AI analytics portion, which is linked to the ability to go deep in verticlas and integration.”
He cited Tookitaki’s analytics of patterns of bank compliance teams dispelling false-positive alerts as one example.
Integrating these functions throughout a service, or into a financial institution, is also about more than just selling a better fraud-detection machine.
As Silicon Valley Bank put it, “Every fintech is, ultimately, a compliance company.” Which points to the direction that funding will take by VCs such as True Global Ventures.
