Fintech AI revenue to grow 960% by 2021 according

Fintech platform revenues for unsecured consumer loans issued using machine learning technology are set to see a jump of 960% during the period 2016-2021, rising to $17 billion globally in the latter forecast year, found in new Juniper Research. According to their Juniper analysts this rise is mainly driven by advances in analytics and accessible computing power.

The new study named “AI & Machine Learning: Fintech Dynamics, Disruption & Future Opportunities 2016-2021“ found that machine learning spend in Fintech will advance rapidly, owing to the highly data‑driven nature of the market, meaning that AI integration is likely to spell substantial benefits.

“Where Big Data analytics offered retrospective business intelligence, machine learning offers predictive and even prescriptive capabilities“, noted research author Steffen Sorrell. Data is key - and industries able to draw expertise from data scientists will be the first to capitalise on the AI opportunity, Sorrell concluded.

Machine learning a subset of AI (artificial intelligence) has seen a tremendous leap in activity since 2011, with substantial increases in VC and R&D investment. For example, two Fintech start-ups Kabbage and ZestFinance, have collectively raised $500 million in funding alone. Meanwhile vendors analysed in Juniper’s research have spent a total of $83 billion in R&D during 2015. Each of these vendors names AI as a part of core strategy.

Until recently, machine learning was too expensive and computationally time-intensive to break into the mainstream. Meanwhile access to extensive data sets for algorithm training were limited. Presently, the ability to use GPU (graphics processing unit) hardware for processing massive and highly available data sets, along with unlimited affordable computing power in the form of distributed architecture has opened the market to a swathe of disruptive new players.

AI is particularly useful for risk-assessment purposes, where variables from numerous financial and non-financial datapoints are assessed by algorithms to approve loans, pointed out in the new Juniper study. Complexity of the algorithms and possibility of use of AI widens the addressable market for financial institutions considerably over traditional FICO credit scoring, where lack of credit history may mean loan rejection despite a real low risk for the lender.