Digital Lending, $100mm+ fund raises and ChatGPT | Fintech Inside Edition #70 - 20th Feb, 2023
Serving hot takes on Payment Aggregator application status, fund raises in Indian fintech sector, an Interesting SPAC by an Indian fintech startup and Generative AI+Fintech.
Hi Insiders, I'm Osborne, Principal at Emphasis Ventures (EMVC).
Welcome to the 70th edition of Fintech Inside. Fintech Inside is the front page of Fintech in emerging markets.
This past week, India's central bank - RBI, released FAQ's on the Guidelines on Digital Lending and the status of applications for Payment Aggregators. Three fintech startups in India and SEA raised $100mm+ in funding EACH. There was also news of a little known Indian insurance startup that is going public via a SPAC! And lastly, can generative AI benefit the financial services industry?
All this and more in today's edition of Fintech Inside. Hot takes in this one, you don't want to miss! :)
If you're reading this on Substack.com you would have seen I've enabled "Pledges". It's an experiment I'm running and all editions of Fintech Inside will be free to read forever. A Pledge is a good way for you to show your support if this newsletter has ever added value to your understanding of the fintech sector.
Raising funding for your early stage fintech startup? reach out to me at os@em.vc
Enjoy another great week in fintech!
3️⃣ Fintech Top Three
1️⃣ The RBI is back, this time with more clarity
RBI (India's central bank) released FAQ's on its recent Guidelines on Digital Lending (GDL) and it released the current status of applicants for Payment Aggregator license.
Takeaways:
Increased clarity on digital lending: After last week's kerfuffle on the ban on digital lending apps, this week, RBI came out to shift the conversation, offering more clarity on digital lending operations. IMO, these are some really well thought out guidelines, but I wonder why these guidelines are only applicable to Digital Lending? Why not all lending? Some of the FAQ answers that stood out for me:
Digital lending need not be entirely digital (qts. 1): The RBI states that lending undertaken "largely by use of seamless digital technologies" will be governed by the Sept, 2022 Guidelines on Digital Lending.
APR must include all charges (except penal charges) (qts. 5, 6, 14, 15): Borrowers should be informed of the Annual Percentage Rate (APR) at the time of sign up and on an ongoing basis (in case of floating rates) and must include cost of funds, credit cost, insurance costs and so on. Penal charges (payment bounce, late payment fees) should not be annualised but should be disclosed at the time of on boarding and the maximum penal charges can only be up to the amount under default.
LSP's and other intermediaries cannot be in the flow of funds (qts. 7, 8, 9, 10, 11): In answering several questions, RBI repeatedly makes it amply clear that intermediaries and Loan Service Providers (that are not regulated) cannot be in the flow of funds i.e. LSP's (except payment gateways) cannot pool lending/borrower funds, cannot automate lender/borrower bank accounts, cannot act as a gateway/switch and so on. LSP's are limited to one thing and one thing only - customer acquisition channels. If you're a fintech founder - do what you have to, to get a license.
Payment gateways facilitating digital lending are now loan service providers (qts. 8): The only exception to the previous point, payment aggregators that facilitate loan payouts, collections and other services to the lender will have to comply with these GDL.
Disclose collection agent contact information before the borrower is contacted (qts. 17): This is a very important piece of lending that the RBI and Ministry of Finance is most worried about - collection agencies sending thugs to collect, leading to suicides and other consequences. By making the lender disclose the agent's contact information before the borrower is contacted, is a step in the right direction. Except it's kinda tough to practically execute.
Clarity on Payment Aggregator (PA) applications: After close to a year of playing a game of whodunnit, RBI finally released the list of applicants for the PA licenses and the related status. There were 139 applicants in total - 54 existing PA's with some having restrictions, 28 new PA's but not allowed to operate yet and 57 PA's whose applications were returned/withdrawn. Below is a very high level summary of the applicants: (updated as on 20th Feb, 2023 at 10PM)
Existing PA’s with some restrictions: Razorpay, Cashfree etc.
Existing PA’s without restrictions: Digio, Google, Stripe, Pine Labs, Infibeam, Billdesk, Innoviti and others.
New PA’s: Juspay, Groww, mSwipe and others.
2️⃣ Three $100+mm funding rounds were announced in India and SEA and three acquisitions in India
InsuranceDekho, an insurance product aggregator, raised $150mm.
PhonePe, Walmart-owned payments company, raised $100mm.
Aspire, a Singapore and Indonesia-based small business banking startup, raised $100mm.
India Acquisitions:
Roadzen, an Insurance as a SaaS startup, was acquired by Vahanna, a SPAC, and is expected to list on the NASDAQ at a valuation of $965mm.
Kotak Mahindra Bank acquired Sonata, micro finance company, for INR 537cr ($66mm).
InCred acquired Rain, a earned wage access startup.
Takeaways:
Funds are chasing quality startups: In this polar environment of a funding winter with record dry powder lying with VC/PE funds, it's clear that larger pools of capital will be invested in default alive/market winning companies. The margin for error with these funding rounds will have to be really small. Other than PhonePe ($12bn valuation), it's unclear what the valuation of InsuranceDekho and Aspire are, but the investors there would have got a pretty good deal.
Found it interesting that this week alone, it was announced that Sequoia led investments in two small business fintech startups: Aspire (Singapore and Indonesian digital banking) and Shift (Australian small business lending and payments).
That one surprising SPAC: Out of nowhere, Delhi and New York based Roadzen, a little known insurance SaaS startup, announced that it was going public via SPAC on NASDAQ at a $965mm valuation. Great on the founders, investors and team for going through with this given SPACs have largely fallen out of favour. Founded in 2015, Roadzen raised a total of $10mm (per CBInsights) in funding, has ~250-300 employees and most importantly, did $9.8mm in total consolidated revenue as on 31st March, 2022 and expects to close 31st March, 2023 with $59mm in total revenue. I have so many questions on this one, but it's not my place, so I'm going to leave it at that for now.
3️⃣ ChatGPT has touched down on Indian fintech startups
Yubi, a wholesale credit marketplace, launched YubiBERT, an open source, fintech focused large language model, based on Google's BERT.
Velocity, a revenue based finance startup, launched a ChatGPT-powered AI assistant.
Takeaways:
Early days for AI in all segments of technology: We are at the bleeding edge of the second wave of AI - not just in India but globally. This second wave is different from the 2016/2017 wave of AI in that one uses Large Language Models (LLM's) to generate text and content within seconds while being aware of the context. The earlier wave was largely a massive step forward for natural language processing. The concept of large in LLM's is also vastly different and these language models could be trained on datasets large enough to include nearly everything that has been written on the internet over a large span of time.
Generative AI and Fintech: Generative AI being applied in fintech is even more on the bleeding edge of AI. Globally, everyone's still figuring out what generative AI can be used for in fintech. The sharp minds at Bain Capital Ventures have a good thesis on how generative AI can be applied for the broader financial services industry - highly recommended. Personally, I think, we are in the fad stage and we are too early in that arc of figuring out how generative AI can benefit the financial services industry but we need experiments by excellent founders to get us there.
YubiBERT and Lexi are experiments: YubiBERT is an Indian, fintech-focused LLM, trained on 50GB (reduced from 220GB) of "news, pdfs, reports, wiki, speech-2-text data & transliterated text". The use cases of YubiBERT, as per the data science team, is to (a) tokenise (classify) and index any Indian fintech text/document, (b) replace GPT-Tokenizer or BERT-Tokenizer in LLM's as YubiBERT is trained on Indian fintech data, and (c) to process Indian languages. Lexi by Velocity on the other hand is a direct application of ChatGPT in the form of a chatbot. Lexi is integrated into Velocity's Insights product where business owners can glean insights on their business. I imagine that Lexi forms as a different interface (chat rather than dashboards) but business owners to ask questions and get answers based off data on their business. Velocity has always had a quick iteration, experimentation culture, hope this one pays off.
🌏 International
The US SEC finalised rule changes to cut the settlement cycle to one business day. US banks are backing away from crypto companies and re-evaluating any exposure to the sector. Discover card network reported rising credit card delinquency rate.
🏷️ Other Notable Nuggets
🎵 Song on loop
Fintech updates can get boring, so here's an earworm: Can't Get You out of My Head (Cover) - AnnenMayKantereit x Parcels (Youtube only). Remember Kylie Minogue? This is a beautiful take on that classic song by her.
👋🏾 That's all Folks
If you’ve made it this far - thanks! As always, you can always reach me at connect@osborne.vc. I’d genuinely appreciate any and all feedback. If you liked what you read, please consider sharing or subscribing.
1-min Anonymous Feedback: Your feedback helps me improve this newsletter. Click UPVOTE 👍🏽 or DOWNVOTE 👎🏽
See you in the next edition.