What's Going On In Fintech?

Hi, and welcome to another episode of What's Going On in Banking. I'm your host, Ron Shevlin, chief research officer at Cornerstone Advisors and author of the Fintech Snark Tank blog on Forbes.

 

We're usually talking about what's going on in banking, but today we're going to narrow the focus a little and talk about what's going on in fintech. To help me do that, I've got Nigel Morris, co-founder and managing partner of QED Investors, one of the premier fintech venture capital firms in the space.

 

Before QED, Nigel co-founded Capital One, where he served as president and chief operating officer. Nigel, I have to tell you, my podcast bucket list had two names on it: Nigel Morris and Jamie Dimon. So I get to check one off. Thanks a lot for joining me today.

 

Ron, it's a pleasure. I'm not sure about "premier venture capital firm," but it's nice to hear you've heard of us.

 

And look, Jamie Dimon is in a class of his own. I don't know if you read the introduction to his earnings report last week, but if he's not the most accomplished and perceptive banker of my generation, I'm not sure who is. He's a real class act. So I'm delighted you put me in the same category, even if I'm not sure it's deserved.

 

Absolutely. I did read it and commented on it because his remarks about AI were interesting. I was also surprised that he mentioned only two major areas of technology this year. Usually, he talks about technology at length. One was cloud computing, which he's discussed before. The other was AI.

 

What really caught my attention was JPMorgan's plan to hire a chief data and analytics officer. He didn't call it a chief AI officer. He framed the role around data and analytics, which I thought was interesting. What stood out to you?

 

What struck me was how much energy he put behind AI and the idea that it could create a major step change in productivity and in how financial institutions serve customers.

 

If you take a step back, I think we're going to see very substantial cost reductions as companies use AI to drive down their efficiency curves. At its core, AI is an advanced statistical tool. That's why I found the use of the term "data science" rather than simply "AI" interesting. In many ways, that's what we're really talking about.

 

Financial-services companies, including Capital One and JPMorgan, have been using advanced analytics for a decade or more. So the underlying idea isn't completely revolutionary. What's changing is the magnitude of the impact.

 

We're starting to see substantial cost reductions, especially in operations and contact centers. Companies like Albert, which is in our portfolio, and many others are using chatbots to reduce costs and improve efficiency.

 

The second opportunity is one-to-one marketing. At Capital One, we used to say we were able to surgically offer the right product to the right customer at the right time. The truth is that we couldn't really do that because we were largely using direct mail.

 

In a digital environment, where you can tailor an offer in real time, true one-to-one marketing becomes much more realistic. That means less noise for the consumer, a lower cost to acquire, and, all else being equal, better economics.

 

As these tools become more common across financial services, the institutions that learn to use them effectively will gain an advantage. Those that don't will be at a substantial disadvantage.

 

A lot of this isn't about developing your own AI. It's about execution. How do you take advantage of the tools that are becoming available across the industry over the next two, three, or five years? The world is changing quickly, so there is real value in having people focused on this now.

 

I think that's what JPMorgan is doing. They've spent years building the infrastructure. Now the focus is on leveraging it and executing.

 

Nigel, you come out of the credit and lending world, and you've got a number of portfolio companies focused on that space. Do you have concerns about bias or other downsides when machine learning and AI are applied to lending and credit decisions?

 

I do. Some of the people I work with tell me I'm still a bit of a Luddite. The Luddites, if you remember, were anti-Industrial Revolution workers in the U.K. who would throw their clogs into machines.

 

I genuinely worry about using very advanced statistical tools to predict credit risk. Once the machine takes over, you can't simply ask it why it made a decision. You also can't easily ask how it traded off one variable against another.

 

With traditional regression analysis, you can often say, "These are the variables, this is how they're weighted, and this is how they interact." As the models become more complex, it becomes increasingly difficult to unravel exactly what variables are driving the outcome and how they are being used.

 

That creates a problem with generalizability. The model is built on historical data, but the future may not look like the past.

 

I've always said we should build the basic models with relatively simple tools that people can understand and where the causal relationships are clear. Then you can layer more advanced tools on top of that. But I don't think you should ever let the machine completely take over.

 

Otherwise, you risk ending up in a dystopian environment where you don't understand why the model made a decision and can't tell whether it will continue to behave correctly as conditions change.

 

And conditions are always changing. Interest rates can move sharply. Student-loan payments can restart after a long pause. Models may have been trained in an environment where those things weren't happening.

 

So don't let the machines take over completely. You still need sensible people applying judgment and putting art on top of the science.

 

Great point.

 

You've been traveling quite a bit recently. Are there things you've seen outside the United States that are missing here, or approaches that U.S. banks and fintechs should be paying attention to?

 

That's a great question, and it's one we think about constantly at QED.

 

Going back to my Capital One days, I've always believed that if an idea works in one geography, the default assumption should be that it can probably be tailored, improved, and adapted to another geography.

 

Nubank, in many ways, was a digital version of the Capital One model in Brazil. There will be a Nubank equivalent in India. Hopefully it will be OneCard or Jupiter, both companies we've invested in. One day there may be one in Nigeria.

 

Ideas can move across geographic borders more easily than people sometimes assume. The barrier often looks higher than it actually is.

 

We're seeing a lot of innovation emerge in places like India and Brazil, where the regulatory climate in some areas is more progressive than it is in many Western markets, including the U.K. and the U.S. Systems such as Pix and other low-friction payment infrastructures are accelerating innovation.

 

We're also seeing underserved populations receive better banking services through new relationships between banks and fintechs.

 

In the U.S., I've often said that the traditional banking system is unwilling, or structurally unable, to serve a large portion of the population. Many consumers live paycheck to paycheck and don't hold significant balances. Others have FICO scores below the threshold where traditional institutions are comfortable extending credit.

 

For that group, the traditional banking system often isn't particularly interested in serving them.

 

Banks also carry expensive fixed-cost distribution systems. JPMorgan Chase and Capital One have moved aggressively toward digital, and some regional banks such as Key and Fifth Third have done the same. But across the thousands of banks in the U.S., digital adoption has not kept pace with consumer behavior.

 

Developing markets sometimes have an advantage because they can leapfrog the old infrastructure. They can go straight to digital in the same way many countries skipped landlines and went directly to mobile phones.

 

That's creating an extraordinary amount of innovation in developing markets.

 

Another factor is talent. The United States has educated a tremendous number of people through business schools and firms such as Goldman Sachs, Bain, and BCG. Many of those people gain experience here, recognize an opportunity, and then return to their home countries to build new models.

 

Nubank is a great example. David Vélez had exposure to Sequoia, General Atlantic, Stanford Business School, and the broader U.S. ecosystem. We're seeing more people take that experience and build companies in other markets. It's very exciting.

 

I don't think the problem is limited to consumers. Small businesses are also dramatically underserved. I recently came across an estimate that, using the FDIC's definition of underbanked or underserved, nearly 80% of small businesses in the U.S. could qualify.

 

Are you seeing innovative approaches to small-business banking or lending outside the U.S. that could be brought here?

 

You're absolutely right. Traditional U.S. banks will usually lend to a small business only if they can secure the loan, often against commercial or residential real estate.

 

If you're a startup, a new business, or simply don't have two or three years of financial history, you can knock on the bank manager's door all day and still struggle to get a loan.

 

Companies such as OnDeck Capital tried to deliver small-business credit digitally. OnDeck was acquired by American Express and had some success, but it still only scratched the surface of the opportunity in SMB lending.

 

A major opportunity comes from proprietary data. If you already provide another service to the small business and have permission to use the data generated through that relationship, you can begin doing things like revenue-based lending.

 

Wayflyer in Dublin and Capchase in the U.S. are examples. They can make funding decisions based on future revenue. Frankly, this is a business banks could have been in. Banks often already have the checking relationship and the benefit of low-cost deposits.

 

Yet traditional banking has not engaged deeply enough in this opportunity.

 

We're seeing similar models emerge in many geographies. In Mexico, for instance, lenders can access government tax data from centralized systems. That becomes a kind of system of record, similar in some ways to credit-bureau data, that can be used to make better lending decisions.

 

Companies such as Konfío are operating in that market. Creditas is doing other forms of lending, including auto lending. In Nigeria, Moniepoint is using its own data to serve businesses.

 

SMB lending remains a huge untapped opportunity. It requires more dexterity than many traditional lenders have been willing to bring to the market.

 

Banks also tend to rely heavily on the personal credit history of the business owner rather than the operating performance of the business itself. Part of the problem is that small-business data is harder to access and harder to verify. But that is beginning to change.

 

It seems to me that part of the opportunity is for financial institutions to integrate with vertical SaaS providers that already serve specific small-business industries. That also brings us into embedded finance.

 

I know you've invested in companies that operate in vertical software markets. Are you bullish on that area? And do you see it helping to revive embedded finance, which has taken some hits recently?

 

I think the space is fascinating.

 

If you take a step back, almost anything in a company that still relies on analog processes and pieces of paper is going to become digitized over the next several years. As that happens, costs fall, efficiency improves, and companies generate unique data sets that can support payments, insurance, and lending.

 

The advantage is not only having proprietary data. These software platforms also know the moment when the customer is most likely to need a financial product, which makes highly targeted offers possible.

 

Nuvocargo is one example from our portfolio. It helps move trucks, drivers, and goods across the U.S.-Mexico border. Because it sits inside that workflow and sees the relevant business data, it can offer adjacent financial services such as insurance, payments, and revenue-based lending.

 

That creates a much more seamless experience for the customer.

 

I think we're going to see payments, insurance, and lending increasingly wrapped into vertical SaaS platforms across many industries.

 

When I talk to bank executives, I tell them this is an opportunity they could participate in. They have a low cost of funds, and they often already have relationships with the small businesses through checking and other products. The challenge is figuring out how to insert themselves into these software ecosystems.

 

We're seeing vertical SaaS companies scale within one industry and then expand into adjacent industries. AvidXchange, which is now public, is another example. It has effectively become part of the system of record around money moving into and out of businesses.

 

I think this is a fascinating space with a great deal of growth ahead. It reduces friction and helps businesses operate more seamlessly.

 

Are you bullish on embedded finance and banking as a service more broadly, including on the consumer fintech side?

 

Yes, with some nuance.

 

Before COVID, a lot of technology leaders at large incumbent banks were saying fintech was largely a flash in the pan. They didn't believe many fintech companies were industrial-strength. Compliance teams would say, "We can't work with them because regulators will have a problem with it."

 

Several years later, it's clear that many fintechs have become meaningful parts of the financial-services value chain.

 

The best banks now recognize that they have enormous technology debt and can't build everything themselves. They need to become good at curating outside companies and combining those services into a better experience for their customers.

 

The objective is to knit together the right products with the lowest possible cost and the strongest possible security.

 

Companies such as Alloy, Treasury Prime, and Atomic, all of which are in our portfolio, are examples of the kinds of providers banks can integrate.

 

The answer isn't "never work with fintechs," and it isn't "outsource everything to fintechs." It's about mixing and matching capabilities based on what a particular bank is good at and where it needs help.

 

Buy now, pay later is one of the clearest embedded-finance examples. Klarna, which was an early QED investment, took an old model, point-of-sale installment lending, and turned it into a much better digital experience.

 

It reduces friction. The merchant often sees higher ticket sizes and improved checkout conversion, while the consumer gets a simple experience. Based on the published numbers, credit risk also appears manageable.

 

The important next step is for buy now, pay later providers to report payment behavior to the credit bureaus so consumers who pay responsibly can use that history to move up the credit ladder.

 

We've only got a couple of minutes left, Nigel, but I want to ask about the next generation of fintechs. What are you seeing founders work on now?

 

There's a lot happening. Last year felt like a nine-month pause after the excesses of the fintech boom and bust. A lot of founders went back to basics and worked quietly instead of rushing into the market.

 

Now we're seeing a vibrancy that wasn't there six months ago, and maybe wasn't even there three months ago. That's exciting.

 

Much of the innovation isn't revolutionary. It's focused on problems we've already discussed: financial inclusion, lending to small businesses and consumers, and improving access to money.

 

Earned-wage access is one example. Companies such as DailyPay, Rain, Payactiv, and Wagestream are growing, and we're seeing similar models internationally. Giving consumers access to wages as they earn them is becoming a meaningful category after a period when people questioned whether it would work.

 

We're also seeing continued growth in embedded finance and in efforts to democratize investing. It's hard to know which next-generation players will be able to compete with companies such as eToro or Robinhood, but there are a lot of interesting ideas in the market.

 

Many ideas still emerge first on the West Coast and then move around the world quickly. Credit Karma is a good example. Its model later influenced companies such as ClearScore in the U.K. I expect similar models to continue appearing in new geographies as access to credit-bureau information becomes more important for things like renting a car, getting a job, or applying for a mortgage.

 

We're also seeing more use of alternative data, including rental payments, utility payments, and deposit-account activity. Those data sets can be genuinely predictive, although, as we discussed earlier, you have to be careful about throwing all of them indiscriminately into a massive machine-learning model.

 

What about blockchain? Have you seen anything potentially scalable there that goes beyond creating another cryptocurrency?

 

We've been cautious about the space. We invested in Bitso in Mexico, which has done very well, and in Shakepay in Canada.

 

Our approach has generally been to focus on the banking services around crypto. If someone wants to lend against it, move it, or store it, those are familiar financial-services activities. You can think about them similarly to services built around other assets such as gold, silver, or foreign currency.

 

What we haven't yet seen is crypto becoming a mainstream means of transaction. We're watching it closely, but the breakthrough has not happened yet.

 

What about the metaverse? Anything interesting there, or has that passed?

 

I think it passed. Who knows, but I think it was mostly a flash in the pan.

 

I am interested in how payments and embedded finance fit into gaming. We're having conversations in that area, and fraud management is obviously important there as well.

 

In fact, I think applying AI to fraud may be more interesting than applying it to credit risk for the reasons we talked about earlier.

 

As for blockchain, we're still studying it. We have a number of irons in the fire, but we haven't seen the industrial-strength breakthrough that people have been predicting.

 

One thing I've learned is that we often expect new technologies to happen faster than they actually do. Then, when adoption finally arrives, it happens exponentially. It's slow, slow, slow, and then suddenly quick, quick, quick.

 

We may get there with blockchain. We just aren't there yet.

 

There are some interesting companies doing things with credit unions here in the U.S. Given how collaborative credit unions can be, that may help generate real-life use cases.

 

Absolutely. Credit unions are a fascinating distribution channel.

 

One of our portfolio companies, Amount, recently announced work with a credit-union consortium. It is bringing advanced analytics, lending-management capabilities, and deposit products into that market.

 

As I mentioned earlier, much of mainstream banking in the U.S. has not embraced digital at the same pace consumers have. Companies such as Amount can help accelerate that change.

 

Credit unions also have an advantageous cost of funds. Another QED portfolio company, Caribou, refinances auto loans. Being able to place those loans with credit unions that have a low cost of funds makes a lot of economic sense.

 

Nigel, thank you so much for spending the time today. And for everybody listening, this has been another episode of What's Going On in Banking. Hope to see you on another episode.

 

Pleasure, Ron. Thanks a lot. Take care.