A2Z Fintech
00:12:51 · Zubin Vandrevala, Aman Narain
[00:12:51] S1: Right. And then a new back pays for its ladder. Right? Cost of risk, 520 basis points against Revolut's one fifty eight. Late stage bad bad loans, 6.1% against against Revolut's 2.4. [00:13:04] S0: Yeah. Same flywheel, opposite ladders. Velez at Nubank turns deposits into credit and prices the default. Storonsky turns deposits into a float and prices transactions. The Lehman floor decided which ladder he built. [00:13:21] S1: Right. And let's talk about a third ladder people forget about. SoFi started by refinancing student loans, lends heavily, bought a bank charter in 'twenty two, and is also bought Galileo and and Technises, the plumbing, and sells that plumbing to other banks and fintechs. SoFi's Top Run isn't a product for customers. It's essentially the stack itself sold to its competitors. [00:13:46] S0: Which is heresy in a bank. Every bank I've known guards its core systems like crown jewels. [00:13:53] S1: Right. And SoFi rents the crown jewels out by the account. [00:13:56] S0: Which is the real story of the neobank class of this decade. The ones that made it aren't cheaper banks. Each has picked a different answer to a question banks stopped asking about a century ago. What is a deposit for? [00:14:10] S1: Right. And and neobank says credit, Revolut says float, SoFi says soft.
00:14:10 · Zubin Vandrevala, Aman Narain
[00:14:10] S1: Right. And and neobank says credit, Revolut says float, SoFi says soft. [00:14:16] S0: And JPMorgan says all three on a $4,000,000,000,000 balance sheet, which is sort of why the incumbents aren't dead, I suppose. And and we said that episode that incumbents can run the flywheel to Chase UK, Trust Bank here in Singapore. What they lack isn't the idea. It's the nerve to cannibalize themselves. Storonsky had nothing to cannibalize. [00:14:40] S1: Yeah. No scar tissues from a branch network. Right? His only scar is probably Lehman. Well, let's watch Revolut run the wedge somewhere new, Aman. Right? Because it shows how far he goes to avoid the lending ladder. Take Mexico, January 2026. The launch offer is 15% a year on savings, which, you know, when you think about what the Central Bank has got maybe 12 times to about seven. [00:15:04] S0: Eight points underwater, Zubin, on every peso he doesn't lend, and he isn't lending. [00:15:11] S1: Yeah. Yeah. And it isn't an interest rate. Right? It's a marketing budget. And I think we've kind of mentioned this before. Right? Internal estimates report puts his acquisition costs in Mexico, I think, under $12 an account against more than maybe about 65 for a branch bank. Right? So 550,000 accounts by August on a, what, $167,000,000 of capital, and behind it, the $65,000,000,000 remittance corridor from The US. [00:15:39] S0: And and same move as the 2015 card. Give away the spread to get the customer. In eleven years, he has never once acquired a customer by lending Yeah.
00:15:39 · Aman Narain, Zubin Vandrevala
[00:15:39] S0: And and same move as the 2015 card. Give away the spread to get the customer. In eleven years, he has never once acquired a customer by lending Yeah. [00:15:49] S1: Which is a flywheel adapting to the market. Right? In London, the wedge was a card. In Mexico, it was a deposit rates. The first step changes. The refusal to climb onto the credit rung does it. [00:16:00] S0: And the float has a weather problem. Analysts rule of thumb, every 100 basis points of rate cuts costs a digital bank with a 5% loan book, 15 to 22% of its float revenue. Nubank is naturally hedged. Cheaper money means more credit demand and fewer defaults. Revolut is the one who dressed for one kind of weather. [00:16:24] S1: Yeah. Interest is about a quarter of his revenue. Right? 74% is fees. [00:16:30] S0: And look at the purest float and fee business that's listed, Wise. As rates eased this year, its growth slid from over 30% to 19. Same species, same coal. [00:16:43] S1: Yeah. Then then the honest summary. Right? You removed credit risk and accepted rate risk. A choice, not a free lunch. [00:16:49] S0: He he... He'd say it's cheaper risk. [00:16:52] S1: And and he might be right for a decade at a time. Right? Rates don't default. [00:16:56] S0: And every risk we've named so far, he's either removed or priced. Not one of them is inside the building. So let's go inside.
00:16:56 · Aman Narain, Zubin Vandrevala
[00:16:56] S0: And every risk we've named so far, he's either removed or priced. Not one of them is inside the building. So let's go inside. [00:17:05] S1: Yeah. Let's go inside. Right? 15,000 staff from 110 countries on his own count. And at the top, more than 40 direct reports. Yeah. Confirmed. Right? Including the three people whose job is it to tell him, essentially, finance chief, the general counsel, and the credit risk officer. [00:17:22] S0: Picture that diary. Twenty minutes each of a week is thirteen hours of one to ones before he even opens dashboard. [00:17:30] S1: His version has no layers, nobody between him and the numbers. The version from former executives, it's... Isn't a structure. [00:17:37] S0: You know, we've been trained classically in business school about span of control and things. But before we judge it, put it next to the most valuable chipmaker in the world. Jensen Huang runs NVIDIA with around 60 reported direct reports. [00:17:51] S1: I I knew you would go there. [00:17:53] S0: And no one on ones at all. I mean, I would love to work for Jensen for multiple He says he can't. So so he puts a problem in front of all of them, and and then they will attack it together. [00:18:05] S1: So funny isn't crazy. It's two third of Jensen. [00:18:08] S0: Which is the real question. Is this the new way of running a company, Zubin, where dashboards and AI replace the middle layer, if you'd like, and the chief executive talks to everybody all at once? Or is it that span of control that works right up until it doesn't?
00:18:08 · Aman Narain, Zubin Vandrevala
[00:18:08] S0: Which is the real question. Is this the new way of running a company, Zubin, where dashboards and AI replace the middle layer, if you'd like, and the chief executive talks to everybody all at once? Or is it that span of control that works right up until it doesn't? [00:18:25] S1: Yeah. And Jensen's version, just to kind of get us on the same page for those who follow along, because everyone hears the same thing at the same time. Storonsky runs through one chair on a Monday, same span, different wiring. [00:18:38] S0: The stakes are very different. You know, in a chip company, a red cell is a delay. In a bank, a red cell could be a regulatory incident, a detailed Monday weekly business review, and, you know, endless, you know, meetings and debate. [00:18:53] S1: It's phystech chalkboard run weekly on the whole company. [00:18:57] S0: And, you know, I'll defend Storonsky because every bank I've worked in have had those meetings I just talked about. The difference is that we found the red cell six weeks after the quarter end in a deck that took a fortnight to build, and nobody remembered who owned it. You know that. You worked at a bank. [00:19:15] S1: I remember those days. Yes. Yes. [00:19:17] S0: He's compressed all of that quarter into a week and and the deck into a screen. That's not cruelty. That's that's low latency. [00:19:26] S1: Latency for the number. Right? What's the latency for the person who owns that number? [00:19:31] S0: Which is, I guess, where the karma comes in.
00:19:31 · Aman Narain, Zubin Vandrevala
[00:19:31] S0: Which is, I guess, where the karma comes in. [00:19:33] S1: Yeah. So let's talk about karma. Built in 2020 in the annual report, points across more than 30 compliance workflows offer a late policy, a delayed suspicious activity report, a missing customer document. The scores roll up to the departments and become a multiplier on the team's bonus. So, compliance isn't a memo, it's priced into your colleagues' pay. [00:19:55] S0: You know, I saw this internal data. There was a report, didn't see the internal data. Saw the report on it that said that the adherence rose 25%. [00:20:04] S1: That's pretty wild. Right? Like, it measures whether report was filed on time, not whether it was right. A bank that has to gamify caution has told you its default setting. [00:20:12] S0: Or it's told you something about every bank, and this is the one indelicate enough to staple it to the bonus. I'll add what I hear firsthand because, as I said, I have friends who work there, exhilarating brilliant people with an absolutely ruthless mindset. For your first few years, you're literally only as good as your last month. And after that, it's your last quarter. Man, that sounds exhausting. Yeah. Very different from Google that we grew up at. But the cadence at which they deliver is an order of magnitude faster than banking, and that's the real competitive advantage as long as they don't screw it up. [00:20:56] S1: Let's park that caveat because we'll need it for that... The breach.