Henrique Dubugras - Building the Financial Center of Gravity – [Founder’s Field Guide, EP. 35]
What this covers
Patrick talks with Henrique Dubugras, co-founder and CEO of Brex, about how a startup credit card became a business checking account. The early cards were prepaid because Brex lacked a credit BIN, so merchants such as Instacart declined them, and that continued until a bank partnership arrived after more than a year. About three months after the late-2018 launch, he moved roughly 75% of the company's resources to the account, over board objections. He explains why banks can't make this move themselves: legacy core processing from vendors like FIS and Fiserv makes any change cost around $50 million and a year. He also describes Dynamic Underwriting Versus Static Limits, where Brex re-evaluates every business daily.
The middle of the conversation covers go-to-market and economics. Early marketing used a narrow message and target: "credit card for startups," aimed at about 70,000 US companies that paid ads couldn't reach. On interchange, he says more than half of the roughly 2% goes back into rewards, which is why European caps eliminated points. He doubts the US will follow, given its attachment to points. He compares Brazil and the US, arguing that a small raise forced profitability while the US round of $57 million led to less efficient decisions. He also discusses engineer costs, remote hiring, and paying around the 90th percentile.
It closes on management and temperament. He proposes judging sales and marketing by CAC relative to LTV, G&A by leverage over time, and R&D by whether its products succeed. He wants to spend about 30 years on one problem, citing Marc Benioff on decades versus years. He also says persuasion works best when you show people they are right.
Dubugras argues Brex wins by owning the full financial stack, rebuilt from scratch and centered on a business checking account, rather than stitching together third-party vendors, and that long time horizons justify costly infrastructure and brand investment.
- Legacy bank core systems make change slow and expensive, so rebuilding infrastructure in-house enabled innovation
- Owning the account gives real-time data, instant payouts, and cross-sell ability that aggregators and vendors cannot match
- Compounding requires working on one problem for decades and making investments that pay off in 10 years
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Of roughly 2% interchange banks earn on cards, more than half often goes back into the points system, so capping interchange at 1% would remove the margin for rewards, as happened in Europe where regulation eliminated points programs.
“more than half of that goes back into the point system. If they killed it and say, 'Hey, interchange now just 1%,' they don't have enough margin to do that anymore so it would die.”
Banks do not innovate because of a lack of ideas; they are bound by legacy core processing technology from vendors like FIS and Fiserv built over 30-plus years, where any change costs around $50 million and a year.
“It's because they're bound by this legacy technology, like FIS or Fiserv that has been built over 30 plus years.”
US engineers cost five to ten times more than engineers in other countries, so US companies must become much bigger to achieve profitability, while remote hiring can improve tech margins by reducing that cost.
“Costs of an engineer in the US, it's not like two or three times, sometimes it's five, six, seven, 10 times more than someone outside of the US.”
Early Brex cards were prepaid because they lacked a credit BIN, so cards were declined at many merchants such as Instacart, which angered customers until the company secured a bank partnership, a process that took over a year.
“we couldn't get a credit BIN like the credit card at that point, and it just didn't work at random places.”
Sales and marketing should be judged by CAC relative to LTV rather than absolute spend, G&A should show leverage as a share of revenue over time, and R&D should be judged by whether the products it produces succeed.
“Sales and marketing you're CAC bound. Absolute numbers don't matter. The matter is like, 'Hey, what is the CAC/LTV'”
Banks long assumed their deposit base gave them an inherent advantage because finance was seen as lending, but Stripe, Square, and others showed that payments, accounts, and cards are separate, transactional, less balance-sheet-heavy profit pools, which opened fintech to disruption.
“people started realizing that Fintech is about more than lending.”
Brex's underwriting re-evaluates every business daily and adjusts limits up or down, whereas banks set a fixed limit that rarely changes, which fails startups whose balances can go from millions to zero quickly.
“A dynamic model means that we re-underwrite every business every day.”
Financial services have historically operated in batches because of legacy banking, and Brex is converting these static processes into real-time streaming, a shift comparable to Vanta's continuous compliance monitoring versus annual SOC 2 audits.
“In financial services, everything is in batches now because of banks and we're turning it all into real time.”
Brex's early marketing used the very clear message 'credit card for startups' targeting a market of about 70,000 US startups, because paid Google and Facebook ads could not target startups, so the team used a more creative approach.
“None of the paid ads worked for us. We couldn't target on Google and Facebook startups.”
Brex sold on the simple value proposition of higher limits and rewards rather than the software features, because customers who sign up for rewards end up loving the software anyway.
“let's just make the sale simple, the value proposition super clear and more people will sign up.”
Brex lets employees split a total compensation offer between cash and equity, and it pays at around the 90th percentile of the market, because the company would rather overpay than face unfairness from a top candidate earning more than the team.
“I'd rather, literally just pay people in the 90th percentile of the market or more to get the best people.”
Big banks categorized startups as small businesses, but startups are economically much closer to midsize and mid-market businesses, so no one was giving them white-glove service, which created the opening for Brex.
“big banks always categorized startups as small businesses. Startups and small businesses are the same thing. The reality is they're not the same thing.”
Henrique and his co-founders want to work on one problem for around 30 years rather than be serial entrepreneurs, because they believe in the power of compounding and that sustained focus on one large dream creates outsized impact.
“We just want to work on something for 30 years.”
Marc Benioff's principle that people overestimate what can be done in a year but underestimate what can be done in a decade explains why sustained effort on one vision, as at Salesforce, builds outsized companies that short-term CEOs would abandon.
“people overestimate what we can do in a year, but underestimate what you can do in a decade.”
The most effective way to persuade investors and customers is not to convince them you are right but to show them they are right, by understanding their existing beliefs and framing your proposal within their worldview.
“you don't really need to convince people that you're right. You just need to convince them that they're right”
Companies with large R&D budgets that launch failed products should not invest more, but companies whose R&D produces successful products should keep investing heavily, and Henrique says Brex should be honest about whether its launches gain traction.
“Am I producing products that are successful? Because there are companies that huge R&Ds, but everything they launch fails.”
Because Brex holds all payments, expenses, and bills, it can offer real-time business reporting, while competitors need 10 business days after month-end to reconcile statements from many separate providers, and stitched-together data cannot match the same vendor across channels.
“For Brex, it takes 10 business days after the closing of the month.”
The US startup ecosystem has many executives who have already gone through hypergrowth, such as a head of engineering who had been early at Stripe, which let Brex scale from 100 to 400 people in a year by copying lessons; Brazil lacks this pool because fewer startups have completed growth cycles.
“Every function in the US there's someone who did it and went through that growth phase and that helps you accelerate a ton.”
Brex plans to invest heavily in its brand despite no near-term return, because a strong brand in ten years could rival American Express, Visa, Mastercard, or Chase, and Airbnb's 93% organic traffic came after 11 years of hundreds of millions in brand spend.
“after like 11 years of investing hundreds of millions in brands, I sure hope so.”
Brex's core premise is rebuilding all financial infrastructure in-house, including KYC, AML risk models, core ledger, authorization, fraud detection, and rewards, rather than relying on legacy vendors, and this rebuild is what allowed them to innovate.
“a core premise of Brex is we're going to rebuild all the infrastructure from scratch. We're not going to rely on any legacy technology from anyone else.”
Brex justifies heavy customer acquisition spend in some segments by holding its own conviction about each segment's lifetime value, as with early-stage startups that later become large paying customers such as Scale AI.
“It was our own conviction that, that was going to be true and we were going to do it.”
US consumers have a deep cultural attachment to credit card points and rewards, which makes Henrique think interchange regulation is unlikely in the US because a vocal group would oppose any measure that kills points.
“It's like you would have a pretty vocal group of people that will not like that decision because it would kill the points.”
About three months after the card launched in late 2018, Brex moved roughly 75% of its resources to build a business checking account, against board objections, because it believed the core account was the customer's center of gravity.
“We made a decision to basically get 75% of our resources and invest it in a newer product. It was not obvious at all.”
The checking account is where the customer's financial center of gravity lies, so once Brex holds it, it can cross-sell any other product, whereas without it cross-selling is much harder.
“If you had that, you could cross sell anything. If you didn't have that, it was much harder to cross sell stuff.”
Third-party data aggregators such as Plaid are good companies but fail often because legacy bank connections are unreliable, so customers get a bad experience unless they use Brex's own account.
“We didn't rely on third-party data aggregators like Plaid... it fails a lot. Not because of them, because of banks have these legacy technology.”
Running their first Brazilian company on a small raise forced the team to prioritize reaching profitability, while the US raise of $57 million led to different decisions and less efficiency, which Henrique attributes to abundant capital.
“Getting to profitability was the main thing a company had to do.”
Brex adapts processes to attract talent rather than standardizing them, exemplified by a flexible compensation model in which new hires choose their own mix of cash and equity from a total offer, to avoid rigid pay bands that lose top candidates.
“I think the most important thing for us is have a talent first culture, which is adapt your processes.”
Writing decisions as memos instead of debating verbally produces better decisions, because people cannot bluff in writing and must think carefully before answering, unlike in meetings where they improvise and invent answers.
“It's really hard to bullshit and write.”
Hired CEOs tend not to make long-term investments because they answer to short-term pressures, whereas Brex makes decisions today expecting value in ten years and only pursues them if it is thinking on that horizon.
“which is a problem of hired CEOs is, 'Hey, we're making decisions at Brex today that we think are going to be valuable in 10 years.'”
Brex's remote-first model lets it hire people in unexpected places rather than only in major cities, since Brazilian hires came from many smaller locations and only one was in São Paulo, a benefit Henrique considers worth almost any cost.
“Honestly, there's only one person in Sao Paulo. All the other people we hired, they're like in these random places in Brazil.”
Published B2B card penetration statistics are misleading because they measure dollar volume rather than number of transactions, and large B2B payments in the tens or hundreds of millions are unlikely to move to cards.
“it's all volume. It's not of number of purchases.”
Fintech is capital-intensive, and Brex raised $57 million before launch because banks required demonstrated balance sheet and multiyear runway before partnering, so larger fundraising made partnerships possible.
“They want you to have many years of runway. ... We raised $57 million prelaunch.”
Many Silicon Valley founders dream of software that sells itself, but Brex's business depends on unglamorous work such as financing, securitization, credit, collections, and infrastructure operations.
“a lot of founders in Silicon Valley they're engineers and their dream is just to build software.”
Credit cards hold only low single-digit share of B2B payments, versus around half of consumer payments, because suppliers have long taken checks and ACH, and fees of a couple percentage points are hard to absorb on multimillion-dollar payments.
“It's just legacy. Suppliers always took payments.”
Brex built its minimum viable infrastructure to launch in 2018 by restricting itself to startups, which let it use startup-specific onboarding questions like 'who is your investor' that would make no sense for non-tech businesses.
“to build faster we built it in a way to only work for startups.”
B2B card penetration is rising as high-margin tech vendors such as Slack, Facebook, Google, and AWS accept cards, and COVID accelerated the shift as Brex's GMV from those companies grew sharply after the pandemic.
“As higher margin vendors become a bigger percentage of the total supplier payments, I think that penetration is going to increase.”
Brex's instant payouts give businesses same-moment access to marketplace sales from Amazon, Stripe, Shopify, and PayPal, which is only possible because Brex owns the account and can move money instantly rather than waiting on ACH or wire.
“The only reason that's possible is because we own the account, so we can actually make money available instantly.”
Amazon's warehouses, trucks, and logistics are an unglamorous investment that takes years to pay off but enables one-day delivery, illustrating why engineers who only want to build software miss the long-term value of infrastructure.
“Building warehouses and trucks and all these things are something that's going to take a long time to pay out but it's what allows me to get one day delivery.”
Going remote requires changing nearly everything about how a company operates, and because there is no playbook for remote work, the company must invent processes, which is a larger con the bigger the company is.
“In remote, there's no playbook. We're having to create it from scratch.”
Because interchange revenue subsidizes the product, Brex offers free wires and ACH and charges $29 per month for its premium software, which is far cheaper than competitors such as Expensify and Bill.com at around $30 per user.
“All interchange is subsidized. Because we have interchange, we can give all these seeds for free.”
Sales mistakes come from pitching first, and the correct approach, learned from the book The Sales Bible, is to ask questions and understand the other party's goals and worldview before saying anything about your product.
“the main mistake people do is they start just pitching everything. The right thing to do is you should ask questions first.”
Incorporating a company, getting an office, and setting up internet took about a month and a half in Brazil but only two days in the US, though US bank accounts and credit cards were slow, and founder bandwidth shifted from non-product work to product work.
“In Brazil, it took us a month and a half to get an office and internet set up and incorporate a company.”
Brex spent about $300,000 on a large share of San Francisco billboards, based on the logic that most startups are in or visit San Francisco, and within three months nearly everyone knew the brand.
“We got all the billboards in the city for $300,000. Three months after everyone knew what Brex was.”
Startups start with large sums of money but deplete them quickly, so their needs differ from small businesses, and banks' old systems could not support their rapidly changing credit needs.
“startups have... They start with a million dollars, but they go to zero pretty quickly.”
Patrick notes that high-performing companies often have a pair at the top, one person who opens doors and one who structures things, a pattern he links to Will Thorndike's The Outsiders.
“you had this partnership pair at the top of the chaos person and the order person.”
Brex consistently chooses to own infrastructure rather than use vendors or shortcuts, accepting higher cost and longer build time because the investment pays off eventually.
“We're going to own this. It's going to be awesome. It's going to take longer. It's going to be more expensive right now, but eventually it's going to pay off.”
Patrick notes that Scott Belsky's window dressing idea holds that the offer that gets a customer through the door is often not what they ultimately buy or return for, so simple and attractive entry offers are useful.
“The thing that gets you in the door is typically not even the thing you buy or ultimately care about or go back for”
Brex's first round of $7 million targeted the best fintech operators, bringing in Micky of Ribbit Capital, Max Levchin and Peter Thiel, who helped the company early before it had launched.
“we optimized for getting like, who were the best people in Fintech and how do we get them on our cap table?”
Card grace periods of 30 to 45 days after statement, roughly 60 days total, give customers working-capital funding that would cost about 10% over 60 days at Brazilian rates.
“60 days to pay something is going to cost you 10%.”
Brex started with corporate cards partly because the founders' payments-processing experience made cards within their zone of knowledge, more than term loans or business checking.
“cards were within our zone of knowledge.”
Henrique says the founders must hold their own conviction about where free cash flow will come from, so cash flow measurement is the right metric but must be backed by the company's own view of future customer value.
“We totally agree that free cash flow is the right measure for us.”
Tech companies pay most B2B bills by card because their suppliers take card and have high margins, whereas traditional businesses like restaurants still pay wholesalers by check because those suppliers have always used checks.
“If you look at a tech company, actually, a lot of their B2B payments are on card.”
Asset managers charge percentage-of-assets fees that are scraped from accounts rather than written as checks, and this automatic, hidden structure creates a psychological barrier to change, similar to how interchange is hidden from consumers.
“it's not a check, it's a scrape from the account.”
Brex pays around the 90th percentile or higher to avoid the unfairness of recruiting a top candidate who earns more than existing team members, and Henrique argues this high R&D spend is what sustains four products.
“I'd rather, literally just pay people in the 90th percentile of the market or more to get the best people.”
After selling Pagar.me and moving to the US, the founders tried to pivot into other fintech areas but kept returning to B2B financial services, so they chose to build Brex around that long-standing focus rather than abandon it.
“our head kept going back to B2B financial services.”
Henrique and co-founder Pedro complement each other, with Pedro building systems and process while Henrique focuses on new products, customers, and rounds, a structure that resembles the one-two partnership of high-performing companies.
“Pedro and I are good friends, we're very complementary.”
The kindest thing Henrique recalls is a benefactor, George, who paid about $60,000 per year for the founders' college in the US, and whose foundation they later repaid with Brex stock so it could fund more scholarships.
“he paid for our college when we came to the US, so a lot of money, $60,000 a year.”
Brex's founding began with a personal frustration: when the founders could not get a credit card in the US despite having money, they realized wealthy people could not get a card with a 10,000 dollar limit for AWS, which sparked the first market.
“Why can't you get a card for $10,000 limit to pay for AWS and all these other things?”
Rather than assuming a product will attract customers, Brex took an opposite approach from the start, pursuing every possible customer with aggressive outbound sales and continuing to do so today.
“I think we were the opposite. We were like, 'Let's go after every single customer that we can.'”
Brex's sales organization was built on continuous iteration of messaging and tactics, with the belief that getting customers on the phone raises conversion, and that the economics should be made to work for that.
“If we can get them on the phone, there will be a higher chance that you can convert”
Patrick argues that the best ideas come from making things rather than thinking about them, and he sees Brex as an example in which a founder stumbled on a frustration that became a business.
“all the best ideas come when you're making things, not thinking about things.”
Brex first raised a $7 million round, reached about 100 beta customers within a year, when many credit-card startups had never launched, and then received a preemptive $50 million round from an investor named Anu.
“a lot of people tried to do what we're doing, it never worked. Then we had 100 customers.”
Brex's darkest period was repeatedly missing launch targets for the central account, first six months then a year, which made Henrique feel he was misleading investors even though it was not intentional.
“We promised all these investors that it was going to be the launched and it wasn't. It sucked because I felt like I was misleading people.”
Henrique says remote work let him move from San Francisco to Los Angeles, work more productively without a commute, and walk his dog during the day, and he expects remote work to improve happiness and productivity across society.
“I'm just having like the most amazing life post remote world”
Brex's first real customer was Alex Wang, founder of Scale, a 19-year-old without a FICO score who could not get an Amex, and his adoption convinced the founders that the product solved a real need.
“That was our first real customer, and first time it was like, 'Okay, cool. I guess people need this.'”
Securing a bank partnership as a young company takes a long time, and combined with building the software, Brex's fix for prepaid-card limitations took over a year.
“getting a partnership with a bank when you're still young, it takes a while”
The billboard strategy was the idea of Brex's head of sales, Sam, not Henrique, who credits Sam with the idea and its execution.
“our Head of Sales, Sam, had the idea and executed on it”