The Garage Beneath the Leather Factory
In the third quarter of 2023, a company most Americans have never heard of sold $11 billion in goods — a nearly 60% increase year over year — through a marketplace that stretches from the favelas of São Paulo to the altiplano of Bolivia, from the colonias of Mexico City to the estancias of Uruguay. By the time that number was reported, Mercado Libre had already surpassed Petrobras, the Brazilian state energy colossus, as Latin America's most valuable publicly traded company. Its market capitalization, which floated near $100 billion and would eventually breach it, represented something without clean analogy in global technology: a single platform that had absorbed the functions of Amazon, PayPal, FedEx, a consumer bank, and an advertising exchange, built not atop the reliable infrastructure of the American Midwest or the factory floors of Shenzhen but across a continent where half the population lacked a bank account, cash remained king, roads disintegrated in the rainy season, and inflation could annihilate a business model between breakfast and lunch.
The paradox at the center of Mercado Libre is that every disadvantage of its operating environment became, over time, a structural advantage. Where Amazon entered a market with working logistics, reliable payments, predictable monetary policy, and consumers already comfortable with online commerce, Mercado Libre had none of those luxuries — and was therefore forced to build each one itself. The company that emerged is not a clone. It is something stranger and, in certain respects, more durable: a vertically integrated operating system for an entire continent's economic life, one whose competitive moats were poured from the very chaos its competitors found repellent.
That this machine was conceived in a parking garage beneath a leather factory in the Saavedra neighborhood of Buenos Aires — not as a deliberate homage to Hewlett-Packard or Apple, but because the garage offered the only fast broadband connection available — is the kind of origin detail that resists metaphor by being too literal. The building had terrible offices. But it had good connectivity. Everything that followed flowed from a similar logic: find the constraint, build around it, and let the constraint become the moat.
By the Numbers
The Mercado Libre Machine
$20.8BNet revenue, FY2024
$5.9BQ1 2025 net revenue (+37% YoY)
130M+Monthly active users
18Countries of operation
120,000+Employees
$100B+Market capitalization (late 2025)
18,000Software engineers
30,000Code deployments per day
An Argentine in Palo Alto
Marcos Galperin was not a scrappy outsider. He was the fourth of five children born into a family that owned Sadesa, one of the world's largest leather manufacturers — Argentine business royalty, the kind of pedigree that grants access to the right schools, the right networks, the right runway. He attended Saint Andrew's Scots School in the Buenos Aires suburb of Olivos, played on Argentina's junior national rugby team, then crossed the hemisphere to study economics and finance at the Wharton School of the University of Pennsylvania. He was, by every measure, a member of the elite. What separated him from the many well-connected Latin American heirs who pass through Ivy League institutions and return home to manage family empires was a particular quality of frustration.
After Wharton, Galperin returned to Argentina and took a job at YPF, the state oil company — not because it paid well (it didn't; he'd turned down offers from JPMorgan and Goldman Sachs) but because he wanted to understand how emerging markets actually worked. The experience was formative in ways a Wall Street desk never could have been: the bureaucratic tangles, the currency instability, the vast distance between policy and execution. He spent two years there before enrolling at Stanford Graduate School of Business in 1997.
Stanford in the late 1990s was ground zero for the internet's first delirium. Galperin watched eBay convert American consumers into enthusiastic online traders. He sold his own Volkswagen Golf through an early classified ad. And he felt something crystallize: not just the standard MBA conviction that the internet would change everything, but a more specific and geographically inflected frustration. "I was very enthusiastic about what I thought the internet was going to do to the world," he later recalled. "And I was very frustrated because I had to come back to Latin America, and there was nothing."
The nothing was the opportunity. Latin American retail was brutally concentrated in major urban centers. Rural populations — hundreds of millions of people — were effectively cut off from modern commerce. "People would rather live in a slum next to a big city than live in better conditions in the countryside," Galperin observed, diagnosing a spatial inequality so deeply embedded it functioned almost like physics. An eBay-type platform wouldn't just be convenient in Latin America. It would be structurally transformative.
At Stanford, Galperin met Hernán Kazah and Stelleo Tolda, who would become his co-founders. The three developed their business plan while studying an eBay case — then seized an opportunity that has since become legendary in Latin American startup lore. John Muse, co-founder of HM Capital Partners, was delivering a guest lecture at Stanford. Galperin volunteered to drive Muse to the airport. Somewhere between the parking lot and the terminal, the three students pitched their idea. Muse invested $7.5 million. Mercado Libre was born.
Nuclear Winter and the Education of Survival
The timing was exquisite in its cruelty. Mercado Libre launched on August 2, 1999, into a Latin America where internet penetration hovered around 3%. Galperin and his team erected makeshift cubicles in the garage beneath his father's leather company at Tronador 4890 in Saavedra. (The building eventually kicked them out for violating safety codes.) They had $7.5 million, a handful of engineers, and a continent to wire.
Then the world ended. The dot-com bubble burst in 2000. Capital markets froze. Argentina, already economically fragile, spiraled into its catastrophic 2001–2002 crisis — sovereign default, bank runs, the peso's collapse, riots in the streets. For a technology startup dependent on both internet adoption and consumer confidence, the environment was close to extinction-level.
Mercado Libre's response became foundational mythology. Nicolás Szekasy, the company's early CFO, implemented what he called "nuclear winter" planning: the assumption that capital markets would remain frozen indefinitely and that the company needed to survive on what it had. They cut teams in half. They centralized operations from nine countries into Buenos Aires. They shifted the entire strategic orientation from growth-at-all-costs to a path-to-profitability mindset. And they preserved roughly $50 million in capital — a war chest that would prove decisive.
It has happened to me very badly. There were times, especially in the beginning, when I remember that I did not want to get out of bed. The alarm clock rang in the morning and said: No, I do not want to get out of bed because I have to face all my employees and the company is going to melt.
But the company did not melt. By 2005, Mercado Libre had achieved profitability with substantial reserves intact. The nuclear winter had created something invaluable: institutional memory around capital efficiency, a DNA-level conviction that adversity was not an aberration but the baseline condition of operating in Latin America. Every subsequent strategic decision — every investment in logistics, every extension into fintech, every bet on a new country — would carry this scar tissue. The company built assuming the next crisis was always coming, because in Latin America, it always was.
The eBay relationship was critical during this period. In 2001, eBay acquired a 19.5% stake in Mercado Libre, simultaneously agreeing not to compete in the region for at least five years. The deal provided both capital and credibility during the darkest months. It also provided an invaluable case study in what not to do. Hernán Kazah later described a pivotal "Lake Como moment" — a meeting with eBay's leadership — where the Mercado Libre team recognized that eBay's playbook, optimized for developed markets with functioning infrastructure, was fundamentally mismatched to Latin American reality. The lesson was clear: adaptation, not replication.
The Inconscience Advantage
The creation of Mercado Pago — the payment platform that would eventually become as significant as the marketplace itself — began not as a grand strategic vision but as a desperate hack to solve a specific, crippling problem: nobody in Latin America could pay for things online.
Credit card penetration was minimal. Banking infrastructure was rudimentary.
Trust between strangers transacting digitally was essentially nonexistent. The marketplace was growing, but the disintermediation problem was savage — buyers and sellers found every creative method imaginable to bypass the platform, connect offline, and avoid fees. Mercado Libre was a matchmaker whose couples kept eloping.
Mercado Pago, launched in 2003 (some accounts date the initial concept to 2004), was born from this crisis. Paula Arregui, who would lead the fintech arm, later offered a candid reflection on its origins: "If we had fully understood the complexity of what we were attempting, we might never have started." The early reality was farcical by fintech standards. The team manually processed transactions. They entered card data into spreadsheets. They faxed information. They created physical coupons. It was, in Arregui's formulation, the "inconscience advantage" — the willingness to act when a full understanding of the problem's difficulty might have induced paralysis.
The breakthrough was behavioral, not technological. By holding funds in escrow until buyers confirmed satisfaction — and by building reputation systems that accumulated trust over time — Mercado Pago transformed the single largest barrier to e-commerce adoption in Latin America into a competitive moat. The payment system didn't just solve the trust problem; it solved the disintermediation problem. Once money flowed through Mercado Pago, transactions became visible, commissions became capturable, and the entire flywheel began to turn.
💳
From Fax Machines to Fintech
The evolution of Mercado Pago
2003Mercado Pago launches as an escrow-style payment solution for marketplace transactions.
2009Expands beyond the marketplace into off-platform payments — stores, peer-to-peer transfers.
2016eBay sells its stake; Tiger Global invests $75M, signaling a new growth phase.
2018QR code payments roll out across physical retail in Brazil and Argentina.
2020Pandemic accelerates adoption; Mercado Pago becomes a standalone fintech platform with tens of millions of users.
2025Reaches 72 million monthly active users; processes payments far exceeding the marketplace's own GMV.
What Arregui's team built, through sheer operational grit and regulatory arbitrage — building first, then forcing regulators to adapt because no frameworks existed for businesses like theirs — was something larger than a payment processor. It was the on-ramp to the formal economy for tens of millions of Latin Americans who had never had a bank account, never held a credit card, never been visible to the financial system. Mercado Pago gave them an identity in the digital economy. And once inside, they rarely left.
Ground Zero: Betting the Company on Architecture
In 2008, with the company freshly public — the IPO had raised $289 million on August 15, 2007, at $18 per share, making Mercado Libre the first Latin American tech company to list on NASDAQ — Galperin made what he later described as a "bet the company" decision. He ordered a complete rebuild of Mercado Libre's technology platform.
The existing system worked. It ran on an industrial-grade Oracle stack, handled millions of transactions, provided a functional user experience, and had supported the company through its IPO. By every conventional measure, it was fine. Galperin wanted to destroy it.
The project was internally called "Ground Zero" (and sometimes "New World"). The team would build an entirely parallel system — not just another application, but a fundamentally different architecture based on Web 2.0 standards, open and loosely coupled rather than monolithic and closed, designed for the mobile web that Galperin saw coming before most of his peers. For nearly two years, the company released essentially no new features while the parallel system was constructed. For a public company in a competitive market, this was an act of extraordinary discipline — or extraordinary recklessness, depending on your vantage point.
The gamble paid. The new platform enabled both technological and organizational scaling that would have been impossible under the old architecture. As COO Dani Rabinovich later noted, the hiring curve at Mercado Libre "looks completely flat for 10–13 years, and starts an exponential curve since we finished New World." The connection between technology architecture and human capital efficiency — the ability to add thousands of engineers without the system (or the organization) collapsing under its own weight — was the real dividend. By 2025, Mercado Libre would employ 18,000 engineers shipping 30,000 code deployments per day. That throughput was only possible because of a decision made in 2008 to burn the old world down.
The Logistics of the Impossible
If Mercado Pago solved the problem of how Latin Americans pay, Mercado Envíos — the logistics arm, launched in earnest around 2013 — attacked the problem of how goods actually reach people across a continent where the phrase "last-mile delivery" often meant "last-hundred-mile delivery over unpaved roads through the Amazon basin."
The scale of the challenge is difficult to overstate from the vantage point of a market with functioning postal services. Brazil alone is larger than the contiguous United States. Mexico's geography ranges from dense urban megacities to remote mountain villages accessible only by dirt track. Argentina stretches from subtropical north to Patagonian south. Infrastructure varies not just between countries but between neighborhoods. And the informal economy — street vendors, cash transactions, unregistered addresses — makes the data layer that modern logistics requires almost comically incomplete.
Mercado Libre's response was vertical integration of a kind that would have been irrational in a developed market. The company built its own warehouses, its own delivery fleet, its own sorting centers. In Brazil, it operates its own fleet of cargo aircraft. Same-day and next-day delivery, once unthinkable in the region, became a reality in major metropolitan areas. The investment was enormous, the returns uncertain, and the analysts who questioned the capital intensity were not wrong to do so — at least in the short term.
But the logic was the same logic that had governed every strategic decision since the garage: if the infrastructure doesn't exist, build it yourself, and then watch as the infrastructure becomes the moat. Amazon's entry into Latin America, beginning in 2012, actually accelerated this thinking. Rabinovich's counterintuitive assertion — "Amazon was the best thing that ever happened to us" — reflected the reality that competing against a company with Amazon's operational standards forced Mercado Libre to pursue a level of execution that might never have emerged organically. The threat was real. The response was to benchmark every operational decision against the best in the world and then build for local conditions that the best in the world didn't understand.
Amazon was the best thing that ever happened to us.
By the early 2020s, Mercado Envíos had achieved something remarkable: a logistics network that was not merely competitive with Amazon in Latin America but was, in many corridors, superior — because it had been built from the ground up for the specific pathologies of Latin American geography, regulation, and consumer behavior, rather than adapted from a template designed for a different world.
The Pandemic Accelerant
When COVID-19 arrived in early 2020, Latin America's e-commerce penetration was roughly 5%. Within twelve months, it had doubled to approximately 10%. Mercado Libre's stock surged nearly 200%, pushing its market capitalization to $82 billion and briefly making it the most valuable company in all of Latin America. The pandemic, in the grim calculus of platform economics, was an accelerant of historic proportions.
But the more interesting story is not the windfall — every e-commerce company on earth experienced pandemic tailwinds — but what Mercado Libre did with it. The company even temporarily changed its traditional handshake logo to an elbow bump, a small gesture that became a regional meme. Logistics volumes grew 5x almost overnight. Mercado Pago's off-platform adoption exploded as physical stores that had never accepted digital payments scrambled to go contactless. The credit arm, Mercado Crédito, extended financing to small merchants who had been locked out of the traditional banking system entirely.
The pandemic revealed the full scope of the ecosystem Mercado Libre had spent two decades building. It wasn't just a marketplace with ancillary services. It was the infrastructure layer — the pipes through which an entire continent's commerce and financial life could flow when the physical world shut down. Sellers who had previously sold in person now routed their customers through Mercado Libre, even when the initial contact happened face-to-face. The platform had become, in the language of systems theory, the default path.
Ariel Szarfsztejn, who managed Mercado Envíos through the pandemic's 5x volume shock and would later be named Galperin's successor as CEO, proved himself during this period. The logistics operation didn't collapse. It scaled. The organizational architecture — distributed decision-making, platform-first engineering, a culture of radical candor imported into a region where business culture traditionally avoided direct confrontation — held under pressure that would have broken a conventionally managed Latin American corporation.
The Fintech Flywheel Widens
Mercado Pago's evolution from marketplace payment tool to standalone fintech platform is the strategic pivot that transformed Mercado Libre from a large e-commerce company into something qualitatively different. By the mid-2020s, the majority of Mercado Pago's transaction volume occurred entirely off-platform — between individuals, at brick-and-mortar stores via QR code, through peer-to-peer transfers, and via a growing suite of financial products that included savings accounts, mutual funds, insurance, and cryptocurrency trading.
The numbers tell the story of this metamorphosis. By Q3 2025, Mercado Pago had reached 72 million monthly active users. Total payment volume far exceeded the marketplace's gross merchandise volume. Mercado Crédito, the lending arm, extended credit to millions of micro, small, and medium enterprises that traditional Latin American banks had never deigned to serve — or couldn't, because these businesses existed in the informal economy, invisible to conventional credit scoring.
Mercado Libre's credit underwriting exploited a data advantage that no traditional bank could replicate: years of transactional history on the marketplace. A seller's payment patterns, shipping reliability, customer reviews, and dispute history constituted a behavioral credit profile far richer than anything a FICO-equivalent score could provide. The company could price risk more accurately than incumbents because it could see risk more clearly.
We're always paranoid. We don't think it's game over by any means.
The fintech arm also created a powerful strategic lock-in. Once a small business owner's entire financial life — payments received, loans taken, savings accumulated, insurance purchased — flowed through Mercado Pago, switching costs became formidable. Not because of contractual lock-in or punitive fees, but because of the sheer friction of reconstituting an entire financial identity elsewhere. This was the same dynamic that had made the marketplace sticky, applied now to an even more fundamental human need: the management of money.
The Advertising Wedge
There is a pattern in the evolution of dominant platforms: first they aggregate demand, then they monetize attention. Mercado Libre's advertising business, Mercado Ads, followed this playbook with the precision of a company that had studied Amazon's trajectory carefully.
By the time Mercado Libre had achieved sufficient marketplace scale — hundreds of millions of product listings, over 100 million annual unique buyers — it possessed something that brand advertisers and performance marketers coveted: high-intent commercial data. A user searching for a specific product on Mercado Libre was not idly browsing; they were shopping. The conversion signal was extraordinarily clean. And in a region where digital advertising infrastructure remained fragmented and immature — where Google and Meta dominated but lacked the closed-loop purchase data that a marketplace generates — Mercado Ads offered something genuinely differentiated.
Sean Summers, who led the advertising business, articulated the competitive logic in the company's internal podcast series: the advertising revenue was essentially incremental margin on existing traffic. The marketplace had already paid to acquire the customer. The advertising business monetized that customer a second time, at margins far higher than commerce. This was, not coincidentally, the exact same strategic logic that had made Amazon's advertising business — which grew from nothing to over $40 billion in annual revenue — one of the most profitable segments in all of technology.
For Mercado Libre, still in the relatively early innings of advertising monetization, the trajectory pointed toward advertising becoming a meaningfully larger share of the revenue mix over time — high-margin revenue layered atop a commerce and fintech foundation, each reinforcing the other.
Eighteen Countries, One Operating System
The operational complexity of running a technology platform across 18 Latin American countries is staggering in ways that rarely surface in investor presentations. Each country has its own currency, its own central bank regulations, its own tax code, its own consumer protection laws, its own logistics infrastructure (or lack thereof), its own cultural relationship to commerce and trust. Argentina's chronic inflation requires constant repricing logic. Brazil's Byzantine tax system — which varies by state, by product category, by whether the moon is waxing — demands specialized compliance infrastructure. Mexico's drug cartels create security challenges for last-mile delivery that no American logistics operator has ever contemplated.
Mercado Libre's response was to build a core platform that could be locally adapted rather than locally rebuilt. The 18,000-engineer organization operates with a ratio of roughly 18 engineers to every 1 product manager — a 5% PM-to-engineer ratio compared to the 20–30% typical at American tech companies. This isn't a staffing oversight; it's a deliberate architectural choice. Engineers are hired for product thinking, not just coding skill. The interview process tests for the ability to understand both technical possibility and user need. "We don't let titles determine who owns the product," as Sebastian Barrios, the longtime head of product and engineering, put it.
The result is an organization that can ship at extraordinary velocity — 30,000 code deployments per day — while maintaining coherence across radically different operating environments. An internal platform handles scaling, security, and compliance automatically, freeing individual engineering teams to focus on user value. This platform-first approach is what allowed Mercado Libre to scale from thousands to tens of thousands of engineers without the organizational collapse that typically accompanies such growth.
Brazil alone accounts for approximately 65% of revenue; add Argentina and Mexico, and the concentration rises to roughly 96%. The remaining 15 countries are strategically important — they extend the network, provide optionality, and create barriers to entry for competitors who must replicate this multi-country complexity — but the business is, in practice, a three-country story with a long tail.
The Succession
On May 21, 2025, twenty-six years after founding Mercado Libre, Marcos Galperin announced he would step down as CEO. The transition to Ariel Szarfsztejn — an internal executive who had risen through logistics, proving himself during the pandemic's volume explosion before assuming broader operational responsibility — had been methodically architected over three years.
Galperin's approach to succession revealed the same long-term, systems-level thinking that had characterized every major strategic decision. Stelleo Tolda, his Brazilian co-pilot of more than two decades, was moved aside — not as a demotion but as deliberate positioning to give Szarfsztejn visibility and scope. Szarfsztejn was tasked with quarterly investor meetings, introductions to every major shareholder, gradual assumption of company-wide decision-making authority. "For me, if it were someone from the outside, it would've been a sign of personal failure," Galperin said. "Because I have always found it very important to work with people who I think are better than me."
For me, if it were someone from the outside, it would've been a sign of personal failure. Because I have always found it very important to work with people who I think are better than me.
The transition, by multiple accounts, was shockingly fast in practice. "We announced a period of transition, which in practice lasted a week," Galperin told the Brazilian business magazine Exame. "Quickly, Ari began leading, and we understood him to be the new CEO." Galperin moved to the chairman role, maintaining strategic oversight while ceding operational control to a 44-year-old executive who had already been making the decisions. The CEO title simply formalized what was already true.
What made the succession possible — what separated it from the founder-to-professional-manager transitions that have destroyed value at countless technology companies — was that Galperin had built an organization where leadership was genuinely distributed. Sean Summers, a senior executive, described being "shocked by how 'leaderless' decisions felt" upon joining the company — not because nobody was leading, but because ideas mattered more than titles. The founder was systemically replaceable even as his imprint remained indelible.
Szarfsztejn inherited a business with 120,000 workers serving 130 million clients monthly, with a market value exceeding $100 billion. Still, as he told CNN in December 2025: "We have a lot to do in Latin America."
The Paranoia Principle
There is a phrase that recurs in every interview, every podcast, every earnings call, every internal meeting — a phrase that functions less as corporate motto than as operating system: "Only the paranoid survive." Galperin borrowed it from
Andy Grove's Intel, but the application is distinctly Latin American. In a region where macroeconomic stability is the exception rather than the rule, where political risk can materialize overnight, where a currency devaluation can erase a quarter's earnings before the quarter is over, paranoia is not a psychological disorder. It is a survival strategy.
This paranoia manifests as constant competitive benchmarking, obsessive NPS measurement, quarterly reviews that interrogate every business unit with genuine intellectual honesty, and — crucially — a willingness to cannibalize existing revenue streams before competitors do. The marketplace evolved from auctions to fixed-price retail. Mercado Pago evolved from marketplace escrow to standalone fintech. Mercado Envíos evolved from third-party logistics aggregation to a proprietary fleet with its own cargo aircraft. Each evolution required destroying a business model that was working in order to build one that might work better.
There's still so much to do, there's still so many opportunities and the fact that we're willing to take risks and make big bets to continue growing and the fact that we're growing at the same rates that we were growing 25 years ago is amazing.
The culture that enables this — hard work, meritocracy, excellence, entrepreneurship, long-term focus, in Galperin's own enumeration — is deliberately countercultural in a Latin American business context. Direct feedback, radical candor, the willingness to kill profitable features because they confused users (with revenue actually going up long-term as a result): these are norms that Silicon Valley takes for granted but that required active construction in a region where business culture traditionally prizes hierarchy and conflict avoidance.
Galperin's personal routine offers a small window into the texture of this intensity. He thinks best at five in the morning, "half asleep, half awake," a notepad on the nightstand. Ideas come during exercise. The CEO of Latin America's most valuable company, worth over $7 billion personally, operating from his office in Montevideo — he relocated from Buenos Aires to Uruguay, a move that itself spoke to the political volatility of the Argentine operating environment — still runs with the metabolism of a founder.
More than one million families now derive their primary income through the Mercado Libre ecosystem. The company itself employs over 120,000 people across the continent. The stock, which IPO'd at $18 per share in August 2007, has compounded at rates that place it among the great wealth-creation stories in the history of public markets. For 27 consecutive quarters — nearly seven years — revenue grew above 30%, a streak unmatched by any other public company among over 80,000 globally.
And still the e-commerce penetration rate in Latin America sits at roughly 14%. In the United States and China, it is closer to 30%. The gap is the opportunity. The paranoia is the fuel. In the garage in Saavedra, there was a leather factory above and a dial-up connection below. Twenty-six years later, the planes fly at night across Brazil, carrying packages through the dark toward addresses that didn't exist on any map when Marcos Galperin first sold his Volkswagen Golf online.
