MercadoLibre and the Spreadsheet Trick That Decides Everything

Quick question before we begin. I'm going to give you one number — MercadoLibre's free cash flow last quarter — and I want you to guess what the stock is worth.

Ready? It was negative $56 million.

If you just mentally filed MercadoLibre under "avoid," you're not wrong to have that instinct — you're just missing the plot twist. That number is real, it's GAAP, and it's also one of the more misleading data points in large-cap tech-adjacent investing right now. Understanding why it's misleading is worth more than the number itself. That's the whole piece.

The business, in one paragraph, because you deserve context before I provide my reasoning.

MercadoLibre is the dominant e-commerce and fintech platform across Brazil, Mexico, and Argentina — Amazon and PayPal fused into one company, operating in a region where the average shopper makes 7 online purchases a year versus 41 in the US. Revenue grew 49% last quarter, the fastest pace since 2022. Unique buyers hit 84.1 million, up 26%. Brazil alone added 17 million new buyers in twelve months — roughly the population of the Netherlands deciding to start shopping on one app, in one year. This is not a company running out of runway. If anything, it's a company that just found a second one underneath the first.

That second runway is Mercado Pago, its fintech arm, which is quietly becoming Latin America's largest digital bank while most of the market still mentally files MercadoLibre under "the eBay of Brazil." Fintech users are up 29% to nearly 83 million. The credit book nearly doubled to $14.6 billion. And that credit book is where our negative-$56-million mystery lives.

Every line above is a green light except the last one. That's not an accident of formatting — it's the entire tension of this piece sitting in one table.

Where the cash actually went

Building a loan book fast costs cash up front — you hand out money now, collect interest for years. MercadoLibre spent $1.29 billion in a single quarter funding new loans. That spending shows up on the income statement as a cash outflow, which is technically true and also the least useful way to think about it, in the same way that a landlord buying a tenth apartment building "loses money" the month they buy it.

Management has drawn the obvious comparison themselves: this looks a lot like Amazon funding AWS's build-out in the early 2010s. Reported cash flow looked anemic while the actual earning asset compounded quietly underneath it. Nobody now thinks that period should have scared Amazon investors off. The question is whether MercadoLibre's credit book deserves the same benefit of the doubt — and unlike Amazon in 2011, we don't get to know the ending in advance.

Two spreadsheets, two very different companies

Here's the part that should make you smarter about valuation generally, not just about this one stock: the exact same business, run through two defensible cash flow models, produces two wildly different verdicts. Neither model is dishonest. They just disagree about what "cash flow" means.

Model one — take the GAAP number at face value. Use the trailing free cash flow as reported, roughly $1.37 billion. Run a standard discounted cash flow. Fair value comes out around $2,310 a share. Against today's price of about $1,854, that's a real but unimpressive 19.7% discount — a "fine, I guess" number that falls short of the 25% margin of safety most disciplined value investors would want before buying more.

Model two — normalize for the credit-funding drag, treating the cash consumed building the loan book as investment rather than expense, the same logic applied to Amazon's capex or Meta's Reality Labs losses. Normalized core free cash flow comes to roughly $4 billion. Run the same DCF machinery — 30% growth for five years, moderating to 18% for the next five, a 4.5% terminal rate — and fair value jumps to $6,727 a share. Against $1,854, that's a 72% discount, the kind of number that makes a value investor's pulse do something undignified.

Same company. Same quarter. A $4,400-a-share gap in "fair value," entirely explained by one modeling choice about whether a number on the cash flow statement represents a cost or a purchase.

Two spreadsheets, side by side

That's the whole disagreement in one table. Same discount rate, same company, same quarter — and a purchase decision that flips from "pass" to "top priority" depending entirely on row one.

So which spreadsheet is lying to you?

Neither, technically. That's the uncomfortable part. Both are internally consistent DCFs built on real inputs. The disagreement isn't about math — it's about a judgment call: is MercadoLibre's credit book a temporary, self-funding investment phase, or a structurally cash-hungry business that just got bigger and hungrier?

The market, by pricing the stock roughly halfway between the two fair values, seems to be hedging exactly the same way I am. Nobody's fully pricing in either model. That's not the market being inefficient — that's the market being appropriately unsure, which is a more useful thing for a stock price to communicate than false confidence in either direction.

I currently lean toward the normalized case, for three reasons that are checkable rather than vibes: reserve coverage on past-due loans sits at 103–149%, which is not what an underwriting problem looks like; the loan yield (NIMAL) is a healthy 17.8%; and the funding pattern matches a genuine reinvestment cycle rather than a business quietly bleeding cash to stay afloat. But "I lean toward it" and "it's confirmed" are different sentences, and I'm not going to write the second one before the data exists.

The three numbers that settle the argument

Q2 earnings resolves this, not vibes, not my priors. Watch exactly three things:

1. Does Adjusted FCF flip positive again? One bad quarter is a data point. Two is a pattern.

2. Does the credit-funding pace moderate, or get visibly offset by the loan book maturing? This is the difference between "investment phase" and "permanent cash drag with better PR."

3. Does loan yield hold near 17.8%? If NIMAL starts compressing while the book keeps doubling, that's underwriting discipline cracking under growth pressure — the first symptom, not the last.

Clean beat on all three, and this graduates from "conviction call" to "confirmed thesis." A miss — especially a second negative FCF quarter — and the honest fair value slides back toward $2,310, which doesn't make MercadoLibre a bad company, but does make it a considerably less exciting stock at today's price.

Why I'm not waiting for the answer to have an opinion

I hold a position sized for real conviction, not maximum conviction — because the marketplace business alone, growing high-40s% in a region a third as digitally mature as the US, is a legitimately good business even under the boring spreadsheet. The credit book is the difference between "good position" and "best position I own." I think the odds favor confirmation. I don't think the odds are so overwhelming that pretending there's no risk would make me, or you, any more informed about it.

Fair value estimates above reflect my own independently built DCF models under both the as-reported and normalized cash flow scenarios.

Disclosure: I own MercadoLibre shares and may increase my position if the share price reaches levels that provide a larger margin of safety. This article reflects my personal investment research and is not financial advice.