Fastest Growth in Four Years, Lowest Margin in Nine Quarters: MercadoLibre Sets the Dial and Tells You It Is Staying There
Key Takeaways
- Revenue of $8,845M grew 49% YoY, the fastest rate since 2022, and beat consensus by roughly 5%. Income from operations fell 20% to $611M and operating margin compressed 595bps to 6.9%, so the stock fell 12.7% to $1,632.52 on 6.0x normal volume.
- The sell-off was triggered by a sentence, not a number. The shareholder letter said management does "not anticipate this changing materially in the near term," which converted a bad quarter into a bad year and removed the recovery catalyst the Street had been underwriting.
- Consensus is arithmetically incompatible with that sentence. The Street's FY26 EPS requires a 199bps net-margin recovery over the remaining nine months; hold Q1's margin flat and FY26 EPS lands near $36 rather than $47.84, which puts the shares at 45x, not the 34x the headline multiple advertises.
- Credit is doing the damage and is the least-discussed risk. Provisions doubled to $1,244M, annualised cost of risk reached roughly 37%, allowance coverage rose to 26.2% of the gross book and total past-due loans reached 27.8%, while management deliberately extended Brazilian consumer loan duration from five months to eight.
- Rating: Initiating at Hold. The investments look genuinely productive and the franchise is strengthening, but estimates have to fall before the multiple means anything, and nothing in management's language suggests a 2026 margin inflection.
Results vs. Consensus
MercadoLibre does not issue formal financial guidance, so there is no prior guide to score this print against. The comparison that matters is against the Street, and it splits cleanly: a large top-line beat and a clear bottom-line miss.
| Metric | Q1 2026 Actual | Consensus | Beat/Miss | Magnitude |
|---|---|---|---|---|
| Net revenues and financial income | $8,845M | $8.44B | Beat | +4.9% |
| Gross margin | 43.7% | n/a | Down YoY | -302bps |
| Income from operations | $611M | n/a | Down YoY | -19.9% |
| Operating margin | 6.9% | n/a | Down YoY | -595bps |
| Adjusted EBITDA | $857M | n/a | Down YoY | -8.3% |
| Net income | $417M | n/a | Down YoY | -15.6% |
| EPS (diluted, GAAP) | $8.23 | $8.78 | Miss | -6.3% |
| Adjusted free cash flow | ($56M) | n/a | Negative | vs. +$58M |
Consensus providers disagreed on the pre-print bar by a wide margin this quarter: revenue estimates ranged from $8,292M to $8,436M and EPS estimates from $8.47 to $8.79. The direction is unanimous across every provider, which is what matters. Revenue beat on every basis. EPS missed on every basis. The figures used above sit at the most widely cited end of each range, and no line-item consensus exists for gross margin, operating income or EBITDA, so those rows are scored against the prior-year actual and labelled accordingly.
Year-over-Year Comparison
| $M unless noted | Q1 2026 | Q1 2025 | YoY | % of revenue Q1'26 | % of revenue Q1'25 | bps change |
|---|---|---|---|---|---|---|
| Net revenues and financial income | 8,845 | 5,935 | +49.0% | 100.0% | 100.0% | n/a |
| Cost of net revenues and financial expenses | (4,983) | (3,164) | +57.5% | 56.3% | 53.3% | +302 |
| Gross profit | 3,862 | 2,771 | +39.4% | 43.7% | 46.7% | -302 |
| Product and technology development | (699) | (551) | +26.9% | 7.9% | 9.3% | -138 |
| Sales and marketing | (982) | (599) | +63.9% | 11.1% | 10.1% | +101 |
| Provision for doubtful accounts | (1,244) | (603) | +106.3% | 14.1% | 10.2% | +390 |
| General and administrative | (326) | (255) | +27.8% | 3.7% | 4.3% | -61 |
| Total operating expenses | (3,251) | (2,008) | +61.9% | 36.8% | 33.8% | +292 |
| Income from operations | 611 | 763 | -19.9% | 6.9% | 12.9% | -595 |
| Adjusted EBITDA | 857 | 935 | -8.3% | 9.7% | 15.8% | -606 |
| Net income | 417 | 494 | -15.6% | 4.7% | 8.3% | -361 |
| Diluted EPS | $8.23 | $9.74 | -15.5% | n/a | n/a | n/a |
Sequential Comparison
MercadoLibre does not publish a standalone fourth-quarter income statement. The Q4 2025 column below is derived as the reported full year less the reported nine months ended September 30, 2025, and it is internally consistent: gross profit of $3,784M less the four operating expense lines totalling $2,895M gives $889M, the derived operating income.
| $M unless noted | Q1 2026 | Q4 2025 | QoQ | % of revenue Q1'26 | % of revenue Q4'25 | bps change |
|---|---|---|---|---|---|---|
| Revenue | 8,845 | 8,759 | +1.0% | 100.0% | 100.0% | n/a |
| Gross profit | 3,862 | 3,784 | +2.1% | 43.7% | 43.2% | +46 |
| Product and technology development | 699 | 584 | +19.7% | 7.9% | 6.7% | +124 |
| Sales and marketing | 982 | 1,036 | -5.2% | 11.1% | 11.8% | -73 |
| Provision for doubtful accounts | 1,244 | 983 | +26.6% | 14.1% | 11.2% | +284 |
| General and administrative | 326 | 292 | +11.6% | 3.7% | 3.3% | +35 |
| Income from operations | 611 | 889 | -31.3% | 6.9% | 10.2% | -324 |
| Net income | 417 | 559 | -25.4% | 4.7% | 6.4% | -167 |
| Diluted EPS | $8.23 | $11.03 | -25.4% | n/a | n/a | n/a |
The sequential table carries the single most under-appreciated fact of the quarter: gross margin rose 46bps QoQ. The shipping investment that dominated the narrative for three quarters is no longer the marginal pressure. The entire sequential operating-margin decline of 324bps is provisions at 284bps plus product development seasonality. Anyone still modelling this as a free-shipping story is modelling the previous quarter's problem.
- Revenue: High quality and broad. FX flattered the headline by 3.5 points (49.0% reported, 45.5% FX-neutral) and on an FX-neutral basis growth decelerated slightly, from 47% to 46%, so the reported acceleration is currency rather than volume: items sold grew 46.7% against GMV of 42.2%. No acquisitions, no one-time items.
- Margins: Low quality in the sense that matters least and high quality in the sense that matters most. The compression is deliberate and self-inflicted rather than competitive, but it is also structural rather than one-time. Of the 595bps, 390bps is provisions and 302bps is gross margin, partly offset by 138bps of product development leverage and 61bps of G&A leverage.
- EPS: The miss is smaller than the operating decline because two items below the line helped. Foreign currency swung to a $6M gain from a $55M loss, and the effective tax rate fell to 28.0% from 30.0%. Strip those and the operational miss is wider than the reported $0.55.
Revenue
This is the strongest top-line quarter MercadoLibre has printed in four years, and the composition is better than the headline. FX-neutral growth of 45.5% means most of the reported growth survives the currency adjustment, though none of the acceleration does. Underneath, Brazil accelerated to 39% FX-neutral revenue growth, Mexico held at 39%, and Argentina decelerated to 65% as its inflation-driven base effect continues to unwind. The commerce and fintech split inverted for the first time in several quarters: commerce accelerated to 39% FX-neutral while fintech decelerated to 54%, which is the arithmetic of the free-shipping investment finally showing up as revenue rather than only as cost.
Margins
The operating margin bridge decomposes cleanly and every component ties to the reported income statement.
| Driver of the 595bps operating margin decline | bps | Character |
|---|---|---|
| Provision for doubtful accounts | -390 | Credit-book growth outrunning revenue growth, plus Brazil consumer underwriting |
| Gross margin | -302 | Lower Brazil free-shipping threshold plus 1P mix |
| Sales and marketing | -101 | Affiliate channel scaling, buyer protection, headcount |
| Product and technology development | +138 | Structural leverage from flat engineering headcount plus AI productivity |
| General and administrative | +61 | Scale leverage |
| Total | -595 | Two-thirds credit, one-third commerce |
The framing management used on the call and in the letter attributes the compression to investment choice. That is accurate but incomplete. Provisions at 390bps are the largest single driver, and provisions are not an investment in the sense that free shipping is: they are the expected-loss charge on a book that grew 87% YoY while revenue grew 49%. Under current expected credit loss accounting the full lifetime loss is booked at origination, so as long as originations accelerate the charge stays elevated. That is a real economic cost with an accounting timing overlay, and it does not reverse simply because management decides to stop investing.
EPS
Diluted EPS of $8.23 fell 15.5% against $9.74, on a share count that has not moved. Weighted average diluted shares were 50,697,298 against 50,697,375 a year ago, and shares outstanding of 50,697,182 are unchanged from December 31. There were no buybacks, so every dollar of the EPS decline is earnings, not denominator. For a company generating $2,075M of operating cash flow in the quarter, the absence of any repurchase is a deliberate capital-allocation statement: the cash is going into the credit book, which absorbed $1,949M of net funding in the quarter.
Segment Performance
MercadoLibre reports by geography, and the geographic contribution table is where the investment cycle becomes legible. Direct contribution fell 3.3% on revenue growth of 49.0%, so segment contribution margin compressed 859bps to 15.9%. Every geography gave up contribution margin, Mexico by only 34bps, and the damage is heavily concentrated elsewhere.
| Segment | Revenue Q1'26 | Revenue Q1'25 | Revenue YoY | Direct contribution Q1'26 | DC YoY | DC margin Q1'26 | DC margin Q1'25 | Change |
|---|---|---|---|---|---|---|---|---|
| Brazil | $4,774M | $3,082M | +54.9% | $389M | -28.2% | 8.1% | 17.6% | -943bps |
| Mexico | $1,976M | $1,222M | +61.7% | $344M | +58.5% | 17.4% | 17.8% | -34bps |
| Argentina | $1,698M | $1,382M | +22.9% | $607M | -6.3% | 35.7% | 46.9% | -1,113bps |
| Other Countries | $397M | $249M | +59.4% | $64M | +42.2% | 16.1% | 18.1% | -195bps |
| Total segments | $8,845M | $5,935M | +49.0% | $1,404M | -3.3% | 15.9% | 24.5% | -859bps |
| Unallocated opex and indirect costs | n/a | n/a | n/a | ($793M) | +15.1% | 9.0% of rev | 11.6% of rev | -264bps |
| Income from operations | n/a | n/a | n/a | $611M | -19.9% | 6.9% | 12.9% | -595bps |
One structural point deserves emphasis before the segment detail. The unallocated corporate cost pool grew 15.1% against revenue growth of 49.0%, delivering 264bps of leverage. This is the only large cost line in the business behaving the way a scaling platform's cost lines are supposed to behave. If the segments were holding contribution margin flat, that corporate leverage alone would have expanded operating margin materially. The entire margin story sits inside the segments.
Brazil
Brazil is simultaneously the best operating story and the worst margin story in the company, and both facts descend from the same decision. Revenue grew 54.9% in dollars and 39% FX-neutral, GMV grew 38% FX-neutral, and items sold grew 56% against 45% last quarter, 42% the quarter before and 26% the quarter before that. Unique buyer growth accelerated to 32%, described as the fastest in five years. Contribution margin fell 943bps to 8.1%, and direct contribution in absolute dollars fell 28.2% even as revenue grew by $1,692M.
"Brazil GMV growth accelerated to 38% YoY on an FX-neutral basis, supported by items sold growth of 56% YoY, up from 45% in Q4'25, 42% in Q3'25 and 26% in Q2'25. Our volume growth has doubled from a high base in just nine months."
— Letter to Shareholders
The offsetting mechanism is working faster than the cost. Brazil unit shipping cost fell 17% YoY in local currency, accelerating from an 11% decline in Q4, and management disclosed that variable contribution per shipment in the R$19 to R$79 band has recovered enough that some price brackets are now at breakeven. That is the specific number that determines whether this investment ever pays back, and it moved the right way.
Assessment: The elasticity response in Brazil is genuinely exceptional and the unit-cost curve is bending faster than management originally guided, which is the strongest evidence in the print that this is investment rather than erosion. The problem is arithmetic rather than strategic: at an 8.1% contribution margin, Brazil now generates less absolute contribution on $4,774M of revenue than it did on $3,082M a year ago, and it is 54% of the company's revenue. Brazil has to inflect for the consolidated margin to inflect, and management has just told the market it will not be pushed to.
Mexico
Mexico is the quarter's quiet standout and the only geography that grew contribution roughly in line with revenue. Revenue grew 61.7% in dollars and 39% FX-neutral, direct contribution grew 58.5%, and contribution margin held at 17.4% against 17.8%, a 34bps decline that is effectively flat. Mexico now generates $344M of direct contribution against Brazil's $389M on 41% of Brazil's revenue.
The operational picture is more mixed than the contribution line suggests. Items sold grew 34% but FX-neutral GMV growth of 28% decelerated sequentially, which management attributed to this year's tax reform pushing small and medium sellers to raise prices into an already cautious consumer. Management is simultaneously increasing credit card issuance in Mexico on the strength of what it called attractive projected paybacks, so the country is absorbing an incrementally larger share of the credit investment even as its commerce growth cools.
Assessment: Mexico is proof that MercadoLibre can grow at these rates without giving up contribution margin, which weakens the argument that the compression elsewhere is unavoidable. Watch the GMV deceleration rather than the revenue line: a tax-driven seller price increase into a cautious consumer is a demand problem, and it is the one genuinely exogenous headwind disclosed this quarter.
Argentina
Argentina lost 1,113bps of contribution margin, the largest decline of any segment, and it did so on the slowest revenue growth in the company at 22.9%. FX-neutral revenue growth of 65% decelerated from 77% in Q4 and 97% in Q3, continuing the unwind of the inflation-driven base that flattered 2025. Direct contribution fell 6.3% to $607M, which is still the largest contribution pool in the company on the third-largest revenue base.
Two things are happening at once. The FX tailwind that inflated Argentine revenue in dollars has become a headwind, and the country is now absorbing early-stage credit card investment. Management launched the Argentine card in August or September of last year, and early-stage card cohorts are loss-making by construction because the full expected credit loss on the committed line is provisioned at issuance. The credit quality commentary was notably confident: 15-90 day non-performing loans fell sequentially while spreads increased, against a banking system where management said delinquency is rising.
"While the financial system in Argentina is seeing rising delinquency, our 15-90 day NPL fell sequentially in Q1'26 while spreads increased. This demonstrates the strength of our underwriting and our agility."
— Letter to Shareholders
Assessment: A 35.7% contribution margin is still extraordinary and Argentina remains the profit engine that funds the rest of the investment programme. But this segment has now given up 1,113bps of margin while decelerating, and it is the geography most exposed to a macro reversal. The credit data is genuinely reassuring; the margin trajectory is not.
Other Countries
Revenue grew 59.4% to $397M with contribution up 42.2% to $64M and margin down 195bps to 16.1%. Chile delivered 40% FX-neutral GMV growth on record free-shipping penetration and faster deliveries. Colombia and Peru were called out as beneficiaries of cross-border trade, where a thinner local seller base means imported assortment expands the addressable catalogue directly rather than substituting for it.
Assessment: Small enough not to move consolidated results at 4.5% of revenue, but it is the cleanest read on whether the playbook travels. Growing contribution 42% while investing is a better outcome than either Brazil or Argentina managed.
Revenue Composition: Commerce vs. Fintech
The geographic segments are the reported structure, but the product disaggregation explains the margin. It shows that the fastest-growing revenue line in the company is also the one carrying the provision charge.
| Revenue stream | Q1 2026 | Q1 2025 | YoY | Mix Q1'26 | Mix Q1'25 |
|---|---|---|---|---|---|
| Commerce services | $3,756M | $2,701M | +39.1% | 42.5% | 45.5% |
| Commerce product sales (1P) | $1,112M | $602M | +84.7% | 12.6% | 10.1% |
| Total commerce | $4,868M | $3,303M | +47.4% | 55.0% | 55.7% |
| Financial services and income | $1,947M | $1,473M | +32.2% | 22.0% | 24.8% |
| Credit revenues | $2,012M | $1,146M | +75.6% | 22.7% | 19.3% |
| Fintech product sales | $18M | $13M | +38.5% | 0.2% | 0.2% |
| Total fintech | $3,977M | $2,632M | +51.1% | 45.0% | 44.3% |
| Total | $8,845M | $5,935M | +49.0% | 100.0% | 100.0% |
Commerce
Commerce grew 47.4%, with 1P product sales growing 84.7% and now 12.6% of total revenue against 10.1% a year ago. 1P generated almost $5bn of GMV over the trailing twelve months and grew 69% FX-neutral. Management is explicit that 1P remains a drag on income from operations, having achieved positive direct contribution in 2024, with the remaining gap being corporate cost allocation that grows as 1P outgrows total GMV. Cross-border trade grew 68% FX-neutral. Advertising grew 63% FX-neutral and 73% in dollars, described as four times market growth in 2025.
Assessment: The commerce mix is deliberately shifting toward the two lines that dilute gross margin, 1P and free-shipped low-ticket orders, and away from the pure-take-rate marketplace revenue that carried the historical margin structure. Advertising is the offset that should eventually re-rate commerce gross margin, and at 73% dollar growth it is compounding fast, but it is not disclosed in dollars so its contribution cannot be sized from outside.
Fintech
Fintech grew 51.1%, but the internal composition matters more than the total. Credit revenues grew 75.6% to $2,012M and are now 22.7% of total company revenue, while the lower-risk financial services and income line grew 32.2%. Fintech monthly active users reached 83mn, up 29.7%, with assets under management approaching $20bn, up 77%. Credit card total payment volume grew 90% and card monthly active users grew 68%, with 2.7mn cards issued in the quarter.
Assessment: Fintech's revenue growth is increasingly credit growth, and credit revenue is booked gross of a provision charge that is growing faster still. Provisions consumed 61.8% of credit revenue this quarter against 52.6% a year ago. The revenue line looks like a fintech platform compounding; the net contribution looks like a lender in an origination-growth phase.
Key Operating KPIs
| KPI | Q1 2026 | Q1 2025 | YoY | Read |
|---|---|---|---|---|
| Fintech monthly active users | 83mn | 64mn | +29.7% | Nearly 20mn net adds in a year |
| Unique active buyers | 84mn | 67mn | +25.4% | Brazil buyer growth of 32% led |
| Gross merchandise volume | $18,951M | $13,330M | +42.2% | +36% FX-neutral |
| Items sold | 722mn | 492mn | +46.7% | Outgrew GMV by 4.5 points |
| Total payment volume | $87,186M | $58,303M | +49.5% | +55% FX-neutral |
| Acquiring TPV | $55,993M | $40,317M | +38.9% | Share gains in all four core markets |
| Total payment transactions | 4,640mn | 3,344mn | +38.8% | |
| NIMAL | 17.8% | 22.7% | -490bps | Card mix plus Brazil consumer spread |
| Capital expenditures | $271M | $256M | +5.9% | Intensity fell to 3.1% from 4.3% |
| Depreciation and amortisation | $246M | $172M | +43.0% | Prior fulfilment capex depreciating |
Unit Economics Underneath the KPIs
| Derived metric | Q1 2026 | Q1 2025 | Change | What it shows |
|---|---|---|---|---|
| Average selling price per item | $26.25 | $27.09 | -3.1% | The signature of the free-shipping threshold pulling in lower-ticket orders |
| Items per unique buyer | 8.60 | 7.34 | +17.0% | Frequency, the metric the investment is buying |
| GMV per unique buyer | $225.61 | $198.96 | +13.4% | Frequency gains outrun the ticket decline |
| Commerce services yield on GMV | 19.8% | 20.3% | -44bps | Take-rate proxy only; numerator includes ads and subscriptions |
| Non-acquiring TPV | $31,193M | $17,986M | +73.4% | Wallet and off-platform volume growing twice as fast as acquiring |
| Average payment ticket | $18.79 | $17.44 | +7.8% | |
| Capex intensity | 3.1% | 4.3% | -120bps | Capex is not where the investment is going this cycle |
The last row is the one that reframes the whole debate. Capital expenditure grew 5.9% against revenue growth of 49.0%, and capex intensity fell 120bps. Whatever else is happening, this is not a capital investment cycle in the conventional sense. The spending is running through the income statement as shipping subsidy, marketing and credit provisions, which is why it hits margin rather than the balance sheet, and why it can be reversed faster than a fulfilment build could be. That cuts both ways: it is more discretionary, and therefore management's decision to keep spending is more genuinely a decision.
The Credit Book
Credit deserves separate treatment because it drove 390bps of the 595bps margin decline and because it is the part of the business where the disclosure is richest and the market's attention was thinnest. The portfolio reached $14,555M gross, up 16.4% in a single quarter and 87% YoY, with the largest nominal quarterly increase in the company's history.
| Item | Mar 31, 2026 | Dec 31, 2025 | Change |
|---|---|---|---|
| Loans receivable, gross | $14,555M | $12,508M | +16.4% |
| Merchant | $2,285M | $2,009M | +13.7% |
| Consumer | $5,319M | $4,559M | +16.7% |
| Credit cards | $6,622M | $5,656M | +17.1% |
| Asset-backed | $329M | $284M | +15.8% |
| Allowance for doubtful accounts | $3,820M | $3,143M | +21.5% |
| Allowance as % of gross portfolio | 26.2% | 25.1% | +112bps |
| 15-90 days past due | $1,157M / 7.9% | $950M / 7.6% | +35bps |
| 90-360 days past due | $2,557M / 17.6% | $2,098M / 16.8% | +79bps |
| Total past due | $4,041M / 27.8% | $3,295M / 26.3% | +142bps |
| Renegotiated loans as % of portfolio | 1.4% | 1.6% | -20bps |
| Write-offs in the quarter | $658M | $330M (Q1'25) | +99.4% |
| Off-balance-sheet unused card commitments | $11,892M | $9,001M | +32.1% |
| Derived credit metric | Q1 2026 | Q1 2025 | Read |
|---|---|---|---|
| Annualised cost of risk | 36.8% | n/a | Provision x4 over average gross portfolio; the figure an analyst put to management as "around 37%" |
| Annualised credit revenue yield | 59.5% | n/a | Gross yield remains far above cost of risk |
| Provision as share of credit revenue | 61.8% | 52.6% | The key deterioration; 920bps worse YoY |
| Gain on sales of credit card receivables | $367M (18.2% of credit revenue) | $255M (22.3%) | Reliance on this item fell, which is a quality positive |
| Credit cards as % of portfolio | 45.5% | 45.2% (Dec) | Mix shift toward the lower-NIMAL product |
Two readings are available and the truth is genuinely between them. The bull reading is that a 59.5% gross yield against a 36.8% cost of risk is a wide, profitable spread; that 15-90 day non-performing loans at 8.0% are broadly stable YoY; that renegotiated loans fell as a share of the book; and that the reliance on gains from selling card receivables actually declined, which means less of the credit revenue is coming from balance-sheet management. The bear reading is that total past-due loans rose 142bps QoQ to 27.8% of the book, allowance coverage rose 112bps to 26.2%, write-offs doubled, and the 90-360 day bucket, which is the one that converts to loss, grew faster than the 15-90 day bucket that management chose to highlight.
The disclosure asymmetry is worth naming. Management repeatedly cited the 15-90 day metric, which was stable, and never addressed the 90-360 day bucket, which deteriorated 79bps. Both are in the filings. Only one was in the narrative.
Key Topics & Management Commentary
Overall Management Tone: Unusually direct and, on the central question, unusually pre-emptive. Rather than framing the margin as a temporary artefact, management volunteered forward language stating the level would not change materially in the near term, then defended that choice consistently across thirteen questions without once offering a recovery timeline. Confidence on the commerce investments was evidence-rich and specific; confidence on credit was assertive but leaned on the one asset-quality metric that had not deteriorated. The posture was that of a management team that has decided the market's reaction is a cost worth paying rather than a misunderstanding to be corrected.
1. The Margin Dial and the Sentence That Moved the Stock
The letter contained a forward-looking statement of a kind MercadoLibre almost never makes. The company does not guide, and its standard posture is to decline forward margin commentary entirely. This quarter it volunteered a directional statement, and the market repriced on it.
"We have the ability to dial margins up or down as circumstances and opportunities evolve"
— Letter to Shareholders
The clause that followed is the one that mattered: management does "not anticipate this changing materially in the near term." A single quarter at 6.9% is a data point that a recovering business grows out of. A statement that 6.9% is roughly where the dial is set converts that data point into a run-rate, and the Street's entire FY26 earnings structure was built on the former reading.
Assessment: This was a deliberate act of expectation management, and it was the correct one. Management chose to take the de-rating now rather than miss consensus three more times. The cost is that it removed the sole remaining catalyst for the multiple and gave the market permission to mark FY26 estimates down hard. Investors should read this as an honest signal, not a bearish one, but honest signals still require the model to change.
2. Brazil Free Shipping: the Elasticity Worked and the Cost Curve Is Bending
The lower free-shipping threshold in Brazil is now three quarters into its life and is the most-evidenced investment in the print. Items sold growth went 26%, 42%, 45%, 56% across the four quarters since the change. Unique buyer growth accelerated to 32%. Conversion, frequency, retention and net promoter score were all described as at record highs, and management noted new cohorts joining since the change purchase more items across broader categories with higher retention than older cohorts.
The economics moved in the same direction. Unit shipping cost in Brazil fell 17% YoY in local currency, accelerating from 11% in Q4, achieved while absorbing 56% volume growth and a 49% increase in MELI+ subscribers since Q3.
"Almost half of the decline in VC per shipment from May (immediately prior to lowering the free shipping threshold) to June (immediately after the change) has been offset by scale, productivity, efficiency and some pricing. This means that some ASP ranges below R$79 are now breakeven."
— Letter to Shareholders
Assessment: This is the strongest evidence in the print that the compression is investment rather than competitive erosion, and it is why the rating is Hold rather than Underperform. Half the unit-economics gap closed in roughly nine months, with management guiding to continued but non-linear improvement. If this curve keeps bending, the gross margin line repairs itself without any strategic change. The sequential gross margin improvement of 46bps is the first visible instalment.
3. Brazil Take-Rate Reductions Hit the P&L Next Quarter
A detail that surfaced only in Q&A and materially affects Q2 modelling: MercadoLibre lowered take rates for competitively priced sellers in selected Brazilian categories and price ranges, conditional on sellers maintaining competitive listing prices. Management framed this as a continuation of targeted reductions begun in 2024, citing 62% unique buyer growth since then, record live listings and record effective sellers.
"There is a complement those investments that Ari mentioned regarding lowering shipping take rate to merchants were implemented towards the end of Q1. So they were not -- did not flow through our P&L in Q1. They will flow through in Q2."
— Martin de Los Santos, CFO
Assessment: This is the most actionable disclosure on the call and it was volunteered rather than asked for. A revenue-reducing action already taken but not yet in the numbers means Q2 commerce revenue and gross margin carry a headwind that Q1 did not. The magnitude was never quantified. Anyone extrapolating Q1's 46bps sequential gross margin improvement into Q2 is extrapolating through a known, unsized negative.
4. The Credit Card as the Anchor Investment
Management positioned the credit card as the fintech equivalent of the logistics network build, which is a deliberately large claim. The portfolio grew 104% YoY to $6.6bn on 2.7mn cards issued in the quarter, card total payment volume grew 90%, card monthly active users grew 68%, and the card's 15-90 day non-performing rate fell 80bps YoY. Management said Brazilian cohorts after five years of investment are maturing at a pace increasingly offsetting the dilution from new cohorts, which is the specific cohort mechanic the entire investment case rests on.
"Investing in our credit card is as pivotal for Mercado Pago as the launch of our managed logistics network was for our marketplace ten years ago."
— Letter to Shareholders
The accounting drag is structural and was explained plainly: the full expected loss on the committed line is provisioned at issuance, so an accelerating issuance rate guarantees an elevated provision charge regardless of underlying credit quality.
Assessment: The cross-sell logic is sound and the improving card non-performing rate is real evidence that underwriting is holding as the book scales. The honest difficulty is that a five-year payback investment funded through the income statement is indistinguishable, quarter to quarter, from deteriorating credit. The company is asking investors to accept cohort maturation on faith for several more years while the reported margin absorbs the cost. That is a reasonable ask from this management team and still a large one.
5. Provisions: the Two-Thirds and One-Third Decomposition
Management decomposed the 390bps provision drag consistently in the letter and on the call. Two-thirds is the mechanical consequence of the credit book growing 87% against revenue growth of 49% under expected-loss accounting. The remaining third is higher provisioning on the Brazilian consumer portfolio, driven by two deliberate underwriting decisions.
"Two thirds of that compression is a function of the credit portfolio growing at a much faster pace (87% YoY) than our consolidated revenue (49% YoY). The remaining third mostly relates to higher provisions in the consumer portfolio in Brazil due to our decision to offer longer loan terms and increase the reach of personal loans."
— Letter to Shareholders
Assessment: The decomposition is credible and it reframes the charge usefully: the larger part is a growth artefact that self-corrects when origination growth decelerates, and the smaller part is a genuine underwriting decision. But note what the first part implies. Provisions only normalise when the credit book stops outgrowing revenue, and management just committed to accelerating card issuance in Mexico and Argentina. The mechanical two-thirds is not going away in 2026 by management's own plan.
6. Brazil Consumer Underwriting: Duration Extended, Credit Box Widened
The most consequential credit disclosure came in response to questioning rather than in the letter. Average Brazilian personal loan duration was extended from five months to eight, and the personal loan product was deliberately extended to previously declined segments at lower spreads.
Q: "I wanted to discuss a bit the Brazil NIM compression. You mentioned that 1/3 of the provisions came from higher provisions in Brazil. Is that part of the upmarket move, like you say you're taking longer loans."
— Marcelo Santos, JPMorgan
A: "We used to have loans typically on average of 5 months, and that has moved to 8 months. And also, we are expanding the reach on our personal loans portfolio. And that is in part what you mentioned is we are reaching out to customers who had a line of credit in the past and we were not -- they were not taking it. So we are lowering the spread to see if we entice them to start trying our personal loan products and also reaching out to segments where they are either more risky or where we have to work with smaller spreads."
— Osvaldo Giménez, President, Fintech
Management characterised the portfolio as continuing to earn double-digit margins, though less profitable than a year ago, and separately acknowledged that longer duration brings higher early-repayment risk which shortens realised interest collection.
Assessment: This is the disclosure that most deserves scrutiny and it received the least. Extending duration by 60% while simultaneously lowering spreads to reach previously declined borrowers is, by construction, a credit-box expansion. Management calls it an experiment that confirmed profitability. Eight-month loans written in Q1 will not have completed a full loss curve until Q4, so the confirmation is provisional by definition. The 90-360 day past-due bucket rising 79bps is the first place any error would appear.
7. NIMAL: Mix, Seasonality and Spread
Net interest margin after losses fell to 17.8% from 22.7%, down 490bps YoY and lower sequentially. Management attributed the sequential decline to normal Q1 seasonality and the YoY decline to two causes: a higher mix of the structurally lower-spread credit card, and Brazilian spread compression from higher provisions.
"there is some seasonality in Q1 when typically we have lower NIM in Q1, but it's -- I say it's normal to see the sequential compression that you see in the results."
— Osvaldo Giménez, President, Fintech
Management also explained the counter-intuitive point that consumer and merchant loan spreads are better than credit card spreads, because the card requires provisioning the entire committed line at issuance, producing an upfront loss that reverses only as the cohort seasons.
Assessment: The mix explanation is arithmetically sound and means NIMAL will keep falling as long as the card grows faster than the rest of the book, which management intends. NIMAL is therefore not a useful signal of credit health going forward; it is a mix indicator. The metric to watch instead is provision as a share of credit revenue, which deteriorated 920bps YoY to 61.8%.
8. Mexico Tax Reform: the One Exogenous Headwind
Mexican FX-neutral GMV growth of 28% decelerated sequentially, and management attributed this to tax reform prompting small and medium sellers to raise prices into a consumer already exercising caution. This is the only genuinely external demand headwind disclosed in the quarter, and management explicitly characterised it as short-term while reaffirming the long-term opportunity.
Assessment: A seller-driven price increase into a cautious consumer is a real demand risk and it is the one deceleration in this print that management cannot fix by spending more. Mexico held contribution margin nearly flat this quarter, which is exactly why a volume slowdown there is more concerning than a margin decline in Brazil: it threatens the segment that is currently proving growth and profitability can coexist.
9. 1P and Cross-Border Trade: Acknowledged Margin Detractors
1P grew 84.7% in dollars and 69% FX-neutral, reaching almost $5bn of trailing GMV, and cross-border trade grew 68% FX-neutral. Management stated plainly that 1P still detracts from income from operations, having reached positive direct contribution in 2024, with the remaining shortfall being corporate allocation that grows as 1P outgrows overall GMV. The strategic case rests on assortment and price competitiveness, with a cited example of Brazilian cellphone share tripling from 12% three years ago to market leadership.
Assessment: Intellectually coherent and consistent with the fulfilment playbook that worked. The uncomfortable structural point is management's own: 1P's path to operating profitability requires diluting corporate allocations, which gets harder the faster 1P grows. This is a line that becomes accretive slowly and only if growth eventually moderates, and no timeline was offered.
10. AI: Real Leverage in Product Development, Early Revenue in Search
Product development expense fell from 9.3% to 7.9% of revenue, contributing 136bps of the offset to margin compression, on headcount growth of 8% described as a carryover from 2025 hiring. Management stated an intention to keep engineer headcount broadly stable through 2026, which makes this the most durable margin offset disclosed. The company rolled out its first LLM-based marketplace search in Q1, live in Brazil, Mexico and Argentina, with reported uplifts in conversion and in sponsored-listing click-through.
"we deployed LLMs in search in commerce for the first time this quarter. And basically, that is live in Brazil, Mexico and Argentina. So now we are using this technology to better understand users' intent"
— Ariel Szarfsztejn, President, Commerce
Management flagged that third-party model costs will partially offset this leverage, framed as a deliberate choice to prioritise adoption over cost optimisation.
Assessment: This is the most credible structural margin offset in the business and the market gave it no weight. Flat engineering headcount against 49% revenue growth is a mechanical 130bps or more of annual leverage. The search deployment also lands in advertising, where better relevance raises sponsored-listing yield on the same traffic. Neither effect is large enough to counteract provisions in 2026, but both are real and both compound.
11. Competitive Intensity in Brazil
Asked directly about Amazon's recent changes in Brazil, management declined to concede any impact and pointed to its own metrics: conversion up one percentage point YoY, record market share, growing supply, accelerating items sold and improving retention. The framing offered was that intensifying competition is expanding the online market faster than it is redistributing it.
"So the pie is increasing at a faster pace than it was before, and we are taking an even larger slice of that pie."
— Ariel Szarfsztejn, President, Commerce
Assessment: The metrics support the claim: a company losing a share war does not simultaneously post record net promoter scores, accelerating buyer growth and a one-point conversion gain. The unfalsifiable part is counterfactual. Free shipping thresholds and take-rate cuts are competitive responses whether or not they are framed as offensive investments, and the timing of both alongside intensifying competition is not a coincidence a careful investor should ignore.
Guidance & Outlook
MercadoLibre issues no formal financial guidance, and none was introduced this quarter. What the company did provide was an unusual set of directional forward statements, which functioned as de facto guidance and drove the share reaction.
| Item | What management stated | Read |
|---|---|---|
| Operating margin | Does not anticipate the Q1 level changing materially in the near term | The de facto guide; caps FY26 at roughly Q1 levels |
| Brazil take rates | Reductions implemented late Q1, flow through in Q2 | Unsized Q2 revenue and gross margin headwind |
| Engineering headcount | Intend to keep broadly stable in 2026 | Durable product development leverage, partly offset by third-party AI model costs |
| Unit shipping costs | Direction of travel downward, explicitly not linear | Gross margin tailwind, with expectations managed down |
| Credit card issuance | Increasing pace in Mexico, early stage in Argentina | Sustains the mechanical provision drag through 2026 |
| Payroll loans in Brazil | Integrated with government, product to launch soon | New lower-risk secured product; no size or timing given |
| Energy and labour costs | No Q1 impact; some Q2 logistics cost pass-through, mostly passed to consumers | Monitored, not currently material |
Street positioning after the print: Post-print consensus carries FY26 EPS of $47.84 on revenue of $38.8B, with Q2 at $10.48 and $9.28B.
Implied ramp required: FY26 EPS of $47.84 on 50.6972M shares implies net income of roughly $2,425M, a 6.25% net margin. Q1 delivered $417M at 4.71%. The residual nine months therefore need about $2,008M on roughly $29,955M of revenue, a 6.70% net margin. That is a 199bps net-margin recovery over the balance of the year, in a year in which management explicitly said the margin dial is not moving.
Guidance style: MercadoLibre has never guided and has historically refused forward margin commentary entirely. That it broke that pattern to deliver a negative directional statement is itself the signal. Managements that expect to beat do not pre-commit the market to a depressed level.
Analyst Q&A Highlights
What Changed Between February and May
The first question of the call went straight to the sentence that moved the stock, asking what new opportunities had been identified since the Q4 report to justify a deeper investment posture, and whether Q1's margin was the right level to assume for the rest of 2026. The response reframed the premise: nothing changed, because the investment philosophy and the specific initiatives are the same ones funded for several quarters. What changed was the strength of the evidence supporting them.
Q: "could you just help us understand what perhaps changed from the fourth quarter that you reported in late February to now, after the first quarter? What new opportunities have you identified to invest behind? And whether this sort of this margin level that we saw in the first quarter, again, I know you don't provide guidance, but is that roughly the right level that we should think about for the remainder of 2026?"
— Irma Sgarz, Goldman Sachs
A: "I think the investment philosophy hasn't changed. In fact, we decided to invest behind initiatives -- similar initiatives that we have been investing over the past several quarters. What we have seen, and we tried to be very explicit about that on the letter this quarter, as a result of those investments. We're seeing very good results in terms of our credit card portfolio."
— Martin de Los Santos, CFO
Assessment: The answer addressed the first half of the question and left the second half untouched. Whether 6.9% is the right level to model for 2026 was asked directly, in plain terms, by the first questioner, and was not answered here or in the two subsequent attempts. The letter's language was as specific as management was willing to be, and the refusal to confirm it verbally is itself informative.
What Could Move the Dial in Either Direction
A later question isolated the unusual nature of the forward statement and asked what would cause the level to change. The answer confirmed the framing was intentional, restated that margin is an output of investment intensity rather than a target, and then volunteered the symmetric point that improving margins in the short term would be straightforward if management chose it.
Q: "It's unusual for you to give that near-term clarity. What has prompted that? And then what could cause it to change? What unforeseen circumstance could cause margins to be lower or to be higher in the near term?"
— Geoffrey Elliott, Autonomous
A: "And then the flip side would be if we wanted to improve margins in the short term, it will be fairly easy for us to slow down certain investments, but we don't think it's the right way to go given the large opportunity that we have in front of us for both commerce and fintech."
— Martin de Los Santos, CFO
Assessment: The most important exchange on the call for anyone assessing downside. Management asserts the compression is fully discretionary and reversible at will, which if true means the margin floor is a policy choice rather than a competitive outcome. That materially limits the bear case, and it is also precisely the claim that cannot be verified until the company chooses to demonstrate it. The credit provision component, roughly two-thirds of the drag, would in any case take several quarters of slower origination to unwind.
Whether Q1's Margin Is the Run-Rate for the Year
A third attempt at the same question tried a different angle, noting that Q1 carries seasonal credit headwinds that ease through the year, and asking whether incremental investment was now masking that seasonal recovery such that full-year margin should be modelled at Q1 levels. Management declined the frame, restated the investment logic, and offered a qualitative endorsement of the Q1 level without confirming it as a run-rate.
Q: "There are obviously seasonal headwinds in 1Q from credit business that partly eases off through the year. But are you now ramping investments in certain areas that's incremental that the seasonal effects are somewhat masked and we should expect EBIT margin for the year to be around 1Q levels?"
— Deepak Mathivanan, Cantor Fitzgerald
A: "as you know, we rather not talk about margins on a quarterly basis and forward-looking. But let me say that in Q1, we decided to invest, we said this -- the level of intensity of our investment based on the performance of the different investments. [...] Again, we're not managing the business to a particular margin level. We feel very optimistic and comfortable about the level that we delivered this quarter"
— Martin de Los Santos, CFO
Assessment: Three analysts asked the same question three ways and none received a numerical answer. The phrase describing management as comfortable with the level delivered this quarter is the closest thing to confirmation offered, and combined with the letter it is enough to model 2026 at approximately Q1 margins. The refusal to say so directly while writing it in the letter suggests management wanted the statement on the record but not amplified.
Cost of Risk at Roughly 37% and Whether It Is the Going Rate
The sharpest credit question of the call computed annualised cost of risk independently, put the figure to management, and asked whether it was a one-off or the sustainable rate as the credit business scales. The answer restated the two-thirds and one-third decomposition and asserted that credit remains accretive to consolidated margin, without addressing the forward rate.
Q: "the cost of risk increased quite substantially this quarter. It's now around 37% as per my calculation. Could you zoom in on which particular loan segment or any particular region that might have led to this increase in cost of risk? And do you see it as a one-off? Or do you see closer to 37%, 38% as a going rate for cost of risk as you continue growing the credit business?"
— Neha Agarwala, HSBC
A: "The fact that our credit book grows at a faster pace than revenues, the credit book grows at 87% year-on-year and our revenues for MercadoLibre growth at 49%, that generates margin compression. And the reason for that is because as we issue any new loan, we have to provision for the full amount of the expected loss of that loan. And when we accelerate growth, we need to provision more. So 2/3 of the margin compression comes from that. That's natural."
— Martin de Los Santos, CFO
Assessment: A precise question that received an explanation of the past rather than a view on the future. The one-off versus going-rate distinction was the entire point and went unanswered. Management's own logic supplies the implied answer: if the charge scales with origination growth, and issuance is accelerating in two new markets, then roughly 37% is closer to a going rate than a one-off for as long as the book compounds near 87%.
Confidence Behind the Credit Box Expansion
A separate line of questioning probed the timing of the decision to lengthen duration and widen the credit box in Brazil, given the macro backdrop, asking specifically how the risk and reward were balanced. The response separated the card, where repayment behaviour was described as unchanged, from personal loans, where the expansion was explicitly deliberate.
Q: "I wanted to dig in a little bit further on the decision to both go longer duration and to expand the credit box when it comes to the credit card in Brazil and personal loans in Brazil. [...] I just want to make sure I understand how you balance the risk/reward here."
— Craig Maurer, FT Partners
A: "given that we are growing well and it's profitable, we wanted to reach out to segments where we believe we can make money even if the -- on the margin, the spread we make is smaller than with the segments we were already serving. And then, again, those segments which sometimes are -- we're not willing to take a credit before to start taking them."
— Osvaldo Giménez, President, Fintech
Assessment: A candid answer that concedes more than the letter does. Lending to segments previously declined, at lower spreads, on longer duration is a credit-box expansion described accurately and without euphemism. The candour is a governance positive. The substance still means the loss curve on the fastest-growing consumer vintage will not be observable until late in the year, and the reassurance offered rests on a metric that cannot yet have registered the change.
The Next Step Down in Unit Shipping Costs
On the commerce side, questioning focused on the durability of the 17% unit-cost reduction and what drives the next leg. The answer disaggregated the improvement into three mechanisms and, unusually for this call, actively managed expectations downward on the pace of further gains.
Q: "It sounds like unit costs were down, I believe, 17% year-over-year. I assume a lot of this has to do with better utilization of idle capacity. As we look later into this year and next, how do you kind of think about maybe the next step down in terms of unit costs?"
— Josh Beck, Raymond James
A: "looking forward, I would say that we expect the direction of travel in unit shipping costs to continue to be downwards, but this is not going to be linear, right? So we will be adding more capacity for sure, given the growth rates that we are having -- and clearly, some of the incremental gains may take longer time to achieve, but we are confident on the trajectory that we'll continue to see."
— Ariel Szarfsztejn, President, Commerce
Assessment: The most operationally detailed answer of the call, and the only place management actively lowered expectations. Volume density, the slow-shipping network absorbing idle capacity, and route productivity are three distinct mechanisms with different durability, and the explicit non-linearity warning is a signal that the easy gains are behind. The gross margin recovery will be slower than the 17% print implies.
Take-Rate Investment: Sellers Versus Buyers
Questioning on the Brazilian seller promotions sought to understand the decision and how investment dollars are allocated between seller incentives and buyer-facing initiatives. The answer characterised the reductions as targeted rather than platform-wide, conditional on seller pricing behaviour, and framed as passing through to buyers the value created by prior buyer-side investment.
Q: "One other item I'd like to understand more about was the Brazil seller promotions you announced in recent months, lowering take rates for competitively priced sellers in certain price points. So I'd like to understand what drove the decision? And when you're thinking about allocating price investment dollars, how you balance that between seller investments versus buyer initiatives such as free shipping?"
— Andrew Ruben, Morgan Stanley
A: "we lowered take rates in some categories, in some specific price ranges. So this is not a platform-wide take rate cut. It's a targeted investments where we see the greatest opportunity. And by opportunity, I mean both elasticity of demand and elasticity of supply. And basically, all the discounts that we are providing to merchants are conditional on sellers maintaining a competitive pricing on their listings in Mercado Libre."
— Ariel Szarfsztejn, President, Commerce
Assessment: The conditionality is the sophisticated part and is genuinely differentiated: a take-rate discount contingent on competitive listing prices converts a margin giveback into an enforced price-competitiveness mechanism. The disclosure gap is size. A targeted, conditional, unquantified take-rate reduction landing in Q2 is a modelling problem no analyst can solve from outside, and the follow-up quantifying it never came.
What They're NOT Saying
- Any timeline for margin recovery. Three analysts asked in three different ways whether Q1's margin is the 2026 run-rate. None received a numerical answer. Management confirmed the direction and refused the level, which is the maximum ambiguity consistent with the letter's own language.
- The size of the Brazil take-rate reduction. Management volunteered that it lands in Q2 rather than Q1, which was helpful, and never quantified it, which was not. A known revenue headwind of unknown magnitude sits in the very next quarter.
- Advertising revenue in dollars. Growth of 63% FX-neutral and 73% in dollars was disclosed for the fastest-growing high-margin line in the business, with no absolute figure and no segment disclosure that isolates it. Advertising is the most credible structural offset to gross margin compression, and its size cannot be determined from outside the company.
- The magnitude of the 1P and cross-border drag. 1P was described as still detracting from income from operations and cross-border as expected to be profitable over time. Neither carries a number, so the portion of the 595bps attributable to deliberately loss-making growth lines cannot be separated from the portion attributable to the core.
- The off-balance-sheet credit card commitments. Unused agreed card commitments grew 32.1% in one quarter to $11,892M. The figure appears in the filings, was absent from the shareholder letter, and was not raised by any of the thirteen analysts. Because expected losses on committed lines are provisioned at issuance, this balance is the clearest forward indicator of the provision charge.
- The 90-360 day past-due bucket. Management cited the stable 15-90 day non-performing ratio repeatedly across the letter and the call. The 90-360 day bucket, which is where losses actually crystallise, rose 79bps to 17.6% of the portfolio and was never mentioned. Both figures are in the same filed table.
- Negative adjusted free cash flow. Adjusted free cash flow of negative $56M against positive $58M a year ago appeared in the letter, framed as seasonal, and was not raised on the call by management or by any analyst. Net debt rose 22.8% in a single quarter to $5,748M.
- Sales and marketing headcount. The letter attributes sales and marketing deleverage to scaling the affiliate channel. The filings add that sales and marketing headcount rose 43% YoY, alongside a $250M increase in marketing expense and $75M in buyer protection. The headcount component of a supposedly variable, performance-based channel is a fixed cost that was not part of the narrative.
- Any capital return. No buyback occurred in a quarter generating $2,075M of operating cash flow, and share count is unchanged. Capital allocation was never discussed on the call. With the credit book absorbing $1,949M of net funding, the answer is implicit, but it was neither stated nor asked.
Market Reaction
- Pre-print setup: Closed at $1,870.01 on May 7 ahead of the after-market release. Down 7.2% year to date against an S&P 500 up 7.2%, down 17.3% over trailing twelve months, but up 7.3% over the prior thirty days. The 52-week closing range entering the print was $1,599.52 to $2,613.63, so the shares came in 28.4% below their 52-week closing high and 16.9% above the low.
- After-hours move: Down 7.2% on the print evening, with the release at approximately 4:05 p.m. ET and the call at 5:00 p.m. ET.
- Next-day session: Opened at $1,685.00, a 9.9% gap down, traded a range of $1,623.00 to $1,704.19, and closed at $1,632.52, down 12.7% or $237.49. The close sat 2.1% above the pre-print 52-week closing low.
- Volume: 2.3M shares against a 30-day average of 0.4M, roughly 6.0 times normal.
- Market context: The S&P 500 rose 0.8% on the reaction session, so essentially the entire move was company-specific.
The shape of the move is as informative as its size. A 7.2% after-hours decline on the print widened to a 9.9% gap at the open and then a 12.7% close. The market kept selling after the call rather than stabilising, which is the signature of a reaction driven by forward language rather than by reported numbers. The numbers were available at 4:05 p.m.; the elaboration of what they implied for the rest of the year came during the 5:00 p.m. call and in the letter's closing section.
What repriced was the earnings base, not the growth rate. A 49% revenue beat with accelerating volume in every major market is not a print that produces a 12.7% decline on its own. The decline is what happens when a market carrying FY26 EPS near $47.84 is told the margin structure that number depends on is not returning. On the pre-print close the shares traded at 49.4x trailing twelve-month earnings of $37.89; at the reaction close, 43.1x. Neither multiple is distressed, which is the reason the selling could continue: the de-rating removed the premium for margin recovery without yet pricing in the estimate cuts that management's language implies.
The setup made it worse. Shares had rallied 7.3% over the preceding thirty days into a print with no guidance and a wide dispersion of consensus estimates. Positioning was constructive on a name that had underperformed for twelve months, which is the configuration in which forward-looking disappointment produces gap risk rather than orderly repricing.
Street Perspective
Debate: Is This Moat-Deepening Investment or a Competitive Response Being Narrated as Choice?
Bull view: The evidence for genuine investment is unusually concrete. Items sold in Brazil more than doubled their growth rate in nine months, unit shipping costs are falling 17% and accelerating, buyer growth is the fastest in five years, and net promoter scores are at records against both retail and banking incumbents. A company losing ground does not produce those metrics simultaneously. Management stated it could improve margins quickly by slowing investment and is choosing not to, which frames the compression as a policy setting rather than a competitive tax.
Bear view: Every one of those investments is also exactly what a market leader does when a well-capitalised competitor intensifies in its most important market. Free shipping thresholds, conditional take-rate cuts, affiliate expansion and 1P price competitiveness are defensive instruments regardless of framing. The claim that spending is discretionary is unfalsifiable until it is tested, and the test would require accepting slower growth.
Our take: The bull case is stronger on evidence and the bear case is stronger on epistemics. The engagement metrics are real, independently disclosed and hard to fake, and the cohort behaviour of post-threshold buyers is the kind of detail a struggling business does not volunteer. But the distinction largely does not matter for a 2026 investor: whether the spending is offensive or defensive, it is happening, management has said it continues, and the earnings base resets either way. The distinction matters enormously for a 2028 investor, and there the evidence favours the bulls.
Debate: Is the Credit Book a Compounding Engine or an Under-Reserved Cycle?
Bull view: A 59.5% annualised gross yield against a 36.8% cost of risk is a wide spread on a book compounding at 87%. The 15-90 day non-performing ratio is broadly stable, card non-performing rates improved 80bps, renegotiated loans declined as a share of the portfolio, and reliance on gains from selling receivables fell. Provisions are front-loaded by expected-loss accounting, so an accelerating book mechanically overstates the run-rate charge; growth deceleration alone would produce a sharp margin recovery.
Bear view: Total past-due loans rose 142bps in one quarter to 27.8% of the portfolio, allowance coverage rose to 26.2%, write-offs doubled, and the 90-360 day bucket deteriorated faster than the bucket management chose to discuss. Duration on Brazilian consumer loans was extended 60% while the credit box was widened to previously declined borrowers at lower spreads, and no vintage written under those terms has completed a loss curve. Off-balance-sheet card commitments grew 32% in a quarter.
Our take: The bull mechanics are correct and the bear timing concern is also correct, and they are not in conflict. Expected-loss accounting genuinely does front-load the charge, so a decelerating book would produce real margin relief. The difficulty is that management has committed to accelerating issuance in Mexico and Argentina, so deceleration is not the plan. The right posture is to treat roughly 37% cost of risk as the going rate while the book compounds near 87%, and to watch the 90-360 day bucket rather than the 15-90 day metric management prefers, because the eight-month Brazilian vintages will not season until late in the year.
Debate: Does the Multiple Already Reflect the Reset?
Bull view: A 12.7% single-session decline took the shares to 2.1% above their 52-week closing low, 43.1x trailing earnings and roughly 34x the FY26 consensus, on a business growing revenue 49% with accelerating volumes and a stated intention to hold engineering headcount flat. Enterprise value of $88.5B against consensus FY26 revenue of $38.8B is 2.28 times sales for the dominant platform in a region where online purchase frequency is roughly a sixth of the United States level.
Bear view: The forward multiple is built on an estimate that management just contradicted. Consensus FY26 EPS requires a 199bps net-margin recovery across the balance of the year; hold Q1 flat and the number is closer to $36, putting the shares at 45x rather than 34x. On that basis the stock is more expensive after a 12.7% decline than it appeared before, and the estimate revisions have not happened yet.
Our take: The bear case is arithmetically unanswerable on the forward multiple and the bull case is right about the asset. The decisive point is sequencing. Consensus has not yet been marked to management's own language, so the multiple investors are quoting is not the multiple they are paying. Until FY26 estimates come down toward the high $30s, the apparent de-rating is an illusion produced by a stale denominator. That is a reason to wait rather than a reason to sell, because the franchise quality and the volume acceleration are genuine and the eventual entry point is likely to be at a similar price on a lower, credible number.
Model Update Needed
This is an initiation, so the comparison below is against post-print Street positioning rather than a prior published model of our own.
| Item | Street (post-print) | Our estimate | Reason |
|---|---|---|---|
| FY26 revenue | $38.8B | $38.5B to $39.5B | Q1 beat by 4.9% and volume is accelerating; Q2 carries an unsized take-rate headwind. Revenue is not the disputed variable. |
| FY26 operating margin | Implied ~9% | 7.0% to 7.5% | Management stated the Q1 level does not change materially in the near term. Modest H2 relief from shipping unit costs and product development leverage only. |
| FY26 net margin | 6.25% | 4.9% to 5.3% | Requires no margin recovery beyond gross margin drift; 2025's pattern had Q1 above, not below, the balance of the year. |
| FY26 EPS | $47.84 | $37 to $41 | The central disagreement. Roughly 15% to 23% below consensus. |
| Provision for doubtful accounts | Implied normalising | 13.5% to 14.5% of revenue | Cost of risk near 37% persists while the book compounds near 87% and issuance accelerates in two new markets. |
| Gross margin | n/a | 43.5% to 44.5% | Sequential improvement of 46bps is the trend; 1P mix and the Q2 take-rate reduction cap the recovery. |
| Product and technology development | n/a | 7.5% to 8.0% of revenue | Flat engineering headcount against high-40s revenue growth, partly offset by third-party AI model costs. |
| Capital expenditure | n/a | 3.0% to 3.5% of revenue | Intensity fell 120bps YoY; this cycle runs through the income statement, not the balance sheet. |
| Share count | 50.70M | 50.70M | No buyback while the credit book absorbs roughly $2B per quarter of net funding. |
Valuation impact: At the reaction close of $1,632.52 the shares carry a market capitalisation of $82.8B and an enterprise value of $88.5B against net debt of $5,748M. On our FY26 EPS range of $37 to $41 the stock trades at 40x to 44x, against 43.1x trailing twelve-month earnings of $37.89. That is a market multiple for a market-level earnings outcome in 2026, with the growth optionality unpriced and the estimate risk unresolved. We see no valuation support for a constructive stance until consensus migrates toward the high $30s, and no valuation case for a negative stance given a 49% revenue growth rate, a strengthening competitive position and a management team that has demonstrated it can reverse the spending if it chooses.
Thesis Scorecard Post-Earnings
This is initiation coverage, so the pillars below are established here rather than graded against a prior standing thesis. Status tags set the baseline that subsequent quarters will be scored against.
| Thesis Point | Status | Notes |
|---|---|---|
| Bull #1: Demand elasticity is real and the ecosystem flywheel compounds | Confirmed | Brazil items sold +56% and accelerating for four straight quarters, unique buyers +32%, items per buyer +17.0%, record net promoter scores. New cohorts buy more across more categories with higher retention. |
| Bull #2: The unit-cost curve offsets the shipping subsidy over time | Confirmed | Brazil unit shipping cost -17% YoY, accelerating from -11%, while absorbing 56% volume growth. Some sub-R$79 price bands now at breakeven. Gross margin rose 46bps sequentially. |
| Bull #3: Fintech cross-sell converts marketplace users into banking relationships | Confirmed | Fintech monthly active users 83mn (+29.7%), AUM near $20bn (+77%), card portfolio +104% to $6.6bn, card TPV +90%, card 15-90 day NPL improved 80bps. |
| Bull #4: Corporate and technology cost leverage is structural | Confirmed | Unallocated costs +15.1% against revenue +49.0% (264bps leverage); product development from 9.3% to 7.9% of revenue with engineering headcount to stay flat in 2026. |
| Bear #1: Margin compression has no stated recovery path | Confirmed | Operating margin 6.9%, down 595bps. Management pre-committed to holding the level and declined to answer the run-rate question from three separate analysts. |
| Bear #2: Credit-book risk is building faster than the disclosed metrics suggest | Confirmed | Cost of risk ~37%, allowance coverage 26.2%, total past due 27.8% (+142bps QoQ), write-offs +99%, Brazilian consumer duration extended from 5 to 8 months into a widened credit box. |
| Bear #3: Consensus estimates are incompatible with management's own language | Confirmed | FY26 consensus EPS requires a 199bps net-margin recovery over the balance of the year. At a flat Q1 margin, FY26 EPS is roughly $36 against $47.84. |
| Bear #4: Competitive intensity in Brazil is rising | Neutral | Management denied any impact and the metrics support that, but the investment programme is consistent with a defensive posture and the counterfactual is unobservable. |
Overall: The operating thesis is stronger than the earnings thesis. Every bull pillar was confirmed with specific, verifiable evidence, and the business is demonstrably widening its lead in its largest markets. Every bear pillar was also confirmed, and the binding one is not operational: it is that the published earnings expectations for 2026 no longer describe the company management has just described. The gap between an improving franchise and a falling earnings base is the entire investment question for the next twelve months.
Action: Initiate at Hold. The business merits a premium multiple and is not currently being valued on a credible number. Wait for FY26 consensus to migrate toward the high $30s, which would put a similar share price on an honest multiple, and revisit toward Outperform if the 90-360 day past-due bucket stabilises and gross margin continues its sequential recovery. The signposts that would change the rating are specific: two consecutive quarters of sequential gross margin expansion, a decelerating rate of credit-book growth relative to revenue, or any quantified framework for the take-rate reduction and advertising contribution.