Shopee Prints a Record and Monee Gets a Listing Window, but a Turn in Credit and an Unpriced AI Build Keep Sea's Discount Wide
Key Takeaways
- Revenue beat cleanly, and whether profit missed depends on which definition the Street used. Revenue of $8,582.3M grew 43.4%, about 6.2% above the roughly $8.08B consensus. Adjusted EBITDA of $931.3M grew 28.9% on the new definition that excludes Garena's deferred-revenue effect, 5.9% below the roughly $990M in Bloomberg's consensus compilation, as reported. On the old definition the quarter would have been $983.3M, within 0.7% of that figure. Net income rose 34.0% to $502.5M, helped by a $65.2M investment gain and held back by a 35.6% tax rate.
- Shopee cleared the bar we set for it. Adjusted EBITDA of $269.2M was up 44.7% and was the most profitable quarter Shopee has reported. At 0.68% of GMV it is now above the 0.58% of a year ago, which removes the reason we downgraded the margin pillar in August. Management reaffirmed the $1.0B full-year target, which now needs only $252.2M in the fourth quarter. The reservation is that the margin was flat sequentially while ad growth slowed, and management said part of future monetisation gains will be reinvested.
- Monee's credit turned in the same quarter that Sea gave it a listing window. The 90-day NPL ratio rose to 1.3% from 1.0% in Q2 and 1.1% a year ago. Management attributed this to an evolving credit environment in Thailand and Brazil and tightened underwriting for selected segments. Provisions rose 82.3%, Monee's adjusted EBITDA margin fell to 18.4% from 26.1%, and allowance coverage of gross loans reached 11.5%. Separately, Sea now hopes to list Monee within one to three years. That is the first concrete path to a market value for a business we estimate earns about $1.2B of annual EBITDA.
- Sea is paying for its AI build-out before it earns any revenue from it. SeaCloud has no enterprise contracts, and management could not yet name its target segment, customers or business model. Other Services lost $83.9M of adjusted EBITDA against $25.1M a year ago. Purchases of property and equipment reached about $547M in Q3, against $356M in Q2 and $139M in Q1. Management says SeaCloud will not significantly affect the financials "for now". The capex line shows the spending has already started.
- Rating: Maintaining Outperform, trimming conviction, and cutting fair value to $95 to $105 from $150 to $165. Our Outperform at $118.32 has been wrong by 35% in eleven weeks, and we cut FY27 adjusted EBITDA by about 20% to reflect the cost of that error honestly. At $76.80 the stock trades on 9.4 times our lower FY27 estimate, with $10.1B of net cash and investments equal to about a fifth of its market value. The new range implies 23.7% to 36.7% upside.
SE financial model
Income Statement · Dollars in millions, except per share
| Income Statement | ||||||
|---|---|---|---|---|---|---|
| Actual | Estimate | |||||
| FY2023 | FY2024 | FY2025 | FY26E | FY27E | FY28E | |
| Service Revenue | $11,942.4 | $15,261.3 | $20,913.1 | $27,673.3 | $30,746.1 | $36,251.2 |
| Sales of Goods | 1,121.2 | 1,558.6 | 2,025.4 | 2,707.9 | 2,856.2 | 3,238.1 |
| Total Revenue | $13,063.6 | $16,819.9 | $22,938.5 | $30,381.2 | $33,602.3 | $39,489.3 |
| Less: Cost of Service | ($6,202.5) | ($8,164.4) | ($10,812.0) | ($14,261.9) | ($15,623.7) | ($18,125.6) |
| Less: Cost of Goods Sold | (1,027.4) | (1,450.4) | (1,882.7) | (2,563.6) | (2,659.3) | (2,979.1) |
| Total Cost of Revenue | ($7,229.9) | ($9,614.8) | ($12,694.7) | ($16,825.6) | ($18,282.9) | ($21,104.6) |
| Gross Profit | $5,833.6 | $7,205.1 | $10,243.7 | $13,555.6 | $15,319.3 | $18,384.6 |
| Less: Sales and Marketing | ($2,779.2) | ($3,472.7) | ($4,492.2) | ($5,838.1) | ($6,128.6) | ($6,910.6) |
| Less: General and Administrative | (1,134.7) | (1,267.7) | (1,357.8) | (1,684.4) | (1,770.8) | (1,974.5) |
| Less: Research and Development | (1,164.1) | (1,206.0) | (1,157.1) | (1,196.2) | (1,199.6) | (1,303.1) |
| Less: Provision for Credit Losses | (633.9) | (776.9) | (1,372.6) | (2,079.0) | (2,189.7) | (2,487.8) |
| Other Operating Income/(Expense), Net | 103.1 | 180.4 | 121.3 | 120.7 | 134.4 | 158.0 |
| Total Operating Expenses | ($5,608.9) | ($6,542.9) | ($8,258.4) | ($10,677.1) | ($11,154.3) | ($12,518.1) |
| Operating Income (Loss) | $224.8 | $662.2 | $1,985.3 | $2,878.5 | $4,165.1 | $5,866.5 |
| Interest Income | $331.3 | $365.8 | $331.1 | $338.4 | $327.0 | $355.4 |
| Interest Expense | (41.1) | (38.3) | (33.6) | (11.0) | (13.4) | (15.8) |
| Other Income/(Expense), Net (incl. equity investees) | (89.7) | (220.6) | (20.8) | (29.6) | (33.6) | (39.5) |
| Pretax Income (Loss) | $425.4 | $769.0 | $2,262.0 | $3,176.4 | $4,445.1 | $6,166.6 |
| Less: Income Tax (Expense)/Benefit | ($262.7) | ($321.2) | ($651.1) | ($1,047.0) | ($1,431.9) | ($1,911.7) |
| Net Income (Loss) | $162.7 | $447.8 | $1,610.9 | $2,129.4 | $3,013.2 | $4,255.0 |
| Less: NI Attributable to Non-controlling Interest | $12.0 | $3.5 | $32.7 | $44.1 | $60.3 | $85.1 |
| Net Income to Sea Limited | $150.7 | $444.3 | $1,578.1 | $2,085.3 | $2,952.9 | $4,169.9 |
| EPS — Basic | $0.27 | $0.77 | $2.65 | $3.38 | $4.70 | $6.50 |
| EPS — Diluted | $0.25 | $0.74 | $2.52 | $3.25 | $4.52 | $6.25 |
| Shares — Basic (M) | 566.6 | 575.0 | 595.0 | 616.2 | 628.6 | 641.3 |
| Shares — Diluted (M) | 594.4 | 604.7 | 638.2 | 641.0 | 653.9 | 667.1 |
| Ratios & Assumptions | ||||||
| Gross Margin % | 44.7% | 42.8% | 44.7% | 44.6% | 45.6% | 46.6% |
| Operating Margin % | 1.7% | 3.9% | 8.7% | 9.5% | 12.4% | 14.9% |
| Net Margin % (consolidated) | 1.2% | 2.7% | 7.0% | 7.0% | 9.0% | 10.8% |
| Effective Tax Rate | 61.8% | 41.8% | 28.8% | 33.0% | 32.2% | 31.0% |
The full workbook adds 16 historical and 11 projected quarters, plus KPI Drivers · Balance Sheet · Cash Flow Statement · Segment KPIs (as reported, memo) — every subtotal a live formula, every projection traced to a driver.
Results vs. Consensus
Q3 2026 Scorecard
| Metric | Q3 2026 | Consensus | Beat/Miss | Magnitude |
|---|---|---|---|---|
| Total revenue | $8,582.3M | ~$8,080M | Beat | +$502M / +6.2% |
| Gross profit | $3,935.4M | n/a | n/a | +51.4% YoY |
| Operating income | $664.2M | n/a | n/a | +39.5% YoY |
| Net income | $502.5M | n/a | n/a | +34.0% YoY |
| EPS (GAAP diluted) | $0.78 | ~$0.99 (normalised basis) | Miss | -$0.21 / -21.2% |
| Total adjusted EBITDA (new basis) | $931.3M | ~$990M (Bloomberg, as reported; basis not stated) | Miss | -$58.7M / -5.9% |
| Total adjusted EBITDA (old basis) | $983.3M | ~$990M (Bloomberg, as reported) | In line | -$6.7M / -0.7% |
| Shopee adjusted EBITDA | $269.2M | n/a | n/a | +44.7% YoY |
| Shopee GMV | $39.7B | n/a | n/a | +23.3% YoY |
Year-over-Year Comparison
| Metric ($M unless noted) | Q3 2026 | Q3 2025 | YoY |
|---|---|---|---|
| Total revenue | 8,582.3 | 5,986.0 | +43.4% |
| Gross profit | 3,935.4 | 2,599.4 | +51.4% |
| Gross margin | 45.86% | 43.42% | +243bps |
| Sales and marketing | 1,886.7 | 1,150.3 | +64.0% |
| General and administrative | 425.2 | 340.0 | +25.1% |
| Provision for credit losses | 681.5 | 373.8 | +82.3% |
| Research and development | 301.7 | 286.3 | +5.4% |
| Operating income | 664.2 | 475.9 | +39.5% |
| Operating margin | 7.74% | 7.95% | -21bps |
| Non-operating income, net | 123.3 | 61.2 | +101.6% |
| Income tax expense | 280.4 | 161.1 | +74.0% |
| Effective tax rate | 35.61% | 29.99% | +561bps |
| Net income | 502.5 | 375.0 | +34.0% |
| Net margin | 5.85% | 6.26% | -41bps |
| EPS (GAAP diluted) | $0.78 | $0.59 | +32.2% |
| Total adjusted EBITDA (new basis) | 931.3 | 722.3 | +28.9% |
| Total adjusted EBITDA (old basis) | 983.3 | 874.3 | +12.5% |
The nine-month picture is consistent with the quarter. Revenue was $23,467.6M against $16,086.6M (up 45.9%) and operating income was $1,907.5M against $1,420.1M (up 34.3%). Net income was $1,398.8M against $1,200.0M (up 16.6%). The effective tax rate for the nine months was 34.5% against 26.7%. Pre-tax income rose 30.5%, to $2,158.1M from $1,653.7M. As in the first half, revenue and operating income grow at more than twice the rate of net income, and most of that gap opens at the tax line.
Quarter-over-Quarter Comparison
| Metric | Q3 2026 | Q2 2026 | QoQ |
|---|---|---|---|
| Total revenue | $8,582.3M | $7,787.8M | +10.2% |
| Gross profit | $3,935.4M | $3,549.9M | +10.9% |
| Operating income | $664.2M | $650.3M | +2.1% |
| Net income | $502.5M | $458.1M | +9.7% |
| EPS (GAAP diluted) | $0.78 | $0.70 | +11.4% |
| Total adjusted EBITDA (new basis) | $931.3M | $896.9M | +3.8% |
| Shopee GMV | $39.7B | $38.3B | +3.7% |
| Shopee adjusted EBITDA | $269.2M | $255.4M | +5.4% |
| Monee revenue | $1,621.9M | $1,402.8M | +15.6% |
| Monee loan book | $12.6B | $11.1B | +13.5% |
| Monee adjusted EBITDA | $298.3M | $288.0M | +3.6% |
| Monee 90-day NPL ratio | 1.3% | 1.0% | +30bps |
| Garena bookings | $890.0M | $763.5M | +16.6% |
| Garena adjusted EBITDA (new basis) | $458.2M | $409.5M | +11.9% |
| Other Services adjusted EBITDA | ($83.9M) | ($46.2M) | Loss widened |
Q2 figures on the new basis subtract the $20.3M Q2 deferred-revenue effect from the reported $917.2M consolidated and $429.8M Garena totals. The sequential shape shows where the quarter's incremental revenue went. Revenue added $794.5M and gross profit added $385.5M, but operating income added only $13.9M. The gap is the quarter's three growth costs: sales and marketing, credit provisions, and the widening loss in Other Services.
Revenue
Revenue of $8,582.3M grew 43.4%, a deceleration from Q2's 48.1% and Q1's 46.6%. Service revenue rose 43.3% to $7,833.7M and sales of goods rose 44.6% to $748.6M. By segment, Shopee contributed $6,070.2M (up 41.3%), Monee $1,621.9M (up 63.9%), Garena $840.1M (up 28.6%) and Other Services $50.1M (up 3.5%). Monee is now 18.9% of group revenue against 16.5% a year ago, and its growth rate accelerated from 58.9% in Q2 even as its loan book growth slowed slightly to 60.0% from 62.5%.
Garena's revenue line needs a caution. GAAP revenue grew 28.6% while bookings grew only 5.9%, because revenue deferred out of last year's exceptional third quarter is now being recognised. The change in Garena's deferred revenue was $49.9M this quarter against $187.6M a year ago. This timing effect flatters Garena's reported revenue and its new-basis EBITDA in a quarter when bookings are slowing. We return to it in the segment section.
Margins
Gross margin was 45.9%, up 243 basis points, and this is the best part of the quarter. Shopee's gross margin rose to 31.2% from 29.5% as core marketplace revenue outgrew logistics cost. Garena's cost of revenue grew only 7.0% against revenue up 28.6%. Operating margin nonetheless fell 21 basis points to 7.7%. Consolidated sales and marketing rose to 22.0% of revenue from 19.2%, and the provision for credit losses rose to 7.9% of revenue from 6.2%.
Overhead discipline held again. General and administrative expense grew 25.1% and research and development grew 5.4%. Share-based compensation fell to $132.1M from $154.2M, which cut unallocated corporate expense by 13.7%. The margin pressure therefore comes from customer acquisition and credit costs, plus the new spending in Other Services, and not from overhead.
EPS and Below-the-Line
GAAP diluted EPS was $0.78 against $0.59, up 32.2%, and basic EPS was $0.81 against $0.63. Two below-the-line items pulled in opposite directions. Non-operating income doubled to $123.3M, including a $65.2M investment gain and $62.2M of interest income. Interest income is falling: it was $197.9M for the nine months against $260.9M, down 24.1%, as cash funded buybacks, the convertible-note redemption and the loan book. Income tax expense rose 74.0% against pre-tax income up 46.6%, taking the effective rate to 35.6% from 30.0%.
The tax rate was the open question we carried out of Q2, and the quarter answered it in the direction we feared. Two consecutive quarters near 35%, and a nine-month rate of 34.5%, look like a new structural level rather than a discrete charge. We said in August that confirming this would permanently lower our earnings estimates without changing the operating story. That is what we do in the model section below.
Segment Performance
| Segment | Revenue | YoY | Segment op. income | YoY | Adj. EBITDA (new basis) | YoY |
|---|---|---|---|---|---|---|
| Shopee | $6,070.2M | +41.3% | $155.4M | +41.1% | $269.2M | +44.7% |
| Monee | $1,621.9M | +63.9% | $288.5M | +16.2% | $298.3M | +15.5% |
| Garena | $840.1M | +28.6% | $453.2M | +46.3% | $458.2M | +45.9% |
| Other Services | $50.1M | +3.5% | ($90.4M) | Loss 3.3x | ($83.9M) | Loss 3.3x |
| Unallocated | n/a | n/a | ($142.6M) | -13.7% | ($10.4M) | -4.7% |
| Consolidated | $8,582.3M | +43.4% | $664.2M | +39.5% | $931.3M | +28.9% |
Shopee Revenue Mix
| Line ($M) | Q3 2026 | Q3 2025 | YoY | % of GMV, Q3 2026 | % of GMV, Q3 2025 |
|---|---|---|---|---|---|
| Core marketplace (transaction fees and ads) | 4,599.6 | 3,055.1 | +50.6% | 11.6% | 9.5% |
| Value-added services (logistics) | 724.4 | 723.6 | +0.1% | 1.8% | 2.2% |
| Marketplace revenue (core + VAS) | 5,324.0 | 3,778.7 | +40.9% | 13.4% | 11.7% |
| Sales of goods | 746.1 | 516.1 | +44.6% | n/a | n/a |
| Total Shopee revenue | 6,070.2 | 4,294.8 | +41.3% | n/a | n/a |
GMV is disclosed rounded to $0.1B, so the take rates above are approximate to within a few basis points.
Shopee: record profit, flat unit margin, and rising depreciation underneath
GMV reached $39.7B, up 23.3%, on 4.5 billion gross orders, up 25.5%. Management put constant-currency growth at close to 26.5%, which means currency cost about three points of reported growth. Revenue of $6,070.2M rose 41.3%. Core marketplace revenue rose 50.6%, more than twice GMV, and the core take rate rose to roughly 11.6% of GMV from 9.5%. Adjusted EBITDA of $269.2M was up 44.7% year on year and 5.4% sequentially.
"Shopee delivered another strong quarter, again achieving new record highs in both top- and bottom-line metrics."
— Forrest Li, Chairman and CEO
Two details temper the headline. First, adjusted EBITDA was 0.678% of GMV against 0.667% in Q2, essentially flat, while ad revenue growth slowed to more than 55% from more than 70%. Management confirmed that part of future monetisation gains may go back into the market rather than into profit. Second, Shopee's segment operating income fell 3.0% sequentially to $155.4M, while its adjusted EBITDA rose. The difference is depreciation and amortisation, which rose 19.6% in the quarter to $113.8M and 49.8% year on year. Adjusted EBITDA excludes the cost of the fulfilment and delivery network that Shopee is building, and that cost is growing faster than the segment's profit.
Shopee's own sales and marketing rose 60.0% to $1,431.6M, or 23.6% of segment revenue against 20.8%. That spending bought something visible. Average monthly active buyers grew 20%, and new monthly active buyers grew more than 50%, accelerating from more than 35% in Q2. Shopee is still widening its funnel, and it is paying for that in the same period that it reports record profit.
Assessment: This is the quarter we asked Shopee to deliver. The $250M bar is cleared, the full-year target is close to certain, and the margin is above its year-ago level for the first time since we moved the pillar to "at risk". We are restoring the pillar. What we are not doing is extrapolating the 44.7% growth rate. Unit margin stalled sequentially, ad growth slowed, management is reserving the right to reinvest, and the rising depreciation shows the business is more capital-intensive than its EBITDA suggests. Shopee's profit is real and still growing. Its path to the 2% to 3% medium-term ambition looks slower than it did in August.
Monee: the book compounds, the spread thins again, and credit turns
Revenue rose 63.9% to $1,621.9M, and the loan book reached $12.6B, up 60.0% year on year and 13.5% in the quarter, comprising $11.2B on-book and $1.3B off-book. Active credit users approached 44 million, up around 28%, with nearly 6 million first-time borrowers added. Average loans outstanding per user grew 23%, and more than 40% of active credit users hold two or more credit products. Personal cash loans and off-Shopee lending together now make up around half of the book.
Adjusted EBITDA grew 15.5% to $298.3M, and the margin fell to 18.4% from 26.1% a year ago and 20.5% in Q2. The causes are explicit in the segment table. The provision for credit losses rose 82.3% to $678.7M, which is 41.8% of segment revenue against 37.6%. Sales and marketing rose 111.8% to $334.6M. Quarterly adjusted EBITDA against the period-end loan book fell to roughly 2.4% from roughly 3.3% a year ago and 2.6% in Q2.
"Our 90-day NPL ratio was 1.3% at the end of September. The modest uptick primarily reflects an evolving credit environment especially in Thailand and Brazil."
— Forrest Li, Chairman and CEO
On the balance sheet, the allowance for credit losses on current and non-current loans receivable rose to $1,382.2M from $1,114.8M at end-June and $842.0M at end-2025. Coverage of gross loans receivable reached about 11.5%, against 10.4% and 9.6%, so reserves are growing faster than the book. Measured against 90-day NPLs, the allowance fell to roughly 8 to 9 times from roughly 10 times at end-June, because NPLs grew faster than reserves. Management also said it has adjusted underwriting criteria and credit limits for selected customer segments, and it described the playbook as "start small, prove economics, then scale".
Assessment: Two quarters ago Monee's spread compression was a mix story on a stable credit book. It is now a mix story on a softening credit book. That is a material change even though 1.3% is still a low absolute NPL ratio, because this book has never been tested by a credit cycle at this size. The ratio rose 30 basis points in one quarter while the book grew 13.5%, and fast growth normally holds the ratio down. Underwriting is being tightened in the two markets driving growth, so loan growth should slow. We take management at its word that the response was early. We also lower our Monee estimates, because a tighter book with higher provisions earns less in the next four quarters, whatever it earns in the long run.
Garena: bookings growth slows to single digits, and the new EBITDA definition hides it
Bookings were $890.0M, up 5.9% against a third quarter last year that grew more than 50% and was Garena's best since 2021. On a two-year stack, bookings are still up roughly 60%. Quarterly active users were 671.7 million against 670.8 million, flat for another quarter. Quarterly paying users grew 4.1% to 68.6 million, a slowdown from 10.2% in Q2, and average bookings per user rose 6.4% to $1.33. Growth came from monetising the existing base harder, and even that has slowed.
"Free Fire anchored this performance with a strong bookings growth, and Arena of Valor continued its strong momentum from its tenth year anniversary celebrations."
— Forrest Li, Chairman and CEO
Content this quarter was a return of last year's Naruto Shippuden collaboration with upgraded gameplay, and Arena of Valor's tenth-anniversary program, including Arena of Valor's largest live event, a concert at the Taipei Dome. Neither of the two licensed titles announced in August, Palworld Online and Monster Hunter Outlanders, was mentioned on the call.
On the new definition, Garena's adjusted EBITDA rose 45.9% to $458.2M. On the old definition, which includes the deferred-revenue effect and tracks bookings more closely, it would have been $510.2M against $465.9M, up 9.5%. Old-basis EBITDA was 57.3% of bookings against 55.4%, so Garena's profitability per dollar spent by players improved again.
Assessment: We noted in August that the definitional change would roughly double the reported growth rate. In Q3 the gap is wider: 45.9% reported against 9.5% on the old basis. The old basis is the better guide to cash generation in a year when bookings are slowing. Garena is a durable, improving cash business growing at a single-digit rate, and we now model it that way. The two licensed titles are the only route back to growth, and their silence on this call is the most important Garena datapoint of the quarter.
Other Services: where the AI bill shows up
Other Services is the residual segment that holds businesses below the reporting threshold. Revenue was $50.1M, up 3.5%, while its operating loss widened to $90.4M from $27.0M and its adjusted EBITDA loss to $83.9M from $25.1M and $46.2M in Q2. The cost of service attributed to Other Services rose 376.9% to $46.5M, and its other operating expenses rose to $140.4M from $75.4M. The release does not say what drove the increase. Its forward-looking-statement language now names "our artificial intelligence, cloud, and infrastructure initiatives" as a risk, and the timing matches the SeaCloud launch.
Assessment: A segment losing $84M a quarter on $50M of revenue is no longer a rounding error. At this run-rate, it costs more than a quarter of Shopee's adjusted EBITDA. We assume most of the increase is the AI and cloud build-out. We model the loss widening further in FY27, because management described SeaCloud as a long-term bet with no contracts, and the capex line is still rising.
Key Operating Metrics
| KPI | Q3 2026 | Q2 2026 | Q3 2025 | YoY |
|---|---|---|---|---|
| Shopee GMV | $39.7B | $38.3B | $32.2B | +23.3% |
| Shopee gross orders | 4.5B | 4.2B | 3.6B | +25.5% |
| Shopee marketplace take rate | ~13.41% | ~12.88% | ~11.74% | +~167bps |
| Shopee adj. EBITDA / GMV | 0.68% | 0.67% | 0.58% | +10bps |
| Shopee monthly active buyers growth | +20% | +18% | n/d | Accelerating |
| Shopee new monthly active buyers growth | +50%+ | +35%+ | n/d | Accelerating |
| Ad revenue growth | +55%+ | +70%+ | +70%+ | Slowing |
| Ad take rate expansion (YoY) | +80bps+ | +90bps+ | +80bps+ | Slowing |
| Ad-paying sellers growth | +35%+ | ~+45% | +25%+ | Slowing |
| ShopeeVIP members | 21M+ | 15M+ | n/d | +40%+ QoQ |
| Monee loan book | $12.6B | $11.1B | $7.9B | +60.0% |
| Monee 90-day NPL ratio | 1.3% | 1.0% | 1.1% | +20bps |
| Allowance / gross loans receivable | ~11.5% | ~10.4% | n/d | Rising |
| Monee active credit users | ~44M | 40M+ | n/d | +~28% |
| Off-Shopee share of SPayLater | ~25% | 20%+ | n/d | Rising |
| Garena bookings | $890.0M | $763.5M | $840.4M | +5.9% |
| Garena quarterly active users | 671.7M | 666.3M | 670.8M | +0.1% |
| Garena quarterly paying users | 68.6M | 68.1M | 65.9M | +4.1% |
| Garena paying user ratio | 10.2% | 10.2% | 9.8% | +40bps |
| Garena bookings per user | $1.33 | $1.15 | $1.25 | +6.4% |
"n/d" means not disclosed at that grain. The Q3 2025 ad metrics are as stated on that quarter's call. VIP members contributed "close to 25%" of Shopee's overall GMV this quarter, while Q2's 24% was measured on Asia GMV only, so the two are not directly comparable. Q3 2025 Garena bookings are derived from the reported 5.9% growth.
Key Topics & Management Commentary
Overall Management Tone: This call was more strategic and less numerical than any in the past year. Management used it to explain two structural decisions, a Monee listing and a new cloud business, and it was unusually candid that it does not yet know SeaCloud's target segment, customers or business model. That candour reassured the market on intent, but it left the size of the spending undefined. Tone on Shopee was as confident as in Q2. On capital it was less precise than in Q2, and on credit management raised the issue itself in prepared remarks, but no analyst returned to it.
1. Sea aims to list Monee "in the next 1 to 3 years"
The most consequential disclosure was structural. Sea aims to give Monee a standalone listing, framed as the logical next stage for a business that has been profitable for three fiscal years and now generates much of its growth away from Shopee.
"We believe Monee has reached the scale and the maturity to stand on its own and would benefit from becoming more independent" … "We hope to list Monee at the right time in the next 1 to 3 years."
— Forrest Li, Chairman and CEO
Management was emphatic that the commercial relationship does not change: Monee stays Shopee's key payment and credit partner, and Shopee stays Monee's key distribution platform. The stated rationale is a dedicated capital base for Monee to move into secured lending, insurance and wealth management. Venue, structure, retained stake and the pricing of the Shopee relationship were not given.
Assessment: A listing is the most direct answer to the sum-of-the-parts discount, because it would put a market price on a business that Sea's consolidated multiple currently values only implicitly. On our FY27 estimates, Monee at 10 times EBITDA would be worth about $13B, or roughly 27% of Sea's current market value. The risks are the timing, given that credit has just softened, and the related-party economics, which will be negotiated between two entities with partly different shareholders for the first time. A listing priced into a weakening credit cycle would set a lower mark than one priced into a stable one. That makes the next two quarters of NPL data more important.
2. SeaCloud: a "frugal startup" with no customers yet
Announced in October and explained in detail on this call, SeaCloud would sell the AI capabilities Sea has built for its own businesses to outside organisations, across infrastructure, models and agentic applications. Management's argument is that token costs are falling structurally and that the durable value lies above bare-metal capacity, in running models efficiently and in agent software and orchestration.
"We are setting up SeaCloud as a frugal startup." … "We will only pursue opportunities with clear customer commitment and attractive returns. We will mainly look to partners and customer prepayment to fund the physical infrastructure. As a result, we do not expect SeaCloud to have a significant impact on our financials for now."
— Forrest Li, Chairman and CEO
In Q&A, management confirmed that no enterprise contracts have been signed. It said it will buy GPUs primarily for internal use regardless of SeaCloud, and it may build data centres for its own needs, citing Monee's data-residency requirements in Singapore. It described SeaCloud as a long-term bet comparable to the decade it took to build the three existing businesses.
Assessment: The strategic logic is defensible. Sea has run AI at scale in fraud, customer service and content generation under regulatory constraints, and very few regional competitors can offer that. The financial framing is the weak point. The promise of no significant impact on the financials "for now" sits beside a capex line that nearly tripled year to date and a segment loss that tripled year on year. The split between internal and external use is the boundary that matters, and it is the one investors cannot see. We give SeaCloud no value and charge its cost in our estimates until a contract exists.
3. The capex step-up nobody discussed
The release's cash-flow commentary shows purchases of property and equipment of $1,042M for the nine months against $351M a year earlier, an increase of 196.9%. Using the first-half disclosure of $495M and the first-quarter figure of $139M, the quarterly sequence is roughly $139M, $356M and $547M. Net property and equipment rose to $1,865.9M from $1,306.8M at year-end. Depreciation and amortisation rose 46.4% to $135.0M.
Assessment: Part of this is Shopee's fulfilment and delivery network, which management says is capital-light because it leases space, and which shows up in Shopee's 49.8% depreciation growth. Part of it is likely GPUs, which management has now said it will buy for internal use. The release does not split the two, and no question on the call asked for the split. Q3 operating cash flow was about $1,501M, so the company can fund this spending. The question is the return on it. Capex running near $550M a quarter is a different capital-intensity profile from the one the stock was valued on a year ago.
4. Credit softens in Thailand and Brazil
Management raised the NPL uptick in prepared remarks rather than waiting to be asked, named the two markets, and described the response.
"We proactively adjusted our underwriting criteria and credit limit management for selected customer segments."
— Forrest Li, Chairman and CEO
The attribution has two parts. One is the external environment in Thailand and Brazil. The other is mix: faster growth in Brazil and in off-Shopee SPayLater, the two higher-yield, higher-loss products that Q2 had already flagged as the drivers of rising provisions.
Assessment: Raising the issue early was the right posture. The substance is less comfortable. The two markets named are also the two growth engines, so tightening underwriting there slows the segment exactly where it was meant to accelerate. Allowance coverage of the book rose to 11.5%, but the allowance fell relative to NPLs, so reserves are keeping pace with the book rather than running ahead of delinquency. We will judge the response on Q4 data: if the NPL ratio stabilises near 1.3% while book growth slows, the tightening worked. A further rise at slower growth would mean the problem is in the book, not the mix.
5. Monee beyond Shopee is now half the book
Personal cash loans and off-Shopee lending make up around 50% of the loan book. Off-Shopee SPayLater more than doubled year on year to nearly 25% of the SPayLater portfolio, and nearly 40% in Malaysia. In Thailand, the ShopeePay unlimited card lets SPayLater users spend their credit limit wherever Mastercard is accepted, and monthly transacting users of the card grew more than 50% in a quarter. Monthly transacting users on the standalone app more than doubled year on year.
"Personal cash loans and off Shopee loans made up around 50% of our loan book at the end of September."
— Forrest Li, Chairman and CEO
Assessment: This is the strongest support for the listing. A lender with half its book outside its parent's marketplace is a standalone financial business. It also clarifies the risk. Off-Shopee lending is where Monee has the least proprietary data and the least control over the transaction, and it is one of the two named sources of the credit uptick. The card rollout "at the pace set by risk performance" is the right discipline. It should also be read as a signal that the pace will slow if risk performance weakens.
6. Shopee's $1B commitment moves to near-certain
Nine-month Shopee adjusted EBITDA is $747.8M. Reaching $1.0B needs $252.2M in the fourth quarter, about 6% below the third quarter's $269.2M and about 25% above the roughly $202.5M of Q4 2025.
"We remain on track to deliver GMV growth of around 25% and adjusted EBITDA of $1 billion for the full year of 2026."
— Forrest Li, Chairman and CEO
Assessment: Barring a disruption, the full-year number is in hand. The more useful signal is that management left the target unchanged rather than raising it, although the fourth quarter can deliver less than the third and still meet it. That is consistent with the Q&A statement that some monetisation gains may be reinvested. The target looks like a figure management is comfortable delivering, not a figure it is trying to beat.
7. Monetisation slows at the margin, and the commission lever is spent
Ad revenue grew more than 55% and the ad take rate rose more than 80 basis points, against more than 70% and more than 90 basis points in Q2. Ad-paying sellers grew more than 35% against about 45%, and spend per seller grew about 13% against more than 15%. Management attributed seller adoption to its AI-driven ad tools.
"Sellers are responding well to the greater value they see in our AI driven ad solutions."
— Forrest Li, Chairman and CEO
Assessment: The core take rate still rose about 210 basis points year on year, and roughly 80 of those came from ads. The remainder reflects higher commissions and other transaction fees, a lever management said in Q2 would slow, and on this call again ranked behind ads. From here, take-rate expansion depends mainly on an ad engine whose every growth metric slowed this quarter. That does not break the pillar, because ad growth above 55% is still exceptional. It does argue for a lower take-rate path in 2027 than the 2026 trend implies.
8. Buyers keep coming, and so does value-seeking
New monthly active buyers grew more than 50% and total monthly active buyers 20%. Management credited offline and online campaigns, distribution through social and video platforms, and a wider range of delivery options. On macro conditions, management acknowledged more value-seeking behaviour and fuel-price pressure but described demand as resilient.
"With these efforts, our new average monthly active buyers grew more than 50% and the total average monthly active buyers increased by 20% year on year."
— Forrest Li, Chairman and CEO
Assessment: Strong buyer acquisition alongside value-seeking behaviour is consistent with Shopee taking share as consumers trade down, which favours the platform with the long tail and the lowest prices. The cost is in the 60.0% growth in Shopee's sales and marketing. The 26.5% constant-currency GMV growth rate is the cleaner measure of demand. Management describes it as similar to Q2, and it does not yet show the macro slowdown that the stock price implies.
9. Logistics: faster at the top, cheaper at the bottom
Shopee is widening its delivery range at both ends. In Indonesia, orders using instant delivery more than doubled with cost per order down, and more than 12,000 offline stores now sell on Shopee. Instant and same-day delivery reached nearly 15% of orders in Greater Bangkok. At the other end, the economical shipping option now carries more than half of Indonesian orders at a cost per order more than 20% below the platform average. Fulfilled order volumes grew around 25% sequentially, and next-day-or-faster parcels grew more than 50% year on year.
"Indonesia, our economical shipping option has gained strong adoption and now accounts for more than half of total orders."
— Forrest Li, Chairman and CEO
Assessment: Tiered delivery is a good margin tool, because it lets Shopee lower the free-shipping threshold without subsidising speed for buyers who do not want it. Value-added services revenue was flat year on year while orders grew 25.5%, consistent with the netting of logistics revenue against shipping subsidies seen in Q2. The capital cost of fulfilment appears in depreciation, not in adjusted EBITDA, so investors should track both.
10. ShopeeVIP passes 21 million and adds a Netflix tier
Membership exceeded 21 million at end-September, up more than 40% sequentially, with monthly retention around 80% and a consistent double-digit uplift in member spending after subscribing. Management said programme unit economics improved sequentially as sellers and partners co-fund more benefits. In September Shopee launched VIP Plus in Indonesia, Thailand and the Philippines, bundling core benefits with a Netflix subscription.
"Membership across markets exceeded 21 million at the end of September, up more than 40% quarter on quarter and members contributed close to 25% of Shopee's overall GMV"
— Forrest Li, Chairman and CEO
Assessment: A paid loyalty programme carrying a quarter of GMV with 80% monthly retention is a real competitive moat in a region where buyers switch platforms easily. The Netflix bundle increases the programme's value but adds a cost Shopee does not control. As in every prior quarter, VIP's cost and its contribution to segment EBITDA are not disclosed.
11. Brazil, content and the agentic channel
Brazil remained the fastest-growing market. Buyer waiting time fell 12%, Shopee added air delivery to the North and Northeast, around 40% of parcels in Greater São Paulo arrive next day, and the Shopee Turbo instant service delivers in under four hours from more than 7,000 partner stores. Fulfilment order volume more than doubled, automotive and home improvement grew more than 80%, and GMV from Mall sellers was up nearly 2.5 times. In content, live and short-video orders grew more than 45% and exceed 25% of Southeast Asian physical-goods orders, and the content business is now positive at the contribution-margin level. Shopee is also an early connector partner for Meta's new AI assistant in Singapore.
"Our new instant delivery service there, Shopee Turbo, already delivers in under 4 hours from more than 7000 partner stores."
— Forrest Li, Chairman and CEO
Assessment: Brazil's move upmarket, through bulky items, Mall brands and faster delivery, is the right way to compete with a repricing incumbent, because it competes on assortment and service rather than price. The content milestone matters because content is one of the three programmes, with logistics and VIP, that Shopee has funded at the cost of margin. On agentic commerce, management's view that checkout and the customer relationship stay on Shopee is plausible for now. It is also the risk that is hardest to disprove from outside.
12. The buyback is done, and it is being doubled
Sea has repurchased 11.1 million shares for $997.2M under the $1.0B programme announced a year ago, an average of about $89.84. The board approved an additional $1.0B with no expiration. Separately, Sea retired the $996.3M of convertible notes it carried at end-June. Nine-month cash used for redemptions was $1,051M, and the balance sheet now carries no convertible debt.
"As of today, we have almost fully completed our $1 billion share repurchase program announced a year ago. We are pleased to announce that we plan to buy back an additional $1 billion worth of shares"
— Forrest Li, Chairman and CEO
Assessment: In August we praised the buyback for buying near $89 when the stock traded at $118. At $76.80 those purchases are 14.5% under water, so the earlier praise was premature. The new $1.0B is about 2.1% of the current market value, and at $76.80 it would buy stock 14.5% below the first programme's average price. It also coexists with a capex budget that is rising sharply and a new business with no revenue. In a capital-allocation debate, a buyback signals that management believes the shares are cheap. It does not answer the debate.
Guidance & Outlook
| Metric | Prior guidance | Current guidance | Change |
|---|---|---|---|
| Shopee FY26 GMV growth | ~25% | ~25% | Maintained |
| Shopee FY26 adjusted EBITDA | $1.0B | $1.0B (9M: $747.8M) | Maintained |
| Shopee medium-term adj. EBITDA / GMV | 2% to 3%, no time frame | Not discussed | Unchanged |
| Garena FY26 bookings | None given | None given | Still absent |
| Monee | None given | Standalone listing in 1 to 3 years | New |
| SeaCloud | n/a | No significant financial impact "for now" | New |
| Share repurchase | $1.0B programme | Additional $1.0B, no expiration | Added |
| Adjusted EBITDA definition | Change flagged for Q3 | In effect; Garena deferred revenue excluded | Basis change |
Guidance was unchanged in numbers and expanded in structure. Management reaffirmed both Shopee targets on the call and added two strategic commitments, the Monee listing window and the SeaCloud funding model, which are not financial guidance but will shape how the stock is valued. The release itself carries no outlook.
Implied fourth-quarter ramp: Full-year 2025 GMV was about $127.3B, and nine-month 2026 GMV of $115.3B is up 27.3%. Growth of 25% for the year therefore needs only about $43.8B in the fourth quarter, up about 19%. The third quarter grew 23.3% as reported. On profit, $252.2M of fourth-quarter Shopee EBITDA would be up about 25% year on year and down 6% sequentially. Both targets leave room to spare. Fourth-quarter growth in line with the third would put full-year GMV growth near 26%.
Street at: Before the print, Bloomberg's consensus compilation, as reported, put third-quarter consolidated adjusted EBITDA at roughly $990M, and the quarter delivered $931.3M on the new basis. We could not source a reliable segment-level consensus. The shortfall against the consolidated figure most likely sits in Monee and Other Services rather than in Shopee, which met the commitment we set for it.
Guidance style: Sea continues to guide one or two Shopee figures a year and nothing for Monee, Garena or Other Services. This year it reaffirmed both Shopee figures with room to spare and raised neither. Given what management said about reinvestment, we read that as deliberate: any upside above the target is more likely to be spent than reported as profit.
Analyst Q&A Highlights
What return SeaCloud should earn, and what it is for
The first question of the call asked for SeaCloud's expected return profile and the demand drivers behind management's confidence. The answer set out a large, long-term demand thesis, an explicit statement of what the company does not yet know, and repeated assurances of capital discipline. No return figure or capital limit was given.
Q: "What are the key use cases or demand drivers that give you confidence that SeaCloud can become a meaningful business over time."
— Alicia Yap, Citigroup
A: "For example, where exactly segment we should focus on, what exact type of customers we should focus on, and what is the best the business model we should use." … "I think the SeaCloud will be really, really a long term bet."
— Forrest Li, Chairman and CEO
Assessment: The candour is useful, and in our experience rare. Management told the market that SeaCloud does not yet have a target segment, a customer type or a business model, which is accurate for a venture-stage business. The difficulty is that the venture sits inside a public company whose capex has already nearly tripled. Until management separates SeaCloud spending from internal AI spending, investors will assume the worst case, and the share price shows they already do.
What Monee still needs before it can stand alone
A question on the milestones Monee must reach to be viewed as a standalone platform, and on the trade-off between integration and independence. The answer described the move as directional and unhurried and committed to keeping the operational integration with Shopee.
Q: "Monee has reached a meaningful scale with over $12 billion in loan outstanding, 44 million active credit user, and also an increasingly diversified business beyond the shopping ecosystem. As you look ahead, what milestone do you believe are still needed?"
— Alicia Yap, Citigroup
A: "In the past, Monee has been benefited hugely from being part of Sea and being integrated, being part of the integrated system." … "Again, so we are not in any rush and I think this is-- we just want to share this directional move with the investors shareholders as early as possible."
— Forrest Li, Chairman and CEO
Assessment: The milestone question was not answered directly. Management gave no profitability, capital or regulatory threshold that would trigger the listing, and the one-to-three-year window is wide enough to wait out a credit cycle. That is sensible for Monee, and it means investors should not price the listing as a near-term catalyst. Its value today is that it states an intention to put a market price on the segment.
What absorbed Shopee's monetisation gains
The sharpest question on Shopee noted that unit margin was flat sequentially despite strong take-rate and ad growth, and asked what absorbed the gains and when operating leverage arrives. The answer pointed to growth investment in both regions and repeated the $1B commitment. It said future monetisation gains will be split between profit and reinvestment.
Q: "We saw that despite the strong growth in take rates and advertising, margins are still flat quarter on quarter at about 0.68%. Could you elaborate what absorbed these monetization gains specifically And which investments should start delivering some sort of operating leverage next year?"
— Divya Kothiyal, Morgan Stanley
A: "We expect going forward, our monetization will also be better primarily from the ad side, less from the fixed commission side. And part of that will flow to the bottom line EBITDA improvement over time. And part of that might go to the investment to grow our market further as well."
— Chris Feng, President
Assessment: The question asked which investments would deliver operating leverage, and the answer did not name one. What it did confirm is the allocation rule: monetisation gains are split between profit and growth, with no fixed share for either. That supports our lower FY27 margin assumption. It also explains why the full-year target was left unchanged when the run-rate could have supported a raise.
Whether AI agents change the economics of a Shopee order
A forward-looking question on agentic commerce asked what happens to advertising revenue, contribution profit and customer data if product discovery moves to an external AI agent. Management said checkout and the buyer and seller relationships remain on Shopee, and compared AI-assistant partners to the affiliate channels it already runs with YouTube and Meta.
Q: "You are an early connector partner for Meta, as you mentioned." … "how do the economics of a Shopee order change? Particularly the advertising revenue and the contribution profit? Do you actually retain the customer relationship and data"
— Divya Kothiyal, Morgan Stanley
A: "At this point in time, all the checkout and seller relationships and buyer relationships happen within the shopping platform. And we did not see a meaningful impact to our economics for the near future"
— Chris Feng, President
Assessment: The affiliate comparison is reasonable, because affiliates already send buyers to Shopee and Shopee keeps checkout, data and ad placement. The phrase "for the near future" limits the answer to the present. If discovery moves into an assistant, the on-platform ad auction loses traffic, and that auction is now Shopee's main source of take-rate growth. This is a long-term risk, not a 2027 risk, and it is worth watching.
Whether peak investment has passed, and whether content is profitable
A two-part question asked when Shopee's investment in VIP, instant delivery and fulfilment peaks, and whether content commerce is now EBITDA-positive. Management said the peak differs by market, because programmes launched at different times, and that content is positive at the contribution-margin level but cannot be measured at the EBITDA level because of shared costs.
Q: "Firstly, on Shopee, when do you think the peak investments would pass considering the ongoing investment in Shopee VIP, instant delivery, fulfillment, And did I hear correctly that content ecosystem is now EBITDA positive?"
— Piyush Choudhary, HSBC
A: "So if you look at our early countries that we started, let's say, some of a program, it does pass the peak investment, but know, there are some other countries came in later." … "Essentially, what we are saying is if we grow the content business more in our platforms, we will have more profit rather than have less profit."
— Chris Feng, President
Assessment: Answering market by market is honest, and it means there will be no single moment when investment peaks: new programmes in newer markets will keep starting. The content answer is a real milestone, because Shopee invested in content for a period of time and its growth now adds profit. The commitment that EBITDA per unit of GMV will keep improving year on year is the one to hold management to.
Whether SeaCloud has customers, and who pays for the hardware
A direct question asked whether SeaCloud has signed any enterprise contracts, whether Sea will own data centres and GPUs or lease them, and what guardrails limit capital spending. Management confirmed there are no contracts yet and that Sea will buy GPUs for internal use in any case. It said Sea may build data centres for its own needs, and that external deals will be judged case by case with partner and prepayment funding.
Q: "But can I check if you have secured any enterprise contracts already and just to clarify, would you invest in data center and GPUs, or would it be capex-light where you will lease these assets?"
— Piyush Choudhary, HSBC
A: "At this moment, we do not have any. We have not secured any enterprise contracts." … "We are going to invest into GPUs, primarily for our internal use cases"
— Forrest Li, Chairman and CEO
Assessment: This was the most important exchange of the call for valuation. Sea's AI capex is driven first by internal demand and only second by SeaCloud. That is reassuring about the venture, because it will not build capacity ahead of contracts. It is less reassuring about total spending, because the internal-use budget has no stated ceiling, and the question about guardrails received a commitment to caution but no number.
How large SeaCloud could become
A question on SeaCloud's likely share of group revenue and profit in three to five years. Management repeated its conviction about AI demand and falling token prices and declined to put a time frame or a size on the opportunity.
Q: "If we look at 3 years or 5 years down the road, in terms of our business mix or our profit mix how significant is SeaCloud to our overall business Would that become another very meaningful business in 3 to 5 years down the road?"
— Thomas Chong, Jefferies
A: "And by the end of this-- will this happen in 3 years or 5 years. I think that depends. I do not have a very, very clear view, and things go very, very fast in this industry."
— Forrest Li, Chairman and CEO
Assessment: An honest answer that gives investors nothing to model. Management expects SeaCloud to generate sizable revenue and profit but will not say when. That keeps the venture at zero value in our model, and it puts the burden of proof on the first signed contract rather than on more strategic commentary.
Whether macro pressure is slowing Shopee
The final question asked whether fuel prices and other macro pressures point to a meaningful or gradual slowdown in GMV growth. Management said constant-currency growth was similar to Q2, acknowledged more value-seeking behaviour, and argued that Shopee's long-tail assortment and price position make it resilient.
Q: "Given that we are seeing the macro is very dynamic, a fuel price and other macro uncertainties how should we think about our GMV outlook or growth rate down the road."
— Thomas Chong, Jefferies
A: "On the Shopee growth, first of all, we do not see a big sharp GMV growth rate slowdown yet." … "we have noticed some more value seeking behavior in our market"
— Chris Feng, President
Assessment: The answer is consistent with the data: 26.5% constant-currency growth and accelerating buyer acquisition do not describe a slowing platform. The word to note is "yet", in management's statement that it does not see a sharp slowdown yet. A platform built on needs-based, price-sensitive purchases tends to gain share in a downturn while its basket size shrinks, which supports the FY26 target and is a reason for caution on FY27.
What They're NOT Saying
- How much of the capex is GPUs. Quarterly purchases of property and equipment rose from about $139M to $356M to $547M across 2026, and nine-month spending nearly tripled. Neither the release nor the call split the total between Shopee logistics, internal AI compute and SeaCloud. Management said SeaCloud will not significantly affect the financials "for now" without quantifying the internal AI budget around it.
- What is in Other Services. The segment's adjusted EBITDA loss more than tripled to $83.9M on flat revenue. The release gives no reason, the prepared remarks do not mention the segment, and no analyst asked about it.
- The effective tax rate, for a second quarter. The rate was 35.6% in Q3 and 34.5% for the nine months, against 26.7% a year ago, and it is still unexplained. Unrecognised tax benefits on the balance sheet rose to $165.2M from $135.7M at year-end, which suggests uncertain tax positions are part of the increase. No one asked.
- Credit performance by market. Thailand and Brazil were named as softening, but no market-level NPL, roll-rate or vintage data was given. The underwriting changes were described only as applying to "selected customer segments". No analyst asked a credit question on a call where the NPL ratio rose 30 basis points.
- Monee listing mechanics. Venue, structure, retained stake, the pricing of the payment and credit relationship with Shopee, and regulatory prerequisites were all left open. The question that came closest, on milestones, received a directional answer.
- The two licensed Garena titles. Palworld Online and Monster Hunter Outlanders were announced in August with a launch targeted this year. Neither was mentioned this quarter, and there is again no Garena bookings outlook.
- The investment gain. A $65.2M investment gain supplied about 8% of pre-tax income and helped net income grow faster than adjusted EBITDA. Its source and whether it will recur were not discussed.
- A 2027 framework. With the fourth quarter close to secured, there is no indication of 2027 GMV growth, Shopee margin trajectory, Monee loan growth after the tightening, or a capex envelope. Sea typically gives full-year Shopee guidance in March. The absence matters more this year because investors are trying to size the AI spending.
- ShopeeVIP economics, again. Management reported 21 million members, 80% retention and improving unit economics, as in prior quarters without numbers. The Netflix bundle adds a new external cost to a programme whose cost has never been disclosed.
Market Reaction
- Pre-print setup: Closed 2026-11-09 at $77.96, down 38.9% year to date against an S&P 500 up 10.3%, down 49.7% over twelve months and down 18.2% over the prior thirty days. The 52-week closing range was $74.01 to $155.05, so the stock entered the print 5.3% above its low and 49.7% below its high. Since our last note's reference close of $118.32 on 2026-08-27, the stock had fallen 34.1% against an S&P 500 down 2.3%.
- The drawdown before the print: The sharpest leg came on 2026-10-20, when Sea's plan for a cloud business and a possible Monee listing became public. The stock fell 9.1% to $80.43 on 18.9 million shares, after a 5.7% fall the previous session and before a further 3.1% decline. From $93.77 on 2026-10-16 to the $74.01 closing low on 2026-10-26, it fell 21.1%.
- Pre-market (2026-11-10, before-open print): Shares traded near $80 before the release, then fell as low as $69.91 shortly after it, down 10.3%. They stayed near $70 to $73 until the call was under way, then recovered to about $78 by 8:05 AM and about $80 before the open.
- Reaction session: Opened $79.75, a gap of 2.3%. Traded up to $85.16, a gain of 9.2%, by mid-morning, then faded steadily to close at $76.80, down 1.5% or $1.16. Volume was 19.5 million shares against a 6.9 million thirty-day average, or 2.8 times normal.
- Benchmark: The S&P 500 fell 0.4% on the session, so the stock underperformed by about one point on the day after a 22% swing from pre-market low to intraday high.
Two prints in one session. The market reacted to the release and to the call separately. The release, read on its own, showed a consolidated adjusted EBITDA figure below the number the Street was carrying, an Other Services loss that tripled, and a 30-basis-point rise in the NPL ratio. A 10% pre-market fall is a rational response to that. The call then supplied two things the release did not: a one-to-three-year Monee listing window, and a SeaCloud funding model built on partners and customer prepayments, with no capacity built ahead of customer commitments. The new $1.0B buyback, which was in the release, added support. The recovery began about 25 minutes into the call and carried the stock to a 9.2% gain within an hour of the open.
A relief rally that did not hold. The fade from $85.16 to $76.80 over the rest of the session is the more informative half. Nothing new came out after the call. What most likely happened is that investors went back to the numbers the call had not changed: capex of about $547M in the quarter, an AI budget with no ceiling, softening credit, and Garena growth in single digits. A stock that closes 9.8% below its intraday high on 2.8 times normal volume is being sold into strength by holders who used the rally to reduce exposure. Options had implied a move of about 14%. The close-to-close move was 1.5%, and the intraday range covered most of that implied move in both directions.
The de-rating came before the print. The 34% decline since our last note is the real market reaction, and all of it came before any of this quarter's numbers were known. It came in two legs. The stock fell 20.7%, from $118.32 to $93.77, between our note and 2026-10-16, before the SeaCloud news. It fell a further 16.9% from there to the $77.96 pre-print close, and nearly all of that came in the three sessions around the 2026-10-20 announcement. The larger part of the decline, about $24.6 of the $40.4 per share, therefore preceded the SeaCloud news, and we cannot tie that first leg to a single event. The second leg reflects a change in how investors view Sea's capital allocation, not in its operating results. The stock entered the print near its 52-week low, priced for an AI-capex problem. The print confirmed that the spending exists and did not resolve how large it will be. Investors who expected the call to cap the spending were disappointed, and the stock is now waiting for evidence on the size of the AI budget rather than on its intent.
Street Perspective
Debate: Is SeaCloud valuable optionality or a capital sink?
Bull view: Sea has run AI at scale in fraud detection, customer service and content generation under strict data-residency rules, and few regional competitors can offer that. Management is funding external capacity with partner money and prepayments and has said it will not build ahead of contracts. The GPUs it buys would be bought for internal use anyway, so SeaCloud is close to a free option on assets the company needs regardless.
Bear view: Capex nearly tripled year to date before SeaCloud has a single customer, Other Services losses tripled, and management cannot yet name the segment, customer type or business model. No significant impact "for now" is not a ceiling. A consumer-internet company entering a market dominated by hyperscalers is the classic diversion of cash from a franchise into a venture, and the stock is right to discount it.
Our take: The bears have the better of the next twelve months, because the cost is visible and the benefit is not. But the bear case depends on spending, and spending is what the next two quarters will show. We assign SeaCloud no value, charge its cost in our estimates, and would treat a disclosed capex envelope, or a first contract on the prepayment model management described, as the evidence that changes the debate.
Debate: Does the Monee listing outweigh the credit turn?
Bull view: A $12.6B loan book growing 60% at a 1.3% NPL ratio, with coverage rising to 11.5%, is a high-quality franchise by any regional standard. Half the book is already outside Shopee, and a listing within three years would put a separate market value on roughly $1.2B of annual EBITDA. That alone would explain much of Sea's sum-of-the-parts discount.
Bear view: The NPL ratio rose 30 basis points in a quarter in which the book grew 13.5%, and the two markets named as softening are the two growth engines. Provisions grew faster than revenue for a second quarter, margins have compressed by nearly eight points, and the book has never been through a credit cycle at this size. A listing priced into a deteriorating cycle would set a low mark, and the window is wide enough for management to wait.
Our take: Both sides are right about different timeframes. The listing is a two-to-three-year value catalyst, and the credit trend is a two-to-three-quarter earnings headwind. We lower Monee's FY27 EBITDA for the tightening and keep the listing as upside we do not include in the fair value. The next two NPL prints decide which side the market listens to.
Debate: Is Shopee's margin improving, or has reinvestment capped it?
Bull view: Shopee just reported its most profitable quarter, its margin is above last year's level for the first time in this cycle, the content programme is profitable, and the full-year target is close to secured. Buyer acquisition is accelerating while competitors cut prices, which is what share gains look like.
Bear view: The margin was flat sequentially, every advertising growth metric slowed, the commission lever is largely used, and management explicitly reserved monetisation gains for reinvestment. Depreciation is growing faster than EBITDA as fulfilment is built out, and segment operating income actually fell in the quarter. The 2% to 3% ambition has no schedule.
Our take: Shopee's profits are now rising year on year, but more slowly than the year-on-year growth rate suggests. We restore the margin pillar on the year-on-year evidence and lower the FY27 margin assumption to 0.80% of GMV from 0.95%. Each ten basis points of GMV margin is worth roughly $190M of FY27 EBITDA, so this assumption carries more valuation weight than any other in the model.
Debate: Has the de-rating gone too far?
Bull view: At $76.80 Sea trades on about 9.4 times FY27 adjusted EBITDA on our reduced estimates, holds $10.1B of net cash and investments, has no convertible debt, and is about to buy back another $1.0B. Revenue is growing above 40%, and the largest segment just reported record profit. The market is pricing a worst-case AI-spending outcome that management has explicitly rejected.
Bear view: Estimates are falling, the tax rate is structurally higher, Garena has stopped growing, credit is turning, and the capex trend has no visible ceiling. A cheap stock with falling estimates and an open-ended capital commitment can stay cheap for a long time, and this one has de-rated for three months without a catalyst to stop it.
Our take: The bears are right that estimates had to come down, and we have cut ours by about a fifth. But the stock has fallen almost twice as far as our estimates, and the gap is now a discount on capital allocation rather than on operations. Discounts of that kind close when management discloses the spending. They do not close on another strong Shopee quarter. That makes the March full-year guidance, and any capex envelope it carries, the most important event for the stock.
Model Update
| Item | Prior assumption | Revised | Reason |
|---|---|---|---|
| Adjusted EBITDA basis | Old (includes Garena deferred revenue) | New (excludes it) | Company definition in effect from Q3 2026 |
| FY26 total adj. EBITDA | ~$3.8B old (~$3.52B new-basis equivalent) | ~$3.56B new basis | 9M new basis $2,656.5M; Garena ahead of plan, Other Services behind |
| FY27 total adj. EBITDA | $5.25B old (~$5.1B new-basis equivalent) | ~$4.1B new basis | About 20% cut across Shopee margin, Monee, Garena and Other Services |
| Shopee adj. EBITDA / GMV, FY27 | 0.95% | 0.80% | Q3 ran 0.68%, flat sequentially; monetisation gains partly reinvested |
| Shopee GMV growth, FY27 | n/a | ~+20% | Constant-currency Q3 ~26.5%; value-seeking behaviour; tougher base |
| Monee loan book growth, FY27 | +45% | +35% | Underwriting tightened in Thailand and Brazil |
| Monee adj. EBITDA, FY27 | ~20% margin | ~$1.3B | Provisions at 41.8% of revenue; NPL 1.3%; margin 18.4% |
| Garena bookings growth, FY27 | +10% | +3% | Q3 +5.9%; QAU flat; paying users +4.1%; licensed titles undated |
| Other Services incl. SeaCloud + unallocated, FY27 | n/a (FY26: ~($215M)) | ~($0.44B) | Q3 Other Services ($83.9M) adjusted EBITDA, 3.3 times the year-ago loss |
| Effective tax rate | ~33% | ~35% | Q3 35.6%; nine months 34.5% |
| Capex, FY26 | n/a | ~$1.6B | 9M $1,042M; Q3 about $547M |
| GAAP diluted EPS, FY26 | ~$2.95 | ~$2.89 | 9M $2.15; higher tax rate |
| GAAP diluted EPS, FY27 | n/a | ~$3.26 | Lower EBITDA, rising depreciation, 35% tax rate |
Resulting estimates. On the new basis, FY26 adjusted EBITDA of about $3.56B comprises roughly $1.01B at Shopee, $1.16B at Monee and $1.68B at Garena, less about $0.29B of Other Services losses and unallocated cost. FY27 of about $4.1B comprises roughly $1.53B at Shopee (0.80% of about $191B of GMV), $1.30B at Monee and $1.70B at Garena, less about $0.44B. Our prior FY27 estimate, converted to the new basis, was about $5.1B, so the cut is about 20%. Shopee's GMV margin is the largest single sensitivity, at about $190M per ten basis points.
Valuation. At the 2026-11-10 close of $76.80 on 633.1 million diluted weighted shares, the equity is worth about $48.6B. Cash and equivalents of $2,905.3M, short-term investments of $5,112.8M and long-term investments of $3,272.9M, less $1,216.5M of borrowings, give about $10.07B of net cash and investments, with no convertible notes outstanding. As before, we exclude restricted cash and deposits payable as customer float, and we treat net loans receivable as working capital of the credit business. Enterprise value is about $38.5B, or 10.8 times our FY26 and 9.4 times our FY27 new-basis adjusted EBITDA. On our prior FY27 estimate on the same basis, the multiple would be 7.6 times. Roughly half the share-price decline since August is lower estimates, and half is a lower multiple. On GAAP earnings the stock trades on about 26.6 times FY26 and 23.6 times FY27.
Fair value range: $95 to $105, from $150 to $165. The range applies 12 to 14 times FY27 new-basis adjusted EBITDA and adds back net cash, which gives $93.40 to $106.32 per share, rounded inward. A sum-of-the-parts cross-check gives about $95.80. It values Shopee at 18 times, Monee at 10 times and Garena at 8 times FY27 EBITDA, charges Other Services and SeaCloud losses at 8 times, and adds net cash. We have lowered the multiple range to 12 to 14 times from 16 to 18 times for four reasons: undefined capital allocation on AI, a credit cycle turning at the fastest-growing profit contributor, a Garena franchise growing in single digits, and a tax rate that converts less EBITDA into earnings. The basis change alone would have justified a slightly higher multiple on a lower EBITDA figure, so the compression is a judgment, not an artefact of the definition. Against the $76.80 close, the range implies 23.7% to 36.7% upside. We include nothing for SeaCloud and nothing for a Monee listing premium.
Thesis Scorecard Post-Earnings
Q2 Commitments Revisited
| What we said to watch in August | Outcome | Grade |
|---|---|---|
| Shopee Q3 adjusted EBITDA above ~$250M, keeping the $1.0B FY26 goal intact | $269.2M; 9M $747.8M; goal reaffirmed | Delivered |
| Shopee FY26 GMV growth of around 25% | Q3 +23.3% reported, ~26.5% constant currency; 9M +27.3%; reaffirmed | On track |
| Adjusted EBITDA definition change; restate history | Implemented; new basis +28.9% vs old basis +12.5% | As flagged |
| Palworld Online and Monster Hunter Outlanders this year | Not mentioned on the call | Silent |
| Whether the ~35% effective tax rate persists | 35.6% in Q3; 34.5% for nine months | Persisted |
| Whether Monee's spread stabilises | Margin 18.4% from 20.5%; NPL 1.3% from 1.0% | Worsened |
| Thesis point | Status | What this quarter showed |
|---|---|---|
| Bull 1: Shopee monetisation | Confirmed, slowing | Core marketplace +50.6% on GMV +23.3%; ad take rate +80bps against +90bps in Q2; ad revenue +55%+ against +70%+. Status unchanged: ON TRACK. |
| Bull 2: Shopee margin normalisation | Confirmed | Record $269.2M; 0.68% of GMV against 0.58% a year ago; $1.0B goal near-certain. Flat sequentially. AT RISK to ON TRACK. |
| Bull 3: Monee multi-market scaling | Confirmed on volume, challenged on earnings and now on credit | Book +60.0%, ~44M users, half the book off-Shopee; margin 18.4%; NPL 1.3%. Volume ON TRACK; earnings conversion AT RISK. Listing window adds a value-realisation path. |
| Bull 4: Brazil | Confirmed | Fastest-growing market again; Mall GMV ~2.5x; bulky-item fulfilment; Turbo instant delivery. Brazil is also one of the two markets where credit softened. ON TRACK. |
| Bull 5: Off-Shopee expansion | Confirmed on scale | Off-Shopee SPayLater ~25% of the portfolio; personal cash and off-Shopee loans ~50% of the book. It is also one of the named sources of the NPL uptick. ON TRACK, with mix risk rising. |
| Bull 6: Garena as a funding annuity | Confirmed as cash, challenged as growth | Old-basis EBITDA +9.5% and 57.3% of bookings; bookings +5.9%; QAU flat. ON TRACK as cash, AT RISK as growth. |
| Bear 1: Investment-phase margin compression | Materialising outside Shopee | Contained at Shopee; Monee margin down ~770bps year on year; Other Services loss 3.3x. CONTAINED at Shopee, MATERIALIZING at Monee and now at group level. |
| Bear 2: Free Fire moderation | Materialising | Bookings +5.9%, paying users +4.1%, QAU +0.1%; two licensed titles silent. Two-year bookings stack still ~+60%. EMERGING to MATERIALIZING. |
| Bear 3: Below-the-line earnings drag | Materialising | Tax rate 35.6%, second consecutive quarter near 35%; net income helped by a $65.2M investment gain. EMERGING to MATERIALIZING. |
| Bear 4 (new): Capital-allocation drift into AI | Emerging | 9M capex +196.9%; Q3 ~$547M; SeaCloud with no contracts; internal AI budget undisclosed. New, EMERGING. |
| Bear 5 (new): Credit cycle turn at Monee | Emerging | NPL 1.0% to 1.3% in a quarter; Thailand and Brazil softening; provisions +82.3%; coverage 11.5%. New, EMERGING. |
Overall: The thesis is stronger at Shopee and weaker everywhere else. Shopee did what we asked: the margin pillar returns to "on track", the profit commitment is effectively met, and the business is growing its buyer base during a consumer slowdown. Against that, two existing bear points moved to "materialising" (Garena's growth and the tax rate), and two new ones opened (AI capital allocation and Monee credit). On balance the thesis is weaker than in August. Revenue growth is not the problem. The problem is that more of each revenue dollar is being spent on things that do not yet earn a return, and less is reaching shareholders.
Action: Maintaining Outperform, trimming conviction, and cutting fair value to $95 to $105. We were wrong at $118.32. Part of that error is that we did not anticipate a capital-allocation shock, but about three-fifths of the decline came before the SeaCloud news, so the shock explains only part of it. Our estimates fell about 20%. The stock fell 35%, and at 9.4 times our reduced FY27 EBITDA with a fifth of its market value in net cash, it now prices a worse outcome than our revised numbers. In the next two quarters we will grade three things. First, whether the NPL ratio stabilises near 1.3% as underwriting tightens. Second, whether Sea discloses a capex envelope or a first SeaCloud contract on the prepayment model it described. Third, whether March brings a 2027 Shopee framework consistent with year-on-year margin improvement. A further NPL rise at slower loan growth, or a second quarter of rising capex without disclosure, would make the Outperform hard to defend.