Sea Limited

IThe business

Sea Limited is three businesses under one founder in Singapore. Shopee is the largest online market in Southeast Asia and Taiwan and a large one in Brazil. Monee lends to Shopee’s buyers and sellers, and more and more away from Shopee’s checkout. Garena publishes Free Fire, a mobile shooting game in its ninth year. In the twelve months to 30 June 2026 revenue was $27.7B and net income was $1.7B C (calculated).²³⁴ The fiscal year ends on 31 December, so TTM on this page means the four quarters to 30 June. Sea has not yet reported Q3, and I expect that in mid-November, though no date has been announced.

Read that table from the right. Shopee is nearly three quarters of revenue and less than a quarter of the profit. Garena is a tenth of revenue and about half the profit. So the part of the company everyone talks about is funded by a game.

Q2 tells the story of 2026. Revenue rose 48.1% to $7.79B. Gross orders rose 27.5% to 4.2 billion and GMV, the value of goods sold, rose 28.4% to $38.3B. Net income rose 10.6% to $458M.³ Shopee’s revenue rose by $1.82B from a year earlier, and its adjusted EBITDA rose by $28M C (calculated).³ That is about 1.5 cents of profit on each new dollar of revenue. Most of it is Sea raising its fees and spending the money on delivery, cashback, members and advertising. Some of the spending doesn’t even show as cost: Sea says value-added revenue fell on "higher revenue net-off against shipping subsidies", so the reported revenue is already lower by part of what it spent.³

Management calls this investing, and says so plainly: 2026 is a year of leaning in.¹⁰ The market has spent the year asking how long it lasts. The stock closed at $74.24 in Q4 2026, down from its 52-week high close of $157.76 on 5 November 2025, and it has fallen 43.5% since the August results C (calculated).¹

Shopee

Shopee takes a commission on each sale, charges sellers for advertising and runs much of the delivery. Core marketplace revenue, mostly fees and ads, was $4.26B in Q2, up 65.6%, or 11.1% of GMV against 8.6% a year earlier C (calculated).³⁴ Ad revenue rose more than 70%, and the number of sellers paying for ads rose about 45%, on a third-party transcript of the call.⁶ Sea’s sellers are paying a lot more, and so far they are staying.

It operates in eight core markets. Revenue by place in 2025 was 62.7% Southeast Asia outside Singapore, 24.1% Latin America, 8.7% the rest of Asia, which includes Taiwan, and 3.5% Singapore.² The 20-F says only that the top three e-commerce markets were more than half of e-commerce revenue, and Sea does not publish GMV by country.² That is a real gap. I can’t see Brazil’s profit, or how much of the group’s margin Indonesia carries.

Monee

Monee is Sea’s lender. Loans outstanding were $11.1B at 30 June, up 62.5%: $10.0B on Sea’s own balance sheet and $1.1B funded by others.³ Most are small, short loans. SPayLater is buy-now-pay-later at Shopee’s checkout, and more than a fifth of it is now used away from Shopee.⁶ Active credit users passed 40 million, with 5.3 million first-time borrowers added in Q2.⁶ Average balances were around $250 in March.¹² Brazil passed $1B of loans in Q1, up more than 250%.¹² Sea runs digital banks and took $4.3B of customer deposits by June.³²

Monee earned $288M of adjusted EBITDA in Q2 on $1.40B of revenue. Revenue rose 58.9% and adjusted EBITDA rose 12.8%.³ Provisions for credit losses and marketing are growing faster than the loans. I come back to this in Financials, because it is the most important page of accounts Sea publishes.

Garena

Garena’s bookings, the cash players spent, were $763.5M in Q2, up 15.5%. It had 666 million quarterly users and 68 million paying ones, and adjusted EBITDA was 56.3% of bookings.³ Free Fire draws more than 100 million players a day.⁶ The 20-F says it was "a significant majority" of the segment’s revenue and profit in 2025.² Two new titles are coming: Palworld Online, developed and published by Garena under license, and Monster Hunter Outlanders, built by Tencent on Capcom’s franchise.⁶³

That chart is the warning. Bookings fell 61% in two years after 2021, helped along by reopening and by the loss of the Indian app stores in early 2022.¹³¹⁵ A game is a hit until it isn’t. Free Fire came back, which most games don’t, and I can’t tell you how many more years it has.

The newest businesses

On 20 October Forrest Li told staff that Monee is approaching the point where it could go public, and that the internal cloud platform, rebuilt over two years by about 500 people for AI work, will be sold to outside customers as Sea Cloud.¹⁶ Bloomberg reported in June, as relayed by the trade press, that Sea was testing a chatbot called Migoo in the United States, led by president Chris Feng.¹⁷ DBS, also secondhand, puts Migoo’s cost at $73M to $175M of 2027 adjusted EBITDA.¹⁸ The stock fell 9.1% on the day of the letter C (calculated).¹¹⁹ Sea gave no budget for either.¹⁶

Sea has one business that pays, one that grows and one that lends, and in October it added a fourth nobody has priced.

IIThe moat

A market protects itself when buyers go where the sellers are and sellers go where the buyers are. Add a delivery network that gets cheaper with volume and a lender that knows who pays, and a rival has to beat you at several things at once. Sea has all of that. The question is how much of it survives a rival with a different way of getting to the buyer.

Scale, and why it hasn’t been enough

Momentum Works puts Southeast Asian platform GMV at $157.6B in 2025, up 22.8%, with Shopee at 53% and first in all six markets it tracks.⁵ That is an estimate from a research firm, and I can’t audit it. The same report has TikTok Shop, with Tokopedia, at 65.7% of Shopee’s GMV.⁵ In Vietnam, Metric data cited by KrASIA has Shopee above 56% and TikTok Shop above 41% in 2025; Momentum had estimated 61% and 24% in 2023.²¹ So the largest player in a market growing over 20% has lost a lot of relative ground to one rival in two years.

TikTok did it by starting from the other end. Shopee waits for you to come looking. TikTok puts the goods in front of you while you watch something else. Shopee answered with its own live selling and short video, and those orders are now more than 25% of physical goods orders in Southeast Asia.⁶ It also struck deals with YouTube, Facebook and Instagram, which tells you it needs other people’s attention.⁶

Logistics

This is the part I rate most. Shopee runs its own delivery in much of the region, and the numbers are good: instant delivery in Indonesia up about 80% with cost per order down about 20%, and delivery as fast as one hour in some cities.⁶ Fulfillment orders rose more than 20% in a quarter.⁶ Sea rents the sites rather than owning them, so this has come with little capital. The cost is in the income statement instead: Shopee’s cost of revenue rose 51.9% in Q2.³ A rival that wants the same promise has to build the same thing, and TikTok has less reason to.

The take rate as a test

There is a simple test of a market’s strength: raise the price and see who leaves. Sea raised core take by about 2.5 points of GMV in a year, and GMV growth stayed near 30% C (calculated).³⁴ Management says it reinvests much of that in the ecosystem, so sellers get some of it back.⁶ Still, that is the best evidence on this page. It says sellers have nowhere better to go at scale. The Morgan Stanley analyst on the Q2 call noted that TikTok Shop raised commissions too, so this may say something about both of them, and about a market that is settling down.⁶

Monee’s edge

Sea sees what you buy, how you pay and how you behave in its apps. It says its newest models, built on the same architecture as large language models, raised approval rates by about 10% at similar risk.⁶ I believe the data is an advantage inside Shopee. I don’t know whether it travels. Off-Shopee lending, Brazil and larger loans to better-off people are all newer than the data that trained the models, and none has been through a bad year at this size.

Where it could fail

Agents. If a chatbot picks the product, Shopee’s ads lose their shelf. Sea has put the Shopee app inside ChatGPT in eight markets, and its own Migoo is partly a defense. The 20-F names AI platforms among its competitors.²⁰² Nobody can size this yet.

Regulators and borders. Sea runs regulated banks and lenders in several countries, and uses contracts with local entities, called VIEs, to control some businesses where foreign ownership is limited.² Free Fire was made unavailable in the Indian app stores in 2022 by government action, and the 20-F says it remains so.² A foreign company selling to Indonesians and lending to them has more than one way to fall out of favor. Indonesia is the market to watch. In April 2026 the Ministry of MSMEs said it was drafting a ceiling on the digital service fees platforms charge sellers, and press named Shopee, TikTok Shop and Tokopedia as affected.³⁶ I found no published level. A Finance Ministry rule also has platforms withhold 0.5% income tax on seller sales, with sellers under Rp500 million a year exempt if they file the right papers.³⁷ A ceiling would hit the 2.5 points of GMV that Sea added to its take rate, and I can’t size it.

Sellers paid 2.5 points more and stayed. That is real strength, and it sits beside a rival now two thirds of Shopee’s size.

IIIThe competitive advantage

Sea’s advantage is density in hard places. Thousands of islands, poor addresses, few credit files and buyers who shop on cheap phones reward a company that has built delivery and payments for those conditions. The advantage is local in each market, so it has to be won separately in each one.

Where Sea wins

It wins on cost. Management says its prices beat other marketplaces and offline shops, and that it stays profitable in Brazil while growing above the market.⁶ It wins on the link between shopping and credit. Most rivals are strong in one or the other, and Lazada is well behind in GMV.⁵ It wins on self-funding: Garena’s cash paid for the war. Few challengers anywhere have a game to pay for a marketplace.

Where it loses

It loses on attention, as above. In Brazil it is up against MercadoLibre, which is spending hard. MercadoLibre’s GMV rose 44% in Q2, 57% in Brazil, while its operating income fell 17%, and it says it is choosing long-term value over short-term profit.⁷ Shopee’s management says it is growing above the market and faster than close rivals in Brazil.⁶ Both statements can’t tell me who is winning, since Sea gives no Brazil figure. Two well-funded companies cutting shipping prices in the same market at the same time is a poor place to look for margin.

It may also lose on price of capital. On DBS’s numbers, reported secondhand, Pinduoduo has entered four Southeast Asian markets, and 10% to 13% of Shopee’s GMV is exposed to it.¹⁸ A rival that can run at a loss on someone else’s cash flow sets the floor for everybody. Sea did that to others from 2019 to 2022, so it knows how it works.

Against MercadoLibre, the nearest business on any page

MercadoLibre is the fair comparison: marketplace, logistics, credit, emerging markets, one strong founder. Shopee’s quarterly GMV is 1.7 times MercadoLibre’s C (calculated), and Sea’s market value is 57% of MercadoLibre’s C (calculated).³⁷⁸ Part of that gap is earned. MercadoLibre’s book is larger, its fintech has more users, and its margin is real and long-standing. Part is Garena, which markets discount as a game. And part of it is the market saying Sea’s reinvestment will not stop.

Sea has built delivery for hard geography, and it fights on two fronts against rivals that are spending as hard as it is.

IVManagement

Forrest Li founded the company in 2009 and is chairman and chief executive. He holds 45.5M Class B shares with 15 votes each. Including proxies from employees and others, he had 57.6% of the votes at 31 March.² The 20-F headline of 16.0% of the shares overstates what he owns, because it counts about 43.2M shares that others own and he only votes, plus options.² What he owns is about 46.8M shares after this year’s trades, roughly 7.7% of the economics C (calculated).²⁹ At $74.24 that is about $3.5B C (calculated), and he also has options on 9.0M shares at $15 and 5.0M at $120.²⁹

How they are really paid

Sea doesn’t publish a pay table by person. It is a foreign private issuer, and the 20-F gives one number: $3.13M of cash for all directors and officers as a group in 2025.² So salaries don’t matter here. Options do. The large grants were at $15 between 2018 and 2020, and then at $120 on 17 April 2022: 5.0M to Li, 2.4M to Feng, 2.0M to Ye, and more to others.² The $15 options are worth a lot at any price. The $120 options pay nothing below $120, 62% above today, and run to 2032 C (calculated). They pay for the stock recovering, with six years left on the clock. That lines up with holders, if they’re granted at that price and not repriced.

Now look at the pay committee. It has two members, Li and Khoon Hua Kuok, and Li chairs it. The nominating committee has the same two, with the same chair.² The committee reviews the chief executive’s pay. Sea follows Cayman practice in place of the NYSE rules on a majority of independent directors, and on the size of its audit committee.² At this year’s annual meeting no proposal was put to shareholders at all.²²

Who can say no

Nobody. In February 2022, four months into a 65% fall in the stock, shareholders approved raising Class B votes from three to fifteen each.²³¹ In September 2022 Tencent, which owns 17.0%, handed its votes to Sea’s board by irrevocable proxy, and a director resigned the same day.²⁴² Directors and officers as a group hold 68.2% of the votes.² Li can launch a cloud business by email to staff, and that is what happened.

Their own money

Since March 2026, Sea’s insiders have filed Forms 3 and 4, which give a look nobody had before. I parsed all 227 filings to 29 October. They show 5.12M shares sold for $539.7M, at an average of about $105, and no open-market purchase by anyone C (calculated).⁹ Ye sold $194.5M. Li sold $147.3M in August, 1.14M shares at about $129, starting on 11 August, the day the Q2 results lifted the stock.⁹ Director David Ma sold $117.2M in March and April.⁹

In fairness: 204 of the 206 sale filings I parsed cite a Rule 10b5-1 plan, set months ahead. The other two are directors Kuok and Heng, who sold $4.4M between them in August C (calculated). Li’s was adopted in September 2025, when the stock was far higher. He sold about 2.4% of his stake, after exercising 1.47M options in March.⁹ Ye sold steady amounts each month and still has 22M shares. In October Feng exercised 1.0M options at $15 and Ye 1.65M, and neither had sold shares since by 29 October.⁹ Insiders can’t easily buy just before results, so the lack of purchases since August means less than it looks. Rich founders sell for many reasons, and only one is that they think the stock is expensive. Still, the record is that the company bought at about $91 while the people running it sold at about $105. Nobody with inside knowledge has put new money in at these prices so far.

Capital over a full cycle

This is where I give Li real credit. He raised money when it was cheap and spent it to win. Sea sold stock at about $192 in December 2020 and about $314 in September 2021, a month before the all-time closing high of $366.99.¹¹¹ That $3.97B raise, at more than four times today’s price, is the best trade in the company’s history. It then lost $7.0B at the operating line from 2016 to 2022 buying Shopee’s position C (calculated).¹¹ When the money stopped being free in 2022 it turned hard. E-commerce guidance was suspended in August, and headcount fell from 67,300 to 62,700 by the end of 2023.²⁵¹¹ Adjusted EBITDA went from −$878M in 2022 to $1.2B in 2023.¹³

The record has two sides. The March 2022 guide for Garena was $2.9B to $3.1B of bookings, and the year came in at $2.8B. The same release said e-commerce revenue would be $8.9B to $9.1B, and that guide was pulled five months later, with the year at $7.3B.¹⁵²⁵¹³ Against that, guidance for 2025 Shopee GMV growth began near 20% and the year was 26.8%.¹⁴¹³ Management has learned to promise less. Its statement in March 2022 that Shopee and SeaMoney would largely fund themselves by 2025 looks about right, if you count deposits.¹⁵

The buyback is small and sensible: $600M at about $91, against a $1.0B plan.¹⁴¹⁰³ It offsets stock pay rather than shrinking the company much. Diluted shares were 638M in 2025 and 634M in H1 2026.²³

The incentive question for the next five years is simple. A founder who controls the votes, who was right to spend in 2019 and right to stop in 2022, has just said he wants to start again. His letter says so: "we build something, get it to the starting line, and then start again".¹⁶ That is a fine trait in a founder. For a holder of the Class A shares it means the capital plan belongs to one person.

He has been right on the two biggest calls in Sea’s history, and there is no one in the room who could stop him on the third.

VFinancials

The headline numbers look better than the business, and the reason is the lender. Read this section as three questions: is the growth real, is the margin coming, and what does an owner actually get?

Growth is real. Revenue rose from $0.8B in 2018 to $22.9B in 2025 C (calculated), and the TTM is $27.7B.¹¹²³⁴ Gross margin has been about 44% for three years. Operating margin went from −12% in 2022 to 8.7% in 2025 and 8.2% in the TTM C (calculated).¹¹²

The margin that went backwards

Shopee’s adjusted EBITDA was 0.92% of GMV in Q1 2025 and 0.67% in Q2 2026 C (calculated).²⁶³ It got that way on purpose. Sea promised at the start of the year that Shopee’s profit would be no lower than 2025 in dollars, which with 25% GMV growth was a promise of a lower margin.¹⁴ In August that became a billion dollars.³ On Sea’s H1 numbers that needs about $522M in H2 against $479M in H1 C (calculated), and the release says it is optimistic.³

On the call, management said "three percent is quite within our reach as a percentage" of GMV and that some markets are above it; the same answer called the path to 2% relatively straightforward.⁶ That is from a third-party transcript, and it gives no date, so I treat it as an aim and not a target. If some markets really are above 3%, others must be far below 0.67%, and I can’t see which.

Monee: where the accounts need the most care

Monee is growing fast and earning less on each dollar. Adjusted EBITDA on average loans, annualized, was 16.1% in Q2 2025 and 11.0% in Q2 2026 C (calculated).⁴³ Management says this comes from mix: Thailand, Malaysia, off-Shopee lending and better-off borrowers carry lower rates by design.⁶ Sales and marketing at Monee rose 139.8% to $294M in Q2, and promotional interest-free loans are part of that.³⁶ Some of this is a good choice. It still means the headline return is falling.

The 1.0% non-performing ratio is the number everyone quotes, and it tells you less than it seems to.³ Gross write-offs of consumer and SME loans were $983M in 2025. Average on-book loans that year were about $6.4B, so about 15% of the book was written off in a year C (calculated).² Of that, $476M was from loans made in 2025 itself.² These are short loans that go bad quickly and leave the book quickly, so few sit past 90 days. The 20-F doesn’t give the day count for write-off. The allowance tells the truer story: $822M at December, or 9.8% of on-book loans C (calculated).² Provisions were $555M in Q2, about 24% of the average book at an annual rate C (calculated).³ I don’t read this as bad. A lender charging high rates on short loans can run this way for years. It means the safety of this book rests on pricing and on how fast Monee can pull back, and it will not look like a bank’s accounts.

The test is whether Monee can stop lending when things turn. Its loans are short, which helps. Sea lived through a smaller version in 2022. Its release said the ratio fell from less than 4% in Q3 to less than 2% by the year end, "mainly" because the write-off period in one market was cut from 180 days to 120.¹³ So this ratio can fall on a change of policy, with no change in who repays. Total loans were $2.1B then, about a fifth of today.¹³

Tax

Income tax took 33.9% of pre-tax profit in H1 2026, and 28.5% in 2025 C (calculated).³² The Singapore rate is 17%. Withholding taxes on cross-border flows and profits in one place that can’t be set against losses in another push the rate up. In 2025 valuation allowances alone added 5.5 points.² Most comparisons of Sea to its peers start from adjusted EBITDA, which skips this. It matters. It is why Q2’s net income rose 10.6% while operating income rose 33.3%.³

What an owner can take out

Screens show Sea with about $4.4B of free cash flow, which would make the stock look cheap at $47B. It isn’t that number. Three things flatter it. Stock pay of $0.6B is real pay C (calculated). Credit provisions of $1.8B are added back in operating cash flow, while the actual loan losses and all the loan growth show up lower down, in investing.³² And operating cash flow takes in seller float and Garena’s deferred revenue, which depend on timing.

So I get two answers, and I show both, because I can’t tell you which is right. If Monee can fund 82% of its loans from outside, an owner took out about $1.4B, or $2.19 a share E (estimate). As the loans were really funded in the TTM, with $3.4B of Sea’s own money, the answer is about $0.5B, or $0.72 a share C (calculated). Net income was $1.7B, or $2.59 a share C (calculated). Against the $6.98 a share of free cash flow on a screen, an owner got between a tenth and a third.

The gap between those two answers is growth. The gap is the money going into new loans. Strip out growth, keep the loan book flat and cut capital spending to depreciation, and an owner would have taken out about $2.5B, or $3.87 a share E (estimate). That is about 46% above net income, and some of it is seller float and Garena’s deferred revenue, which I can’t separate. So at a stand-still an owner gets about the earnings. Growth is what pushes owner cash below them. Shopee is also spending heavily on growth inside the income statement, and if it stopped the profit would be much higher. By how much is the whole debate, and I come to it in Valuation.

The balance sheet

Cash, restricted cash and investments were $14.3B at 30 June. But $4.3B of that belongs to depositors and $3.3B to sellers and customers in escrow. Take those out with $1.2B of borrowings and the $1.0B of convertibles, due in September 2026, and net financial assets are $4.3B C (calculated).³ Sea also has about $4.1B of its own money in the loan book C (calculated). That is 43% of it, far above what a lender normally needs, and about $2.4B of it could come back if outside funding grew. I leave that $2.4B, about $3.75 a share, out of my value E (estimate). I can’t tell how much regulators in each country would release.

The convertibles were struck near $477 and were never going to convert at these prices.² Dilution from stock pay has slowed: $0.7B in 2022 and $0.6B in the TTM, about 2% of revenue C (calculated). Capital spending is low for a company this size, $0.8B in the TTM, or 2.9% of revenue C (calculated). It more than doubled in H1 to $495M.³⁴

Screens show $6.98 a share of free cash flow. An owner got between $0.72 and $2.19.

VIValuation

How this goes badly wrong

Before any numbers I ask what would most likely wreck this, and work back. I see three ways.

1. The margin never comes. TikTok Shop, Pinduoduo, MercadoLibre and Lazada keep buying share, so Sea has to keep handing its higher fees back to buyers. GMV keeps growing and Shopee ends the decade at about 1% of GMV. This is the most likely of the three, and my bear case is built on it. It would not show as a loss. It would show as year after year of "investing".
2. The credit book has a bad year. About 15% of the book was written off in a good year.² Sea is also adding new borrowers, new countries and larger loans, and the models haven’t seen a downturn at this size. A recession in Indonesia or Brazil, or a currency crisis, would arrive with deposits to repay, and with the group’s profit now resting on this book. Deposits of $4.3B can leave faster than loans can be collected, though the loans are short.
3. The game fades while the founder spends. Garena is about half of group adjusted EBITDA, and most of that is one nine-year-old title.¹⁰³² If Free Fire does in the next two years what it did in 2022, the cash that paid for Shopee’s war goes away. If that happens while Sea Cloud is taking $1B or $2B a year, the second business and the fourth are competing for money from a shrinking first.

All three share a feature. None would show in this year’s accounts. And each makes the others worse, because a bad credit year in a downturn would come with weaker consumers on Shopee.

What I can’t judge

I can’t tell you Shopee’s real margin when it stops investing, because I don’t know where the investing ends, and Sea doesn’t give country profit. I can’t judge how Monee’s loans behave in a recession, because nobody has seen it at this size, including Sea. I can’t judge how long Free Fire lasts. And I can’t size Sea Cloud. There is no budget, no customer and no price in anything Sea has published, and an article by Momentum Works says the letter offered little on capital spending.¹⁶¹⁹ So I show a range and don’t pretend the middle of it is a measurement.

What the price implies

At $74.24 Sea’s market value is $47.1B on 633.9M diluted shares C (calculated).¹³ The close is confirmed on a second feed.¹²⁸ Take off $4.7B of net financial assets at 1 January and the price implies $42.3B for the business E (estimate). That is 28 times TTM net income C (calculated). So this isn’t a statistically cheap stock today. It is cheap only against what the earnings might become.

If you believe the screen, owner cash is $3.9B and the price needs almost no growth. That is wrong, for the reasons in Financials. From net income it needs 13% a year for ten years. From owner cash at 18% equity it needs 16%, and from owner cash as loans are really funded, 32% E (estimate). Revenue is growing 48%, so 13% to 16% is not a hard ask. The question is only whether revenue turns into profit.

Another way to see it. Put Garena at ten times my estimate of 2026 profit after tax, $1.3B, and Monee at twelve times its $0.8B E (estimate). That is $23.3B, which leaves $19.0B for Shopee and everything else, or about 12% of 2026 GMV E (estimate). Those multiples are my own, and you can move them. It takes low multiples on both, for good reasons. Even so, the price is not paying much for the largest market in Southeast Asia.

The market is pricing Shopee at about 1.5% of GMV in ten years. It earns 0.67% today, and that gap is what the stock is about.

The discount rate

I use 11.5%. The capital asset pricing model with the 4.94% ten-year Treasury and Damodaran’s 4.20% premium gives 11.4% on Sea’s five-year beta of 1.55, 11.1% on its full history of 1.46 and 9.5% on the three-year beta of 1.08 C (calculated).^2930311 Betas use month-end price returns against SPY. A US beta misses the fact that about 87% of Sea’s revenue comes from places where a bond default or currency crisis is a live risk. Damodaran’s January table gives country premiums of about 2.5% for Indonesia and the Philippines, 3.2% for Brazil and 0.8% for Taiwan, which come to about 2.4 points on Sea’s revenue mix C (calculated).³²² Add those to the premium at a beta of 1.08 and you get 12.1%, so two routes land between 11.4% and 12.1% C (calculated). I take 11.5%. That is a point above Meta and 1.5 above Tencent on this site, and I think it is earned. A point on the rate moves my base by about $12 E (estimate).

Three cases

Each case is built by year, 2027 to 2036, valued at 1 January 2027. Shopee is GMV times a margin. Monee is average loans times a pre-tax return, and 18% of each new dollar of on-book loans is kept as equity, which an owner can’t take out. Garena is bookings times a margin. Then come other losses, stock pay, depreciation, capital spending and tax. The tax rate starts at 33% and falls to 25% by 2035, as more of the profit lands where it can use past losses.

The base case takes management at something like its word, and later.

• GMV grows 19% in 2027 and fades to 6% by 2036, to $458B, 11.1% a year E (estimate). It is growing 28% now, and Sea has guided to about 25% this year.³⁶
• Shopee’s margin on GMV goes from about 0.66% in 2026 to 0.8% in 2027, 1.45% in 2030 and 2.0% by 2035 E (estimate). DBS expects 0.76% in 2027, secondhand.¹⁸ The 2.0% end point is management’s easy step, nine years out, and about where DBS puts Pinduoduo and Alibaba. That is the assumption doing the most work.
• Loans reach $18B in 2027 and $55B by 2036, 15% a year E (estimate). The pre-tax return falls from about 10.5% now to 6.0% as mix moves to safer borrowers and competition arrives. It stays smooth in the base, and it won’t be smooth in life.
• Garena’s bookings grow 5% in 2027, on the new titles, then 2% to 3%, with the margin slipping from 56% to 50% E (estimate). This is a judgment about a game. If Free Fire is still there in 2036 I’d be surprised, and the base assumes something else takes its place.
• Sea Cloud takes $500M of capital a year to 2031 and $300M after, with $300M a year of other losses at first E (estimate). I have no basis for this. It sits here as a placeholder until Sea gives one. Capital spending otherwise stays at 0.4% of GMV, near today’s rate, since Sea rents its warehouses and says it means to continue.⁶

My 2027 earnings come to about $3.55 a share E (estimate). DBS’s $105 target at 21 times 2027 non-GAAP earnings implies about $5.00 before stock pay, and mine would be about $4.46 on that basis C (calculated), so I am below the one estimate I can see.¹⁸

The bear case is way one, with a mild version of the others: GMV growth fades to 5%, Shopee stops at 1.2%, Monee has one bad year in 2028 with returns ending at 5%, Garena’s bookings fall 16% over three years and Sea Cloud takes $1B a year with little to show E (estimate). The bull case is management’s aim: 3.0% on GMV, $576B of GMV in 2036, loans of $73B, and Garena with a second hit E (estimate).

Weighted 25%, 50% and 25%, the cases give $97 E (estimate). At 20%, 50% and 30% it is $104, and at 30%, 50% and 20% it is $91 E (estimate). The value after 2036 is 60% of my base C (calculated). That is lower than most pages on this site, because Sea has so far to go in the first ten years.

In the stress case all three ways happen together. Sea survives it, because it has net financial assets and low capital needs, but $11 a share is an 85% loss from here E (estimate). I put no probability on it. It is on the page because 2022 was not long ago: from its October 2021 high close the stock fell 90%, to $35.11 in December 2023 C (calculated).¹

The one number

Most valuation debates have five moving parts. This one has one.

Move Monee’s return by a point and the base moves about $4. Double Sea Cloud to $1B and it moves about $4 too. Move Shopee’s end margin by half a point and it moves about $19 E (estimate). Every other input is small next to that. The price equals my base with a 1.5% margin and nothing else changed, or with GMV at 6.0% a year instead of 11.1%, ending at $285B E (estimate). A return on Monee of about 1.2% would do it too, which tells you the market isn’t only worried about credit. I also won’t pretend this is one independent number: Shopee’s margin depends on rivals’ choices, and it may depend on how much credit Monee offers at checkout.

What about the odds? My base needs four things to go roughly right at once: Shopee at 2% (I say 55%), Monee’s return holding near 6% with no large accident (65%), something like Free Fire’s bookings holding (65%), and Sea Cloud staying small (70%). Multiplied, that is 16%, about one in six E (estimate). These are my judgments and aren’t independent. So don’t read $93 as what I expect to happen. It is a middle, in a wide range. The good news in that bridge is that three of the four barely matter one at a time. So the argument stays simple: Shopee at 1.5% or better covers the price, and I’d put that at about two in three E (estimate).

The margin of safety

What does the same dollar earn elsewhere? My base returns 13.3% a year at $74.24, against 4.94% on a ten-year Treasury with no business risk and 11.5% that I require E (estimate).²⁹ The closest choice on this site is Tencent, which owns 17% of Sea.² My page has it 24% below its base at a 10.0% rate, reaching the price at 12.4%, with far larger profits and a stake in Sea inside it. Prosus sits 22% below its base and Uber 24% below. MercadoLibre costs 41 times 2025 net income against 29 for Sea, with its own margin under pressure.⁸⁷ So Sea isn’t unusually cheap on my pages. It is 20% below a base that is much harder to pin down than Tencent’s. At equal discounts I would want the steadier cash flows, and I’d want Sea only if it offered more.

How good the business is and how good the price is are two questions. How good is the business? Good, and strange. It has a real franchise in a growing region and a founder with a strong record, and it earns thin margins, depends on a single game and lends unsecured to people with short credit files. How good is the price? Reasonable on my numbers, and the market has reasons for it. This looks like a good business at a price that pays for part of what it could become, which is a different thing from a great business at a bargain.

• Price$74.24Close, Q4 2026 [1]
• Base value$93My estimate E (estimate)
• Margin of safety below$6035% under my base value C (calculated)

My rule for a discount to base value is 25% for a business with net cash and no customer above a tenth of revenue, 30% when one cycle or two customers drive results, and 35% when fixed charges such as leases sit ahead of the shares. Sea has no large customer and has net financial assets. But one cycle, credit, drives a large part of its results, and $4.3B of deposits sit ahead of the shares, which I treat like fixed charges. So I use 35%, which gives $60 C (calculated). The discount covers errors in my base inputs. It does not cover the bear: at $60 the bear value is 38% lower, and at today’s price 50% lower C (calculated). At $60 my base returns about 15.4% a year and the bear about 8.4% E (estimate). The price is 24% above that line, so the stock would have to fall about 19% to reach it C (calculated). So today’s price has a discount to my base, and it does not have my full margin of safety.

The price, and how it got here

Sea listed at a first close of $16.26 in October 2017, closed at $366.99 in October 2021 and at $40.50 at the end of 2023.¹ It has now lost more than 40% this year and about 80% from the top C (calculated).¹ The fall since August is the part that needs explaining, because nothing in the accounts has changed since then.

I can tie part of it to events. The DBS cut said investors feared a new period of subsidies after one-hour delivery launched in Indonesia and Thailand.¹⁸ The letter added two unknowns.¹⁶ MercadoLibre also fell 16% in those weeks, so some of this is about the group and not Sea alone.⁸ But most of the fall, from $131.51 to $88.47, came between 11 August and 19 October, when Sea filed nothing on its business, and the day before the letter the stock fell 5.7%. I could not confirm what drove all of it. Q3 results will test whether the market knew something.

Why the market may be right

I am below the market’s fears on some points and I may simply be wrong. Here is the case against my $93. Net income grew 10.6% on 48% revenue growth. Provisions rose 71.5%. Monee’s return fell by about a third in a year. Shopee has promised a good H2 and has to deliver it into a higher base and weaker currencies, and management said both of those on its own call.³⁶ Insiders sold $540M, and none bought.⁹ A founder with a record of big bets has just announced one, with no number on it.¹⁶ Every other margin debate about Sea has ended with management choosing growth. If you think the 2% never comes because rivals will not let it, $63 at 1.2% is the fair number, and that is 15% below the price E (estimate). The market doesn’t need to be stupid for $74.24 to be right.

Where this could be in 5 and 10 years

This is the same model, read at later dates. Value at the end of 2031 and 2036 is the present value then of the owner cash that comes after, at 11.5%, per share on today’s count and in that year’s dollars. It leaves out the $4.7B of net cash and investments I count today, about $7.50 a share. The owner cash column is the sum paid out from 2027 to 2036, and it is not discounted.

For my base to be right at five years, Shopee needs about $281B of GMV in 2030 and a margin near 1.6% by 2031, up from 0.66% today E (estimate). Owner cash has to reach about $3B a year by 2030. At ten years the base needs 2.0% on $458B, owner cash of $7.7B in 2036, and Monee loans of $55B earning about 6% E (estimate).

The bear is the margin that never comes. Shopee stops at 1.2%, Monee has a bad credit year in 2028, and Sea Cloud takes $1B a year. Its year-10 value is $58, under today’s price, even with $23 a share paid out on the way. The bull is management’s aim: 3.0% on $576B of GMV, loans of $73B and a second hit for Garena. Four numbers drive the spread: growth, margin, capital needed and the discount rate. Margin matters most.

The most likely way my base is wrong is that margin. If Shopee ends at 1.2% and not 2.0%, today’s base value is $63 and not $93 E (estimate). The year-10 figures also assume that in 2036 an owner still asks 11.5% and expects 3% growth after that.

These are estimates of value, not forecasts of the share price. The price can sit above or below value for years.

VIIThe case for the price

What most people miss

Sea’s fees went up by a quarter and sellers stayed. Core marketplace revenue went from 8.6% of GMV to 11.1% in a year while GMV kept growing 28% C (calculated).³⁴ Almost all of it went back out. So the thin margin shows a choice, and it also means the earning power is already in the top line.

The growth takes almost no capital. Sea rents its warehouses, and capital spending is 2.9% of revenue C (calculated).³⁶ That is rare among companies building a delivery network.

Sea has about $2.4B more equity in its loans than an 18% cushion needs E (estimate).³ That is about $3.75 a share, and it isn’t in my value. A Monee listing with its own funding could bring it into the open.

The bull case

In plain words: Shopee has won the region and has just shown it can charge sellers more, so its margin is a choice that can be reversed when rivals tire. Monee lends on data no outsider has, and its loans could be worth more as a listed lender than inside Sea. Garena pays for all of it while that happens. The people making this case most clearly are management. On the Q2 call they said 3% of GMV was within reach and the path to 2% relatively straightforward, and that fixed commissions will rise more slowly but ads have room.⁶ Forrest Li said in August that credit penetration remains low and that only a fraction of Sea’s users take its financial products.³ In October he said Monee could stand on its own and go public.¹⁶

Jianggan Li of Momentum Works, who follows the region closely, wrote on 22 October that the standalone apps and merchants outside Shopee make Monee more attractive as a separate company, and asked the sharper question of whether Shopee can be very profitable without it.¹⁹ He also said most analysts did not change their views after the letter.¹⁹ DBS, the one note I read in any detail, went the other way on 2 October, to a $105 target.¹⁸

Some large shareholders added in the first half. T. Rowe Price went from 20.5M shares in December to 35.6M at June, WCM from 25.9M to 33.9M and Viking Global from 3.7M to 5.7M. Baillie Gifford reported 37.2M ADSs in October, down from 41.0M.³³³⁴² Those are positions, and I can’t say what any of them think. I found no published view on this company from the well-known investors I usually check, and no ARK price model.

Against reality

The price needs less than my base. It needs Shopee at about 1.5% of GMV with the rest of my base intact, and today’s reading is 0.67%. If Shopee reaches about 1% by 2028, with GMV above 19% in 2027, loans near $17.6B and non-performing loans at 1.5% or less, then $93 is possible. The bull case of $167 needs 3% on GMV and 3.6 times today’s volume. Sea hasn’t shown 1% in any quarter yet. The best was 0.92% C (calculated).²⁶³ In some markets it is already above 3%, on management’s word, and I can’t check that.⁶

What a 10x would take. Ten times $74.24 is about $742, or $483B of equity on the share count at that price C (calculated). By 2036 that is about 26% a year C (calculated). At 25 times earnings it needs about $19B of net income, more than the $16.6B in my bull case E (estimate). On my bull tax and mix, that is about $670B of GMV at more than 3%, and a lender and a game on top. At today’s ratio of revenue to GMV, about 19%, that is roughly $128B of revenue at a net margin near 15%, against $27.7B and 6.1% today E (estimate). For companies this size it has been done: in the ten years to Q4 2026 NVIDIA rose 115 times from a public float near $29B, Tesla 30 times from $25B and Lilly 19 times from $77B C (calculated).³⁵ I chose those because they did it, so that list is all survivors, and each had one enormous product wave. MercadoLibre, the closest business, rose 9.7 times, from a float under $4B C (calculated).³⁵⁸ Sea did better than ten times once, from $16 to $367 in four years, then gave back about 80% C (calculated).¹ From here I think it is close to impossible by 2036 without a second business as large as Shopee. Sea Cloud is the only candidate named, and it has no numbers.

What they have to hit

Where I come down

The take rate moved me: it is the strongest evidence that the franchise can charge. The bull case’s 3% didn’t, because I can’t see it in any published figure. I would raise my values if Shopee reaches 1.0% of GMV in 2027 with growth above 19%, and if Sea gives Sea Cloud a budget under $1B a year. At 2.5% my base would be $111 E (estimate).

VIIIRisks, and what would change my view

1. Competition. TikTok Shop with Tokopedia at 65.7% of Shopee’s size, Pinduoduo’s entry on DBS’s account, MercadoLibre and others in Brazil.⁵¹⁸⁷ The 20-F says competition for share is particularly intense.²
2. Credit. Unsecured short loans of $11.1B, gross write-offs near 15% a year, and more new borrowers, on interest-free promotions.³²⁶ A change in write-off timing can move the ratio investors watch.¹³
3. Funding. $4.3B of deposits and $1.2B of borrowings against loans of about $10B on book. Asset-backed funding can come with guarantees and repurchase duties.³²
4. One game. Free Fire was a significant majority of Garena in 2025, and the 2021 to 2023 fall shows the size of the risk.²¹³
5. Sea Cloud and Migoo. No budget, no named customer, and direct competition with the largest companies in the world.¹⁶¹⁷
6. Governance. Dual-class control, Tencent’s proxy to the board, the founder on both key committees and no shareholder proposals at the 2026 meeting. If Li dies his Class B shares convert to Class A, so control ends with him and nobody has said what comes next.²²²
7. Currencies. Reported in dollars, earned in rupiah, baht, reais and more. Management raised this itself in August. Damodaran’s premiums are one measure; the actual loss in any one year could be much more.⁶³²
8. Regulation. Banking rules in several countries, data rules, and foreign ownership limits that Sea works around with VIE contracts. Indonesia is drafting a ceiling on the fees platforms charge sellers.²³⁶
9. Taiwan. A real market for both Shopee and Garena. The 20-F warns about cross-strait tensions.²
10. Tax. A rate near 34%, about twice Singapore’s, and $136M of unrecognized tax benefits on the balance sheet.³²
11. AI. An outside agent could take the shopping decision. Sea is on ChatGPT and also building its own, and I can’t size either.²⁰¹⁷
12. Tencent. It owns 17% and supplies games. Sea’s right of first refusal on Tencent’s games in six markets renews every two years, and was renewed in 2025. A sale of Tencent’s stake would weigh on the shares.²

What would raise my base value: Shopee at 1.0% of GMV in 2027 with GMV above 19%, a Sea Cloud budget near what I assume, Monee’s return holding above 10%, or a second game with staying power. What would lower it: Shopee staying under 0.6%, non-performing loans above 2.0%, Free Fire bookings falling two quarters running, or capital spending above $2B a year without contracts. The first check is Q3, expected mid-November.