Sea Limited: Building the Next Amazon?

ELI5/TLDR

Sea Limited is a Singaporean conglomerate that runs Shopee (Southeast Asia’s largest e-commerce marketplace), SeaMoney (a buy-now-pay-later and consumer lending business), and Garena (a gaming publisher with one blockbuster, Free Fire). The stock has been a rollercoaster — up 2,100%, down 90%, up 400%, now down 55% — because the market keeps re-rating how durable its profitability is. Drew Cohen’s pitch is that the three segments now do roughly $2.4B EBIT combined, the loan book is compounding at 80%, and TikTok Shop is the new bear case. He builds a 10-year reverse Fermi valuation that pencils out to a 17–22% IRR if you believe Shopee can hit 1.5% EBIT-on-GMV at maturity and the lending book scales to ~$83B.

The Full Story

What you’re actually buying

Sea is three businesses stitched together. $22.9B in revenue, 9% operating margin, ~$2B in EBIT. Until 2022 they were burning money every year. They flipped profitable in 2023 after Forrest Li took the axe to a global expansion (France, Poland, India, Argentina, Colombia, Chile, Mexico — all exited), suspended executive cash comp, capped travel, and forced the org to make money. Last year they grew 30%. The crash from $360 to $35 in 2023 was the bear thesis vindicated; the run back up was the pivot working; the current 55% drawdown is fear that TikTok Shop forces them back into subsidy mode.

The segments break out roughly like this:

Shopee (e-commerce): $14.5B service revenue + $2B 1P goods, ~$600M segment EBIT. The big number, the smallest margin.
SeaMoney (fintech): $3.8B revenue, ~$970M EBIT, growing 60%. The hidden compounder.
Garena (gaming): $2.4B revenue, $1.4B EBIT. The cash machine that funded the other two.

Worth pausing on the segment EBIT mix: gaming throws off more profit than e-commerce and fintech almost combined. The “e-commerce story” investors buy is partially funded by a single mobile game.

Shopee — the moat is logistics, not the app

Shopee has 400M buyers, 20M sellers, $127B GMV across seven Southeast Asian markets plus Brazil. They started in 2015 with women’s fashion (Forrest Li’s bet on young women as taste-makers), launched in seven countries simultaneously, and ran the playbook: third-party marketplace first, then layer first-party FMCG (diapers, toothpaste, detergent) on top to drive habituation.

“It’s a high-frequency item, something you’re ordering often all the time. And then it also helps kind of seed more selection.”

The unit economics are what investors fight about. Average order value is low. Southeast Asia is geographically diffuse — thousands of islands, packages literally moving by boat. Incomes are below Korea or even Brazil. Shippers are coupon-sensitive. The bear case is that you can’t make money shipping cheap goods to poor people across an archipelago.

The bull case rests on three things actually working:

SPX Express, their in-house logistics arm, now handles >50% of packages in Southeast Asia and ~70% in Brazil. Owning the lane kills the margin a 3PL would have taken. It also unlocks upstream sorting and consolidation — the kind of efficiencies that compound at scale.
High-intentionality habituation. Shopee is moving from gamified discovery (countdown timers, coupon games, compulsive scrolling) toward Amazon-style trusted utility. The VIP free-shipping program has 7M members — small versus 400M buyers, but those members spend 30–40% more.
Take rate expansion. Commissions run 8–18% by category, averaging 13%. They don’t really make money on shipping yet. The real profit lever is advertising — same flywheel as Amazon, where ads are most of the e-commerce profit. Drew thinks ads are already the bulk of Shopee’s margin.

The GMV asterisk

A lot of GMV in Asian e-commerce is fake. Drew traced this back to JD.com, which used to publish two GMV numbers and quietly killed the tighter one because the media kept reporting the smaller figure against Alibaba’s looser one. The mechanism is “brushing” — merchants ship items to themselves or to fake accounts to inflate reviews and ranking.

When Drew did this analysis on Alibaba years ago, ~40% of GMV looked uneconomic. He thinks Shopee is much cleaner now (platforms have wised up, enforcement is tighter), but he uses 30% as a haircut for valuation purposes — i.e., real GMV is closer to $90B than the reported $127B. This matters because every “EBIT as % of GMV” comparison falls apart if your denominator is fake.

SeaMoney — the underwritten compounder

This is where the equity-research nerd in you should perk up. $9.2B loan book, growing 80% YoY. $970M segment EBIT on $3.8B revenue. They just got bank licenses in Singapore, Malaysia, Indonesia, Philippines (Thailand pending) plus Brazil — meaning they can take deposits and use them as a cheaper funding source.

The product mix is mostly:

Shopee Pay Later (BNPL at checkout, partly merchant-funded — merchant might receive $96 on a $100 transaction)
Personal loans to existing money customers
Merchant loans for working capital

Three risk-management features Drew highlights:

Short duration. Most loans are 3–6 months, almost all under a year. If your underwriting is broken, you find out fast. Compare to a 30-year mortgage where you don’t know for years.
Hostage leverage. If a borrower defaults, Shopee can cut their access to the marketplace. Strong incentive not to skip your $50 BNPL bill if it costs you the place you buy everything.
Spread cushion. Drew estimates net interest margin after losses around 25%, similar to MercadoPago. Disclosures are worse than MELI’s — they don’t give you NIM-after-losses directly — but you can back into it.

The valuation Drew runs is striking. Right now the lending book penetrates ~6% of real GMV. He grows that to 15%, holds off-platform lending flat at 40% of the book, and rolls the e-commerce assumption forward 10 years. Loan book gets to $83B. Apply 25% NIM, 35% margin (loss provision baked in), and SeaMoney throws off $7.3B in pre-tax profit a decade out — bigger than his Shopee number.

The risk frame is honest. They haven’t lived through a real Southeast Asian credit cycle. COVID was a weird stress test because e-commerce merchants kept paying to preserve marketplace access. If interest rates get capped harder by regulators, or if competitors crowd in and compress spreads to developed-market levels, the math compresses fast.

Garena — one game, surprising durability

Sea started life as a game distributor for internet cafes. The Tencent partnership was the unlock. They bought a Vietnamese studio in 2017 that had just made Free Fire — a battle-royale shooter designed for slow phones and bad bandwidth, perfect for the region. It became a multi-billion-dollar franchise.

The volatility is real: Garena did $4.3B revenue in 2021, fell to $1.9B in 2023 (down ~50%), then stabilized and is now growing 26% again. Free Fire is still essentially the only IP they own, but Drew makes the point that a single durable free-to-play title is actually safer than a studio that has to launch a new hit every year — there’s no relaunch cycle, just continuous content updates and cosmetic monetization.

The publisher side (Tencent and EA Games — they recently won FC Mobile) is much lower margin: maybe 20–30% versus ~70% on Free Fire. Two real risks here: Tencent’s Level Infinite is increasingly self-publishing (they took League of Legends in-house via Riot), and India banned Free Fire — which used to be one of the biggest markets.

Competition: TikTok Shop is the bear case

E-commerce competition is mostly settled. Lazada (Alibaba) never localized properly and bled share. Tokopedia got absorbed into GoTo and then 75% sold to TikTok. So the field is Shopee versus TikTok Shop.

The frame Drew uses is intentionality versus discovery:

“When you do have high intentionality, that means I have this idea, I need to buy more laundry detergent. I’m going to just go into the app, type into the search bar, hit buy.”

That’s Shopee’s territory. TikTok is discovery commerce — you weren’t planning to buy a $10 t-shirt, you saw a livestream, you bought it, you’ll wait a week, you’ll return it if it’s bad. Different consumer expectation, different shipping requirements, different category mix. Drew’s view is TikTok Shop is more competitive with Instagram than with Shopee, but it’s clearly stealing some of Shopee’s old “browse and buy” budget.

The numbers: TikTok Shop is doing maybe $50B GMV in Southeast Asia, growing faster than Shopee’s $125B. The flat-profit guidance for next year is the company telling you they’ll spend to defend share. That’s the immediate trigger for the 55% drawdown.

Fintech competition is the opposite — fragmented and crowded. Grab regionally, plus Dana (Indonesia), GCash (Philippines), TrueMoney (Thailand), MoMo and Zalo (Vietnam), TouchAndGo (Malaysia), MercadoPago and NuBank (Brazil). SeaMoney is #3 or #4 in most of these. Drew’s read is that they’re not chasing leadership — they’re using Shopee as a customer acquisition funnel and monetizing the cohort cost-effectively. That’s actually a healthier strategy than burning capital on standalone wallet share.

The 10-year math

Drew runs a Fermi-style reverse valuation. The assumptions are explicit and worth replaying because that’s the whole exercise — change them, see what falls out.

Shopee 10 years out:

Real GMV $90B today → $330B in 10 years (e-commerce penetration 20%→45%, retail spend 5%/yr, flat market share)
1.5% EBIT/GMV at maturity (low end of management’s 2–3% EBITDA target, conservative; Amazon NA does 6%, China comps do ~2% even in brutal competition)
→ $5B EBIT

SeaMoney 10 years out:

Loan book $9.2B → $83B (penetration 6%→15% of real GMV, off-platform held flat at 40% of book)
25% NIM, 35% margin (losses absorbed)
→ $7.3B pre-tax profit

Garena 10 years out:

5% growth (in line with retail spend; no new hit assumed)
→ $2B EBIT

Sum: $14.3B segment profit, minus $1B unallocated overhead, taxed at 23% → $10.2B net earnings in year 10.

Apply a 20–30x multiple → $205–305B market cap. Off today’s $43B market cap (after backing out $10B net cash from $53B EV), that’s a 17–22% annualized IRR.

This isn’t a buy/sell call — it’s a frame. The work is in sensitizing. What if Shopee does 1% instead of 1.5%? What if TikTok Shop pulls 5 points of share a year? What if a credit event takes 30% of the loan book? What if Tencent self-publishes more of Garena’s catalogue? The framework lets you put numbers on each scenario.

“If I’m giving you very aggressive math and the return is low, it’s very easy to pass. If I’m giving you aggressive math and the returns kind of sound decent to you, then maybe you want to see, okay, well, is there a margin of safety?”

Key Takeaways

Sea is a 3-segment business: Shopee (~~$600M EBIT), SeaMoney (~~$970M EBIT, fastest-growing), Garena (~$1.4B EBIT, surprisingly durable cash machine).
The pivot to profitability in 2023 — exit unprofitable markets, suspend exec cash comp, cap travel — is the founder’s playbook, and it worked.
Shopee’s moat is increasingly logistics (SPX Express does >50% of SEA packages, ~70% of Brazil) and ad monetization, not the gamified app surface.
~30% of reported GMV may be brushing/refund inflation; haircut the $127B headline to ~$90B for valuation work.
SeaMoney is the under-appreciated leg: $9.2B book at 80% growth, ~25% NIM-after-losses, short-duration loans (3–6 months) with marketplace-access leverage on borrowers.
Garena throws off more profit than the other two segments combined and is mostly Free Fire — a single 10-year-old free-to-play title that keeps compounding via cosmetic monetization.
TikTok Shop is the live bear case. Estimated $50B SEA GMV growing faster than Shopee. Flat next-year profit guide is management telling you they’ll spend to defend share.
The fintech business is #3 or #4 in most markets — that’s fine; the strategy is monetizing the Shopee funnel, not chasing wallet leadership.
10-year math at conservative assumptions (1.5% EBIT/GMV, 15% lending penetration, 25% NIM) implies $10.2B earnings → 17–22% IRR off current price.
Real risks to underwrite: Southeast Asian credit cycle (untested), regulatory rate caps, Tencent self-publishing, India-style market bans, TikTok subsidy war duration.

Claude’s Take

This is an unusually well-built equity walkthrough. Drew worked at Goldman investment research and Capital Group on the buy-side, sat in meetings with Forrest Li, and it shows — the segment-by-segment unpack is what you’d want from a real research note, not a YouTube essay. The GMV-haircut sidebar alone is the kind of thing that separates analysts who read footnotes from analysts who don’t. He gets the JD.com history right, applies the framework to Shopee, and uses the conservative number throughout. That’s craft.

The 10-year Fermi valuation is the right framework for a business that hasn’t fully monetized. The honest move is that he tells you the assumptions are wrong — he just wants you to see how the math behaves under each lever. The 1.5% EBIT/GMV at maturity is genuinely conservative; if ad revenue scales the way it has at Amazon, 3% is plausible and the math becomes obviously attractive. The risk is the other direction: Southeast Asia is structurally lower-AOV, lower-income, more coupon-sensitive than the US, so 1.5% might also be optimistic. The honest band is probably 0.75–3%, and within that band returns swing wildly.

What I’d push back on: he’s somewhat dismissive of TikTok Shop (“different consumer behavior, more competitive with Instagram than Shopee”). That’s plausible for laundry detergent, less plausible for the impulse-buy fashion and beauty categories that were Shopee’s foundation. The flat-profit guide is management telling you the threat is real enough to spend on. And the Garena assumption (5% growth, no new hit) is conservative on direction but the segment is one banned market or one Tencent in-housing decision away from looking very different — Free Fire is one game.

The real insight buried in here is that SeaMoney might be the actual story. A $9.2B short-duration loan book growing 80% with marketplace-access enforcement and 25% spreads is a remarkable structure if the credit cycle holds. If you believe that, you’re not really buying an e-commerce stock — you’re buying a fintech with an embedded customer acquisition machine. That reframe is worth more than the headline IRR number.

8/10 — high information density, honest valuation framework, primary-source familiarity with management. Loses a point for under-weighting the TikTok Shop threat in the qualitative section even after acknowledging it in the price action.