- The Runway Ventures
- Posts
- 🤯 Watcha Collapse: The ₩49B Bond That Made Korea's Netflix Unsellable
🤯 Watcha Collapse: The ₩49B Bond That Made Korea's Netflix Unsellable
How Korea's AI-driven streaming pioneer peaked at 1.33M users, trapped itself in ₩49B of convertible bonds, and lost the content war to Netflix
Admond Lee
April 26, 2026
P.S. Brands like Notion, Airwallex, Deel, and Intercom sponsor The Runway Ventures to reach founders who actually convert.
Hey Founders,
Welcome to The Runway Ventures — a weekly newsletter where I deep dive into failed startup stories to help you become the top 1% founder by learning from their mistakes with actionable insights.
Today's story is about how Korea's indie AI-driven streaming pioneer got crushed by a convertible bond time bomb, failed acquisitions, and an unwinnable arms race against Netflix and chaebol-backed rivals. Let’s get to it! 🚀
Today at a Glance:
☠️ 1 Failed Startup → Watcha
⚠️ 2 Mistakes → Raising ₩49B convertible bonds at peak valuation
🧠 3 Lessons Learned → In streaming, the library is the moat
🔗 The Runway Insights → How to hire people who are better than you
💰 Southeast Asia Funding Radar → Salmon Group raises $100M to expand digital banking and lending operations across the Philippines through equity and bonds
☠️ 1 Failed Startup: Watcha
🚀 The Rise of Watcha
🇰🇷 Founded by Park Tae-hoon (CEO), Won Ji-hyun (COO), and Kang Seok-hoon (Business Director) in 2011, Watcha was a South Korean streaming and recommendation platform that used AI to serve personalised movies, dramas, books, music, and webtoons to niche audiences in Korea and Japan.
🕺🏻🕺🏻🕺🏻 Founders’ Story
Watcha started when Park Tae-hoon, a KAIST computer science student, got frustrated watching Korean media platforms push the same popular movies to everyone. No taste. No personalisation. Just rankings.
So he teamed up with two KAIST mates — Won Ji-hyun and Kang Seok-hoon — and built Watchapedia, a movie rating and recommendation engine powered by machine learning.
🎬 The bet was simple → if you could predict what someone would love based on their taste, you could own their entertainment life.
The Problem — 🎬 Korean media platforms in 2011 were generic. Everyone saw the same "popular" content.
Naver portal and box-office rankings dominated discovery
ML-driven recommendations barely existed in Korean commercial services
Diverse tastes (arthouse, foreign, LGBTQ+, anime) had nowhere to go
The Solution — 🧠 Watcha built an AI engine that predicted what you'd love based on what you'd rated.
Watchapedia (2012) became Korea's Letterboxd meets IMDb — accumulating rating data
Watcha Play (2016) turned that data into a personalised SVOD streaming service
A curated niche library (arthouse, classic, foreign, queer content) that global platforms ignored
🇰🇷🎬 In short, Watcha was the indie AI-driven OTT trying to beat Netflix not with scale, but with taste.
The one reason I started was that there are diverse people in the world, and I disliked that the world forces everyone to do the same thing.
 | It was not a grand dream to change the world. It was a complaint. And for almost a decade, that complaint looked like a business. 💰 K Cube Ventures (Kakao's early VC arm) bet on them first with an ₩800M seed in May 2012 — Watcha became their very first portfolio company. |
📺 In January 2016, Watcha launched Watcha Play — the SVOD streaming version of their data. Before Netflix had built real muscle in Korea. Before Tving, Wavve, and Coupang Play even existed.
💰 By December 2020, Watcha had raised a ₩36B (Series D) led by eBEST, Kakao Ventures, and Korea Development Bank — cumulative funding hit ~₩59B.
🎵 In August 2021, they acquired Vlending, MBC's music distribution arm. Park's new pitch: Watcha wasn't just an OTT — it was a "comprehensive content data company".
🏔️ At its peak, Watcha:
hit a ₩338 billion (~$257M) valuation via convertible bonds in October 2021
reached 1.33 million MAU in February 2022
posted ₩73.3B peak revenue in 2022 (COVID hit and everyone stayed home, doubling the viewing time)
offered a library of 90,000+ titles
accumulated 600-750 million rating data points
prepared to launch Watcha 2.0 — expanding into music and webtoons
In February 2022, Won Ji-hyun announced Watcha 2.0 as "a comprehensive entertainment subscription service". The team was gunning for an IPO that same year. Park publicly shut down any sale rumours.
The indie bet was working. Or so it seemed.
📉 The Fall of Watcha
💣 But behind the peak was a ticking time bomb.
In October 2021, at that peak ₩338B valuation, Watcha issued ₩49 billion of convertible bonds to Dunamu, Kakao Ventures, and Enlight Ventures. At the time, it looked like smart capital raising at premium terms.
It turned out to be the single decision that killed the company.
When the funding winter hit in 2022 and rates spiked, Watcha's ₩100B pre-IPO attempt at a ₩500B valuation got zero investor interest. The CBs suddenly became a ceiling no rescuer could swallow.
📌 Here’s what happened to Watcha:
🎬 From KAIST project to ₩338B valuation
21 Sep 2011 — 🌱 Frograms, Inc. founded in Seoul by three KAIST founders.
May 2012 — 💰 K Cube Ventures backed the team with ₩800M — Watcha became their first-ever portfolio company.
Aug 2012 — 🚀 Watchapedia beta launched — Korea's first AI-driven personalised movie recommendation service.
Jan 2016 — 📺 Watcha Play launched. The pivot from recommendation app to full streaming service.
21 Dec 2020 — 💰 Series D closed at ₩36B. cumulative funding reached ~₩59B.
Oct 2021 — ⚠️ Issued ₩49B convertible bonds at a ₩338B enterprise valuation. The debt anchor that would sink them.
22 Feb 2022 — 🎯 Watcha 2.0 announced.
IPO targeted for 2022. Sale rumours denied. Peak MAU ~1.33M.
💣 The ₩49B time bomb ticks down
1H 2022 — 📉 Pre-IPO attempt to raise ₩100B at a ₩500B valuation.
But zero investor interest as the market had repriced overnight.
28 Jul 2022 — ✂️ Restructuring confirmed: double-digit layoffs, halted content investment, Watcha 2.0 postponed, IPO shelved.
Sale talks opened with SKT, KT, Ridi, Kyobo, Krafton, Kakao, Coupang.
May 2023 — ❌ LG Uplus abruptly withdrew from an advanced acquisition. The ₩49B CB debt and financial-investor opposition killed the deal.
11 Oct 2023 — ⚖️ Watcha filed an antitrust complaint accusing LG Uplus of stealing its recommendation algorithms during due diligence.
Nov 2023 — 💣 Watcha defaulted on the 2021 CBs at maturity. No extension reached with creditors.
May 2025 — 📉 MAU collapsed to ~470,000 — down 64% from peak.
4 Aug 2025 — 🏛️ Seoul Bankruptcy Court formally approved receivership. Park Tae-hoon was designated as supervisor — no external administrator appointed.
24 Apr 2026 — 🪦 CJ ENM, the last serious bidder, withdrew from the court-led sale citing "financial and business viability".
Watcha's survival hangs by a thread.
The product was genuinely good. The recommendation engine was sophisticated. The curated library was loved by film lovers. The data, 750 million ratings, was real.
🩸 But an independent startup had taken on ₩49B of debt at a peak valuation in a market about to be dominated by conglomerates. Netflix kept pouring billions into Korean originals. Tving (CJ ENM/KT), Wavve (SKT), and Coupang Play all had deep-pocketed parents ready to absorb losses for years.
Watcha had a bond that made it too expensive to rescue and too insolvent to stand alone.
In the end, the math didn't care about taste.
Want to learn more about Watcha’s downfall?
⚠️ 2 Mistakes
Park Tae-hoon (Founder and CEO of Watcha)
Mistake 1: Raising ₩49B convertible bonds at peak valuation
In October 2021, Watcha chose to raise ₩49B as convertible bonds from Dunamu, Kakao Ventures, and Enlight Ventures — pegged to the peak ₩338B valuation. At the time, this looked smart → cash in at a premium during the pandemic streaming boom, avoid dilution, IPO next year.
🤦🏻♂️🚨 But CBs aren't clean equity. They carry a ceiling.
The contract referenced that ₩338B peak with penalty interest up to 15%. Meaning any future rescue (down-round, acquisition etc. had to clear that number first.
🤫 When the tech winter hit in 2022 and the ₩100B pre-IPO at ₩500B attracted zero investors, the CB became a trap:
May 2023 → LG Uplus walked because existing shareholders refused to cede management at a price below the CB reference
Nov 2023 → Watcha defaulted at maturity, no extension reached with creditors
Jul 2025 → Enlight Ventures filed for court receivership over unpaid CB principal
Apr 2026 → CJ ENM, the last serious bidder, withdrew citing "financial viability"
Every rescue died the same death. Nobody would clear that ceiling.
💸 A fleeting market peak had been converted into a permanent financial anchor — in a capital-intensive industry about to reprice violently.
Mistake 2: Treating personalisation as a moat in a content war
🙅🏻♂️ As early as 2017, Park Tae-hoon publicly framed Watcha as anti-Netflix:
Netflix is shifting its focus into self-produced original series, while Watcha Play focuses on providing personally-recommended video lists and diverse lineups.
Park genuinely believed personalisation and a 750M rating dataset would be Watcha's durable moat in SVOD.
🤔 That was a category error.
In subscription video, the real moat is content ownership and exclusive originals. Users go where the hit shows are and tolerate whatever UX surrounds them. You can have the best recommendation engine in the world — if you don't have Squid Game, people don't subscribe.
While Watcha kept polishing curation and ratings infrastructure:
Netflix poured billions into Korean originals
Tving (CJ ENM) and Wavve (SKT) tapped their parents' content libraries
Coupang Play bought sports rights and tentpole K-dramas
🥲 The scoreboard by early 2023 → Netflix 11.6M Korean subscribers. Tving/Wavve 7-8M. Coupang Play 6-7M. Watcha never crossed 1.5M.
In 2022, Watcha burned ₩55.5B on ₩73.3B revenue (a 76% operating loss ratio) because they were outspent in the only battle that actually mattered.
The recommendation engine was genuinely good. In SVOD though, a better algorithm never beats a better library.
🧠 3 Lessons Learned
Lesson 1: Be careful of fundraising with convertible bonds
A convertible bond at peak valuation creates a permanent floor.
Every future rescuer must clear it. Watcha learned this when the market repriced — the ₩49B CB became untouchable, and every acquisition attempt (LG Uplus, CJ ENM) died at that ceiling.
🌮 Key Takeaways:
Peak-valuation debt locks in a price every future rescuer must clear, even years later.
Convertible bonds carry hidden ceilings that penalise every down-round and acquisition attempt.
Clean equity at a defensible price beats premium debt at a fragile peak.
🛠️ Operator Playbook:
🎯 Stress-test CB terms before signing (not after default)
Model three scenarios: bull (IPO), base (down-round), bear (fire-sale). If your CB terms make the bear case unworkable, renegotiate or walk.
If the conversion math at a 50% down-round already wipes out founder equity, you've signed a trap — one that just hasn't been triggered yet.
🧱 Cap CB conversion at a valuation FLOOR (not just a ceiling)
Without a floor, CB holders become the most powerful veto in any M&A talk. Watcha learned this when LG Uplus walked in May 2023, blocked by financial-investor opposition to any sub-₩338B exit.
Build explicit consent rights for down-round acquisitions at pre-agreed discounts; silent veto power is the default you have to negotiate away.
💰 Earmark 30%+ of any frothy-market raise for a debt service reserve
Watcha spent its ₩49B on Vlending, originals, and Watcha 2.0, leaving zero cushion for 2023 refinancing.
Penalty interest in a rising-rate environment (Watcha faced up to 15% annual) converts a solvable maturity into a terminal one within 12 months.
Lesson 2: Read the funding cycle before expanding scope
In late 2021, with a tech winter forming on the horizon, Watcha did the opposite of what survival required.
They acquired Vlending, raised ₩49B in debt, and announced Watcha 2.0 — music and webtoons. Within 6 months, every plan was shelved — the capital was already burned.
🌮 Key Takeaways:
The hardest signal to read is the one that says "stop, narrow, defend".
Adjacent-category expansion requires proven core economics before launch.
A funding winter rewards focus on what already works.
🛠️ Operator Playbook:
📊 Run a quarterly "cycle posture" review
Track 3 macro signals: recent comparable raises, interest rate trajectory, public market multiples in your category.
If 2 of 3 trend negative, shift from growth posture to durability posture within the same quarter.
Netflix paused content spending in 2022 and refocused on retention before resuming aggressive investment in 2023 — that's reading the cycle.
🧊 Apply the "freeze test" before any new product line
Ask: "If new capital disappeared for 18 months, would we still launch this?"
Model the standalone 18-month burn — team, content, marketing, infrastructure — using only existing cash.
If running the new line would force layoffs in the core, kill the launch.
Lesson 3: In streaming, the library is the moat
📺 From 2017, Park Tae-hoon publicly framed Watcha against Netflix as personalisation versus originals, betting a 750M-rating dataset would be the durable moat.
The market voted differently. Netflix reached 11.6M Korean subscribers while Watcha never crossed 1.5M because subscribers go where the hit shows are.
🌮 Key Takeaways:
In subscription video, the moat is content ownership — algorithms are just features.
A better algorithm cannot offset a smaller library when subscribers chase exclusive hits.
Distinctive curation builds niche loyalty but rarely drives the mass-market growth that funds originals.
🛠️ Operator Playbook:
🎬 Audit your moat against the actual buying decision
Survey new subscriber cohorts: "What made you sign up this month?"
If <20% mention recommendations, your moat lies elsewhere.
Benchmark content spend as % of revenue against category leaders. Netflix spent ~50% at peak — that's what kept subscribers locked in.
📚 Pick a moat that compounds with subscriber scale
Content ownership compounds → the same hit episode amortises across 11M subs vs 1M subs at the same production cost.
Recommendation quality plateaus fast. You don't need 750M ratings to recommend well — you probably need maybe 50M.
Coupang Play won by buying content nobody else had — exclusive K-League and EPL Korean broadcasting rights — instead of competing on algorithms.
🔗 The Runway Insights
How to hire people who are better than you (Read)
How to compete with legacy incumbents (Read)
How to angel invest (even if you have no money) (Read)
VC in 2026: 75% of all the money is going to just 5 VC funds. And to just 5 "startups." (Read)
Surviving AI price wars without destroying your business (Read)
💰 Southeast Asia Funding Radar
Salmon Group raises $100M to expand digital banking and lending operations across the Philippines through equity and bonds (More)
PvX Partners secures $10.5M (Series A) to scale cohort-based user acquisition financing platform for consumer and mobile gaming apps (More)
Kitar bags $10M (Pre-Series A) to grow secondhand mobile phone recommerce platform advancing circular economy across SEA (More)
Muun AI closes $700K (Pre-Seed) to scale industrial AI platform converting machine telemetry into real-time operational intelligence for factories (More)
💃 Rate Today’s Edition
What'd you think of today's edition?Your feedback helps me create better content for you! |
That’s all for today
Thanks for reading. I hope you enjoyed today's issue. More than that, I hope you’ve learned some actionable tips to build and grow your business.
You can always write to me by simply replying to this newsletter and we can chat.
See you again next week.
- Admond
Disclaimer: The Runway Ventures content is for informational purposes only. Unless otherwise stated, any opinions expressed above belong solely to the author.
Reply