Part IThe Story
The bell rang in New York in the middle of July 2016, and the man most responsible for it was not there to hear it. Line Corporation, the Tokyo-based messaging company that had grown out of a Korean search portal's Japanese outpost, was listing simultaneously on the New York and Tokyo stock exchanges. No overseas subsidiary of a South Korean company had ever done that. On its first day the shares closed at $41.58 in New York and 4,345 yen in Tokyo, 27 and 32 percent above their offering prices. The market had been expected to value the company at roughly 6.5 trillion won. It went to 10 trillion. The offering put about 1.5 trillion won of fresh capital into Line's accounts, a sum its chairman would describe, with the flat candor of a man reading off a balance sheet, as "the first time we've had some capital leeway since we founded Line."
That chairman was Lee Hae-jin, founder of Naver, and on the day of the listing he was in Chuncheon, in Gangwon Province, at a building called GAK.
GAK is a data center. Naver finished it in June 2013, making it the first Korean internet company to build and run its own. It holds no trading floor and does not photograph well. Lee had arranged to be there because Naver's press corps happened to be touring the facility that week, and one of the few reporters who has written thoughtfully about him, at DongA Science, guessed that he had meant to stand before the media "proudly (?)," the question mark doing a great deal of work, with Line's success as a tangible thing to point to. He told the reporters he had watched the opening bell on television and felt something powerful. Then he started talking about what scared him.
"I'm still afraid of those huge American Internet businesses, like YouTube in video services, Facebook in social networking, Instagram in photo-sharing, and so on," he said. "Every year they record sales of tens of trillions of Korean won with profits also in the trillions. Whenever I think about how we can survive in competition with them, I get nervous and worried. They say Naver is a dinosaur — compare it to Google."
The setting matters. A founder who wanted the cameras could have flown to Manhattan; Line's executives were there. Lee chose a warehouse of servers in a provincial city, a building whose whole purpose is to keep Korean data on Korean soil, under Korean control, out of reach of whatever might arrive one morning from California. He watched his triumph secondhand, on a screen, inside a vault. Then he described it as a reprieve.
Every important thing about Lee Hae-jin is already in that scene: the fear, the absence, the fixation on infrastructure, the arithmetic of a small country against large ones, the wish to be judged by what he built rather than by where he stood. Over the next decade the scene would repeat in new costumes. A parliamentary hearing room. A Fair Trade Commission office he walked into on his own, to argue that he did not own his own company. A hotel ballroom in Palo Alto where, at 57, he compared the global internet to a coffee chain. The man who would not ring the bell kept turning up in rooms where he had to answer for it.
By the Numbers
The Navigator's Ledger
₩500MSeed capital for Naver.com as an in-house venture inside Samsung SDS, 1997
200M+Answers accumulated on Knowledge iN, the Q&A service Naver launched in October 2002
2×Naver's search volume relative to Google in South Korea, 2025
₩12.04TNaver Corporation revenue in 2025 (about US$10.52 billion)
~3%Lee's own description of his stake when he refused the title of 'owner' in 2019
7 yearsTime between Lee leaving Naver's board in 2018 and returning as chairman in March 2025
67Languages supported by Naver Dictionary in 2024, up from two at launch in 1999
Two Hours Out of Eight
The official biography fits in a paragraph, which suits its subject. Born in Seoul on June 22, 1967. A bachelor's degree in computer engineering from Seoul National University, then a master's in computer science from KAIST, the state-funded science institute that was minting Korea's engineering class. One account says that before his first real job he had helped build an information-retrieval system for the national intelligence service, so the young engineer's first serious work on search was done for spies. In 1992 he joined Samsung SDS, the systems-integration arm of the country's largest conglomerate, as a research engineer.
What he did there is the first durable piece of the Lee Hae-jin method, and it is close to an act of quiet embezzlement. He called it the 25 percent rule. Of every eight-hour working day he took at least two hours and spent them on new technology unrelated to the work Samsung was paying him for. He kept this up for about three years. At the end of it he had decided what the two hours were for: a search engine.
In 1994 he recruited three new hires at the company and formed a small search-development team. Accounts differ on details, as accounts of founding myths do, but the shape holds: in 1997 his proposal for a basic search service won an internal competition at Samsung SDS, and the company spun up an in-house venture around it with 500 million won in capital. The project was called Naver.com, a compression of navigate and the agentive suffix -er, the company later explained, meaning "a person who navigates the vast ocean of information on the Internet." In June 1999 the venture left Samsung and became an independent company, Naver Comm, with Lee in control. Its founding date is June 2, 1999.
Decades later Lee would tell this story in a way that is both moving and slightly wrong. "I was a new hire when Google first unveiled its search engine," he said in Palo Alto in June 2025, "and I thought back then that the information gap was going to lead to a gap in competitiveness — that is why I decided to develop our own Korean-language search engine." Google was incorporated in 1998. By then Lee had been at Samsung for six years and running his search team for four. Memory compresses. Every founder's does. What is telling is the shape it compresses into. In Lee's retelling, Naver begins as a response to an American giant, a defensive crouch, before there was a giant to crouch against. Fear came first in his memory, ahead of the facts.
He went further in the same speech, reaching back past Samsung and KAIST to the fifteenth century. He spoke of a "greater sense of responsibility," because "unlike how those with power in the past sometimes tried to monopolize information by seizing control of the language or burning books, Korea has a meaningful history of being the first in the world to invent metal type printing and creating Hunminjeongeum" — the 1446 royal manuscript that introduced the Hangul alphabet so that, as the old preface has it, ordinary people could read. The Korean founder of a search engine had placed his company in a line running back to King Sejong. It is an extraordinary thing to say at a venture-capital networking event. He said it without irony, and the 200 investors in the room seem to have taken it the same way.
A second Lee shows up in the record next to the mythologizer. He is precise and methodical, by every Korean account meticulous to the point of fussiness. Business Post's profile describes him as kkomkkomhan, painstakingly thorough and outwardly calm, yet bold when a business decision arrives, unafraid of adventure. The two hours a day fits both. Taking them was a gamble against the employer. Keeping them up for three years took a ledger-keeper's discipline.
A Country Answering Its Own Questions
The problem Naver faced in its first three years was not technical. The early Korean web was thin. There simply wasn't much written in Korean online, and a search engine with nothing to search is a very expensive mirror. Lee's rivals, domestic portals like Daum and the Korean arms of Yahoo, faced the same scarcity. Google, when it arrived, faced it worse: its algorithm was magnificent at ranking documents that existed, and in Korean too few of them did.
Naver's answers came in two pieces, two years apart. The first was presentation. In August 2000 it launched what it called comprehensive search: one query returned results grouped by type on a single page, websites, images, community posts, news, each in its own band. Naver calls it the world's first integrated search. Whether or not that claim survives every footnote, the instinct behind it lasted. A Korean user did not want ten blue links to a sparse web. She wanted Naver to have already sorted the sparseness for her.
The second piece made the company. In October 2002 Naver launched a service first called Knowledge Search and soon renamed Knowledge iN. Anyone could post a question on any subject. Other users answered. The asker chose the best answer, and points went to the people who wrote the most helpful ones. If the Korean web had no documents, Naver would get Koreans to write them, one question at a time, and then index what they wrote.
It worked on a scale that is hard to take in. By May 2016 Knowledge iN held 100 million questions and 200 million answers. An "Ask an Expert" tier added licensed doctors, veterinarians, pharmacists, lawyers, tax accountants and labor attorneys. Three years after Knowledge iN launched, Yahoo introduced Yahoo! Answers, and Bradley Horowitz, then Yahoo's vice president of product strategy, cited the Korean service as the inspiration. That is the rare case of Silicon Valley copying Seoul and saying so.
The service drew criticism. A 2007 New York Times report noted that Naver tolerated loosely sourced, sometimes unauthorized republishing to keep the answers flowing, and that the quality fell short of Wikipedia's more heavily moderated pages. The criticism is fair, and Lee never really argued with it, because the point of Knowledge iN was never encyclopedic perfection. The point was a body of Korean-language data that no outside crawler could reach, because it lived on Naver's servers and nowhere else. Naver Blog followed in 2003, launched under the name "paper" in June and renamed in October. Naver Cafe, which let anyone start an online community, would host 10.5 million active cafes by May 2017. Webtoon began in 2004. Each of them turned users into a source of supply.
Lee laid out the logic plainly in 2025, and it is worth quoting at length because he was not really talking about 2002:
"In the early days of search engines, many of them competed against each other using their own algorithms. However, with the advancement of technology and training on different search engines, they all began to have similar algorithms. Ultimately, data became the key differentiator in this competition. Similarly, I believe that AI will also progress in a similar manner."
So Knowledge iN was the first draft of an argument he would still be making twenty-three years later, about large language models. Algorithms converge. Data does not, and whoever holds the data holds the moat. He would come back to it in Palo Alto, at the AI Seoul Summit, and at a shareholders' meeting in the Green Factory, with nothing changed but the nouns.
There is a smaller irony buried in the founding years. Naver's first commercial breakthrough in search advertising, a 2001 product called "Plus Site Registration" and billed as Korea's first search ad, depended on exactly the closed garden that Knowledge iN was building. Answers kept users on Naver. Users on Naver saw Naver's ads. In 2017 a Korean lawmaker would accuse Naver of running ads at the top of results without labeling them clearly as ads, and Lee would reply that every search engine in the world does the same thing, Google included. He was right, and it did not help him.
The Man Who Gave Away the Microphone
In July 2000, a year after independence, Naver merged with Hangame Communications, an online game portal, along with several smaller firms including Oneque and Search Solutions. Hangame's founder was Kim Beom-soo, who, like Lee, had been through Seoul National University and Samsung SDS. They were friends. In 2001 the merged company took the name NHN, for Next Human Network, though insiders also read it as Naver, Hangame, Network. Each brand kept its own name. In October 2002, the month Knowledge iN launched, NHN listed on the KOSDAQ.
Then Lee did something few Korean founders of that generation would have done. A Hong Kong profile records that after the listing he admitted openly that he was not good at socializing and did not play golf well, and so he handed the chief executive's job to Kim. It is worth stopping on the golf. In Korean business of that era the golf course was where deals with officials, bankers and conglomerate heirs actually got made, and a CEO who could not work a fairway was missing a whole branch of the job. Lee looked at that job, decided it was not his, and gave away the microphone. By 2004 he had stepped down as an NHN chief executive altogether, keeping the board chairmanship he had held since 1999 and later the title of chief strategy officer.
The partnership worked for a while. Hangame threw off cash, Naver threw off traffic, and by 2008 NHN had the largest market capitalization on the KOSDAQ, made the Forbes Global 2000 for the first time, and moved that November to the main KOSPI exchange. Kim Beom-soo left in 2007 "to do his own thing," as one Korea Times account put it. Three years later, in March 2010, he launched KakaoTalk.
It is hard to invent a more novelistic rivalry. Two Samsung SDS alumni from the same university, co-founders of the same company, part ways, and within four years they are fighting over the same thing, the messaging app on every Korean phone. In 2012 Kim's KakaoTalk had 53 million subscribers and about 200 employees, and was not yet profitable. Lee's NHN had 3,600 employees and 2 trillion won in annual revenue, and an app called Line that had reached 50 million users in half the time it took Kakao. But the users were in different places. Under 20 percent of Kakao's were overseas. More than 90 percent of Line's were.
Kim had won Korea. Lee had lost Korea and gone abroad, and the reasons why are the hinge of the whole career. A man who was bad at golf, who avoided the press, who by his own account was not good with people, had built the dominant portal in a country famous for the dominance of personal networks. Then the Korean market, at the moment of its greatest technological shift, chose his old friend's product over his. He took his best people and his capital and went to the one place the friend wasn't. Later Lee would put it in strategic terms, as urgency and the smallness of the home market. Those reasons were real. They also sound like what a man says when he has been beaten at home.
Scared Every Morning
Lee's most-repeated word in public, by DongA Science's careful count across press conferences, is some variant of afraid. "I'm scared," he says, or "I feel desperate." Not once or twice, the reporter notes. Constantly.
— Lee Hae-jin, as quoted by DongA Science, 2016For the past 17 years, I've been scared every morning. I was scared that a new technology or a new service would emerge from the United States. Recently, I'm scared of Chinese companies that are growing to an enormous scale with the protection of their country and entering the global market. I'm scared whether we can survive the competition with them.
The reporter's follow-up question is the obvious one: what does a man worth more than a trillion won have to be afraid of? The honest answer seems to be: everything that happened to everyone else. In 2012, when Lee gave an internal lecture to NHN staff, the mobile revolution was killing incumbents in public. Nokia, Motorola and Nintendo, three companies that had looked unassailable five years earlier, were collapsing, as the DongA piece puts it, "in an instant." NHN's portal had been built for the desktop. It was not obvious that it would survive the phone.
At that lecture Lee lost his temper, as far as a man like Lee loses his temper, at his own employees. According to the Korea Economic Daily, he had read a post on an internal board in which someone wrote, "I came to NHN to live comfortably after working at Samsung." "I was so dumbfounded and my heart sank," he said. Too many employees, he went on, treated NHN like "some 'neighborhood recreational soccer club.'"
The line about the soccer club has the ring of something said once and remembered for years by everyone who heard it. It also catches the founder's predicament at a particular point in a company's life. A man who had taken two hours a day from Samsung to build something of his own was now running a company that people joined to escape Samsung's pressure. Without meaning to, he had built a refuge, and he could not stand it.
The fear also sat oddly with the facts. By the time he said "17 years," Naver had beaten Yahoo Korea, held Google off, and outlasted Daum's lead as the portal of record. In September 2023 it would still hold 59 percent of Korean search to Google's 31, against a rival with forty times its revenue. In 2025, by one count, it handled twice as many Korean searches as Google. Choi Soo-yeon, the CEO Lee's board appointed in 2022, would tell Fortune in 2024 that Naver was perhaps "the only company in the world that has survived competition against Google and Amazon." No other company the Google of its time came for has ever been able to say that.
So the fear was not a reading of the evidence. It was more like a stance, something close to a discipline, kept up in spite of the evidence. When the JoongAng Ilbo's editorial board wrote about his departure from the board chairmanship in October 2016, it ended on that point with something like awe: "Lee said he is fearful every day. His sense of vigilance and hunger should be a wake-up call to all local entrepreneurs." Most founders would be flattered. Lee's public record suggests he would have found it beside the point. Fear was not his message to other people. It was how he ran himself.
Under the Desks in Tokyo
Lee had stepped back from domestic management in January 2011, resigning as chief strategy officer to concentrate on overseas operations. "Back then, he made the decision in a bid to seek a future growth engine overseas away from the already saturated domestic market," a company official told the Korea Times. Japan was the first target. NHN had folded its Japanese subsidiaries into NHN Japan in 2003, launched a Japanese search engine in 2007, and bought the portal livedoor.com in 2010. For years it went nowhere. Search in Japan belonged to Yahoo Japan and Google, and a Korean portal's ten-year-old trick, getting users to write the web, did not take in a culture that did not want to.
On March 11, 2011, the Tōhoku earthquake struck. Lee was at the Japanese subsidiary. So was a Naver employee who later described the weeks that followed to the Hankyoreh:
"We would get down under our desks every time there was an aftershock. We knew that if we died at home, it might be hard for anyone even to find out about it, so we just stayed at the office, eating and sleeping there as we worked on the messaging app."
There is no better image of the Lee Hae-jin enterprise than that one. Engineers from a foreign company, sleeping on the office floor of a country not their own, crawling under their desks with each aftershock, building a communication tool for a population whose phone networks had just failed. They stayed at the office because dying alone at home meant no one would know. What they built over those weeks was, at bottom, a way for people to find out whether someone was still alive.
Line launched in June 2011. Lee is usually credited as its originator, though the operational builder was Shin Jung-ho, later Line's chief globalization officer, whose own later titles at Z Holdings would include group chief product officer. One account puts the founding team at about 15 people from Korea, Japan, China and the United States. Lee himself framed the origin as fear again, this time more specific. "Urgency," he said, was "the secret that led to Line's success": the urgent need to expand overseas "because the South Korean market is too small to survive in."
The growth was startling. In November 2013, after twelve years without a public address, Lee came out to celebrate Line reaching 300 million users. By June 2014 it had 470 million registered users worldwide, more than the 450 million of WhatsApp, which Facebook had just bought. It had 52 million users in Japan, 27 million in Thailand, 20 million in Indonesia, and only 14 million in Korea, where Kim Beom-soo's KakaoTalk held the ground. That year JP Morgan said Line was the only messenger that could challenge WhatsApp in South America. Naver's overseas revenue doubled in the first quarter of 2014 to 28 percent of total sales. Its market capitalization reached about 27 trillion won, and Lee noted, with a meticulous man's satisfaction, that Naver was the only startup in twenty years to break into Korea's top ten by market value.
The Japanese strategy then went somewhere less dramatic and in some ways more Lee-like. In 2013 Naver had an internal mail system called NaverWorks. Lee decided it could be a business software product and spun it into a company called Works Mobile. Its managers drew up a domestic plan. Lee, as Maeil Business tells it, sent them back to it with a sentence that reads like a koan: "win in the global market if you want to." The product launched in Japan in 2016 as Line Works. In its first month it brought in 200,000 won in sales, about 2.4 million won a year. The localization was close and specific. Japanese workers, the team found, tended not to share personal phone numbers with colleagues, and field crews needed mobile chat more than office suites. By 2017 Line Works led Japan's paid business-messenger market, and it held that position for seven straight years. By mid-2026 its annual recurring revenue topped 16 billion yen, about 150 billion won, with 590,000 customers and 5.8 million users. Monthly revenue had grown something like sixty-thousandfold.
The engineers under the desks and the 200,000-won launch month belong to the same story. Each began with a Korean product failing to win Korea, and each ended with Naver more Japanese than Korean in the place it mattered most. In March 2021, Line merged with Yahoo Japan under a structure that gave Naver and SoftBank 50 percent each of a new holding company called A Holdings, which in turn controlled Z Holdings, the company that ran both. Lee became chairman of A Holdings. In 2023 the combined business became LY Corporation. By then Lee's company held half the controlling interest in the most-visited website in Japan, a country that had colonized his own within living memory.
He has never, in the public record, said a word about that irony. A man who talks about Hunminjeongeum at venture events can be assumed to have noticed it.
Three Percent
In Korea the word that matters most in this part of the story is chongsu. Its roots mean something like "total head," and in practice it means the owner, the family patriarch who controls a chaebol through cross-holdings and dynastic loyalty no matter how few shares he holds directly. Korean law gives the idea a dry bureaucratic name, dongilin, the "same person," meaning the single individual who effectively controls a business group. Once a group's assets cross 5 trillion won, the Fair Trade Commission names that person, and the person takes on duties: disclosing transactions with relatives, reporting every company he and his family touch.
On September 3, 2017, the FTC designated Naver a large business group and named Lee Hae-jin its same person. Naver issued a statement the next day that reads in places like a manifesto:
"That Korea is unilaterally applying the same laws designed to regulate family-owned chaebol to private companies (which have grown independently) signifies that the country's corporate regulations are outdated... To advance, Korea must formally recognize more 'ownerless' private companies and foster a corporate environment where such firms can grow in number."
The company's case was strong on paper. Lee held about 4.3 percent of Naver; DongA Science's tally put it at 4.64. The National Pension Service owned more than twice as much, 10.5 percent. There was no circular shareholding. Naver owned most of its subsidiaries outright, and they owned theirs. None of Lee's relatives worked at Naver or held its shares. DongA's columnist compared him to Steve Jobs, who died holding 0.59 percent of Apple and whose authority came from "ability and trust," not equity.
Before the designation was final, Lee did something entirely out of character. The founder the Korean press called eundun-ui gyeongyeongja, the reclusive manager, walked into the Fair Trade Commission himself to argue that he did not own the company he had founded.
The FTC was not persuaded. Its reasoning was simple. Lee was the only large shareholder who took part in top-level decisions. The pension fund and the foreign institutions held more stock and did not run anything. A stake of 4 percent is a great deal when nearly everyone else holds less than 1. Then the commission turned up the details. Lee personally owned 100 percent of a management consulting company called Jieum. His cousin held half of a restaurant brand called Hwaeum. A tour business called YoungPoong Tour was tied to his relatives. In February 2020 the FTC announced it would refer Lee to prosecutors for failing to report 20 affiliated companies in Naver's 2015 filing. The list included Jieum, Hwaeum, Line Friends (the character business behind Line's bear and rabbit stickers), the digital outlet YTN Plus, and sixteen nonprofits. The FTC said the omissions were knowing, since Lee received regular reports on the companies. The statutory penalty for such a violation was up to two years in prison or a fine of up to 150 million won.
Lee kept working to make the ownerless argument true. In February 2018 he sold 195,000 shares. That spring he left the board entirely, having given up the chairmanship the previous March, and took the title Global Investment Officer, GIO, a job with no domestic operating authority. Then, in June 2019, at a joint symposium of the Korean Sociological Association and the Korean Academic Society of Business Administration in the Four Seasons Hotel in Jongno, the reclusive manager unloaded.
— Lee Hae-jin, Korean Sociological Association symposium, Seoul, June 2019It is completely sad to see expressions like "owner" when I have worked with the employees for decades. I believe that Naver is a new model for governance and transparency.
He had never thought even once in his life, he said, that Naver was his company. "I own only 3 percent of the company's shares, so the system does not allow me to make a decision by myself." Then, in a line that made headlines, he went after the idea of corporate social responsibility itself: "It's too much to ask tractor companies to solve even the unemployment of farmers." Companies should concentrate on surviving competition, he said, and leave the social burdens to "those in social organizations, politics and academia." Treating a company's growth as immoral, he added, was "undesirable."
The two halves of that afternoon pull against each other in a way he must have felt. In the first he insists he is not powerful: three percent, no family, a professional board, decisions made by others. In the second he speaks with the full authority of a man setting policy for an entire industry, telling sociologists what a company owes society. A man who really had no authority would not have been invited to give that speech. The FTC saw the same contradiction and wrote it into law. The commission was not entirely wrong, and Lee was not entirely wrong either. The concept had been built to cage dynasties, and Lee had no dynasty. He did have the power, and what he wanted was for the law to see the absent family and miss the power.
The God of the Republic
For years the National Assembly had asked Lee to testify at its annual audits, and for years he did not come. A Korea Times columnist reported a rumor in August 2018 that Lee was "so thin-skinned that he did not want to face parliamentary inspection," adding that "being away from the heat, Lee may have thought he is safe." By then he had already gone once. On October 30, 2017, after the three main parties agreed to pursue anyone who refused to appear, he sat in the second row of witnesses before the Science, ICT, Broadcasting and Communications Committee, a few seats from the president of Samsung Electronics' mobile division and the head of Apple Korea. It was his first appearance at a parliamentary audit since Naver's founding. One reporter wrote that the session could have been called "the Lee Hae-jin audit."
The occasion was a scandal over news. A week earlier Naver had admitted that, at the request of an official from the K League, Korea's professional football association, it had pushed an article critical of the league off its main sports-news page. It was the first confirmed case of news manipulation by the portal through which, at the time, most Koreans read most of their news. Naver did not write a word of it. It decided which words anyone saw.
The lawmakers did not hold back. Kim Kyung-jin of the People's Party gave the line that would follow Lee for years:
— Rep. Kim Kyung-jin of the People's Party, parliamentary audit, 2017Naver is the god of the Republic of Korea, which controls the consciousness, thoughts, brains, culture and spirit of the people of the country.
Lee apologized. "I believe that the sports news removal case is a serious issue of concern and I sincerely apologize." Then, by most accounts, he surprised the room twice. Rep. Kim Sung-tae of the Liberty Korea Party pressed him on a proposal to make large portals pay into the Broadcasting and Communications Development Fund, a levy traditionally paid by broadcasters and telecom carriers that use public spectrum. The industry, Kakao included, opposed it unanimously. Lee said: "If the Broadcasting and Communications Development Fund is imposed fairly, I am willing to pay it." Asked whether Naver might disclose its news-ranking algorithm, the most valuable trade secret a search company has, he said that "personally" he agreed with the idea: "If the possibility of the algorithm being abused by an external force is excluded, providing the algorithm to the public and having it assessed objectively would be a nice picture in the long term."
The Byline Network reporter covering the hearing saw what the qualifier meant. An algorithm disclosed with no risk of abuse or external attack is a contradiction in terms. Lee had agreed in principle to something that could never happen in practice. It was a beautifully built concession. It cost nothing and sounded like everything.
The rest of his testimony followed the pattern. On reforming news distribution he deferred: those decisions belonged to CEO Han Seong-sook, he said, and to the people running the domestic business, and would take time. On the call to shut down Naver News he made the case for survival: "When Cyworld failed, Facebook took all of its revenue." Cyworld, a Korean social network of the early 2000s, had been dominant at home before Facebook swallowed its users, and to Korean lawmakers the comparison needed no gloss. "We are losing video to YouTube," he said, "and photos to Instagram." On search advertising, the lucrative "power link" placements, he noted that Google held something like 90 percent of the world's internet ad market while Naver held about 70 percent of Korea's search market. By global standards the god of the Republic was a regional deity.
The pattern held in later years. Lee appeared again in October 2018, and in October 2022 at a comprehensive audit of the Korea Communications Commission. In 2021 lawmakers maneuvered to call him over financial-platform regulation; one report noted that if he left the country, Han would be summoned instead. In 2022 a sharply hostile Asia Today editorial put his remark that "media companies can choose how users view their news on Naver" next to the portal's four-tier grading of Korean outlets, a system it compared to the hereditary status ranks of the Joseon Dynasty. It quoted lawmakers from both parties, one of whom likened reading Naver News to "publishing Rodong Sinmun every day," the North Korean party paper. The editorial's headline was "Naver head Lee Hae-jin's inconsistency." It noted the algorithm he had agreed in principle to open remained closed.
The charge of inconsistency deserves to be taken seriously and then set beside the record. Every argument Lee made against Google at the national level, that one company should not decide what a whole people sees and that diversity of sources is a civilizational good, applied with equal force to Naver within Korea. He knew it. In 2019 he called the domestic monopoly criticism "hurtful," and in nearly the same breath said that "if a bigger company than Naver is established, we can call it a big success." He seemed to mean it the way a man means a hope he does not expect to see come true. The defender of diversity ran the least diverse information market in the developed world. That was not hypocrisy exactly. He had kept a small nation's information out of a foreign monopoly, and the price was that the nation got a domestic one.
The Recreational Soccer Club, Revisited
In May 2021 a senior Naver developer died by suicide after accusing the company of fostering a toxic workplace. A labor union investigation found he had been bullied by executives for years. A later government survey found that more than half of Naver employees reported being bullied at least once in a six-month period. Naver acknowledged that "there were some acts of workplace harassment by some executives."
On June 30, 2021, Lee, still GIO with no formal domestic authority, sent an email to every employee. He apologized. He wrote of shock and disappointment. And according to the Maeil Business account, he signaled a complete overhaul of current management within the year and said he himself would step still further away.
It is the only moment in the public record where Lee's habit of absence reads as penance rather than strategy. Nine years earlier he had stood before the same company and complained that its people treated it like a neighborhood soccer club, that they had come to live comfortably. Nothing in the sources connects that lecture to the culture that killed a developer, and it would be glib to draw the line. But the two moments sit in the same building, and the founder had to live with both. He had demanded urgency. The company had produced, somewhere inside it, cruelty. He answered with the move he knew best, which was to leave.
The board's reply was Choi Soo-yeon. She was a Harvard-educated mergers lawyer who had joined Naver in 2019 as head of global business support, was 40 when appointed, and had no background in running a portal. Fortune described the job she was given as caretaker, a CEO chosen to show the company was ready to change its culture. Her first interview with the international press came two years into the role, at headquarters in Seongnam, conducted in Korean and translated. In it she described Naver almost exactly the way Lee had been describing it for a decade: "a company that constantly fights against such a world," a world of one or two search companies and one or two commerce companies, "and strives to preserve diversity." The caretaker had become a true believer. Or she had always been one. That may be why he chose her.
Starbucks, Tea, and the Stone Before the Sling
The large language models arrived in force in late 2022, and Naver, which had launched what it called Korea's first hyperscale model, HyperCLOVA, in 2021, found itself far behind on cost and performance. It released HyperCLOVA X in 2023, a model trained mainly on Korean rather than English or Chinese. A chatbot, CLOVA X, followed in October. It was respectable, built from scratch, and clearly not as good as what came out of San Francisco.
Lee's reemergence came gradually. On May 21, 2024, he spoke at a session of the AI Seoul Summit. It was his first public appearance since the October 2022 audit, and the Maeil Business headline called him the "Secret Manager." He spoke about children.
"Unlike searches in which users select information from various search results with one keyword," he said, "AI that provides immediate answers is a way to get answers without choice, and these AI characteristics will have a very significant impact on children and adolescents, especially." Then he quoted Orwell's 1984: those who control the past control the future; those who control the present control the past. "If the very few AI dominate the present," he said, "the perception of past history and culture will be made only by the answer of the AI, and eventually the AI will have a great influence on the future."
Here the sixteenth-century king and the twentieth-century novelist turn out to be making one argument. Hunminjeongeum gave a people the means to read for themselves. A single dominant AI would give them one reader to read for them. A search engine, even a monopolistic one, still returns a page of results and leaves the choosing to the user. Lee had spent twenty-five years defending the results page. Now the page itself was going away.
Five weeks later, on June 25, 2024, Lee, Choi and Naver Cloud CEO Kim Yoo-won flew to Santa Clara and met Jensen Huang at Nvidia's headquarters. It was the first meeting between Naver's leadership and the man whose chips the whole field depended on. The official photograph shows the three Koreans smiling. Choi holds a framed picture of Nvidia's office building, a gift from Huang. The vocabulary they agreed on was "sovereign AI": every nation, or region, or language running models trained on its own data, in its own data centers, reflecting its own culture. Nvidia liked the idea because it meant selling chips to every government on earth. Lee liked it because he had been building toward it since 2002.
Two days after the Nvidia meeting, on June 27, Webtoon Entertainment, the comics business Naver had launched in 2004 and spun off in 2017, opened on the Nasdaq. This time Lee was in the room. He stood next to CEO Kim Jun-gu at the opening bell, and the photographers caught him smiling broadly. Kim told reporters he had been proud of him and had said to Lee that he had been through a hard time. The shares closed nearly 10 percent up. Eight years after watching Line's bell on a television in Chuncheon, Lee Hae-jin rang one in person. The sources don't say whether he was afraid that morning. They say he smiled.
In February 2025 Naver's board approved his return as an inside director, and on March 26, at the annual meeting held in the Green Factory in Seongnam, the shareholders confirmed it. He was named board chairman the same day, seven years after leaving. Naver had become, the year before, the first Korean internet company to pass 10 trillion won in annual revenue. Even so, at the same meeting shareholders stood up to complain: "Naver has been operating globally for a long time, but no clear results have been made." And of the $1.2 billion purchase of Poshmark, the American clothing-resale platform he had championed as GIO, a deal Reuters said in 2022 "invites skepticism," they said: "We have acquired C2C companies such as Poshmark, but no results have been seen."
Lee's first words to employees on his return were plain: "Naver must play its role in the AI era, beyond the internet and mobile era." To the shareholders he gave the company's founding story back to them as a thesis: "The key to Naver's success in the Internet era was its passion to turn innovative technologies into services most needed by users and Naver's determination to build up in different ways than companies with larger markets and capital power." Afterward he told reporters: "It is very sad that the world uses only one or two search engines or AI."
His first official visit after his return, days before the shareholders' meeting, was to Seoul National University Children's Hospital, where Naver was funding a three-year, 30-billion-won digital-bio research program out of a 2023 donation. "We have to boldly get on the huge wave of AI," he told the forum, "but I think we have to invest in smart people first so that we can move forward well while changing our direction and strategy."
On June 5, 2025, he went to Palo Alto, to a ballroom at the Four Seasons, to announce Naver Ventures, the company's first North American investment arm. Two hundred people from the Valley showed up. It was, one Korean outlet noted, his first appearance before overseas media since Line's listing in 2016. He told the Hunminjeongeum story. He made the data argument. Then he reached for an image that only someone raised in a tea culture would think of:
"Starbucks is known for delivering consistent quality worldwide. But it would be unfortunate if Starbucks were the only place to get coffee everywhere."
Afterward, speaking to a Korean correspondent, he was blunter than he had ever been in public. "Naver's AI is insufficient," he said. "We have always been short. But we are used to the fight because everything, including searches, has been built up in short supply." And then the David line, garbled slightly by machine translation into something that reads, accidentally, like scripture: "If David wants to beat Goliath, he has to concentrate, but I think this is the process of catching the rock." Before the stone, he explained, there has to be a large language model, a cloud, and every basic technology in place.
It was an admission that the stone had not yet been picked up. The rest of 2025 and 2026 was spent picking it up. Naver's search business, which had made up 78.06 percent of revenue at its 2015 peak, fell to 35.9 percent in the first half of 2025, the gap filled by commerce, where Naver's 20.7 percent share of Korean e-commerce was closing on Coupang's 22.7. In November 2025 Naver announced that its payments arm, Naver Financial, would acquire Dunamu, the operator of Korea's largest cryptocurrency exchange, Upbit, in a stock swap valuing the combined company at $13.6 billion. The aim was a stablecoin rail that would let Naver's webtoons, shops and future AI agents settle payments across borders without a card network in between. On June 8, 2026, Jensen Huang came to Seongnam, toured 1784, the second Naver headquarters, finished in 2022 and billed as the world's first robot-friendly office building, and stood with Lee to announce a gigawatt-scale "AI factory" partnership. Brookfield signed on in July. In September Lee met Bruce Flatt to confirm the project was on track, and Naver proposed an AI factory partnership for Europe in France. A digital-twin platform became Saudi Arabia's national standard. The first export of the 1784 smart-building system went to Japan.
Naver's second-quarter 2026 revenue was 3.3888 trillion won. The company was about to spend a sum many multiples of the 500 million won Samsung had once given an engineer with a two-hour-a-day habit, all to build, again, the infrastructure for a Korean answer to an American question. "Diversity," Lee had said in Palo Alto, "is going to shine only brighter."
Sixty-Seven Languages
Some Naver products have died along the way. Camp Mobile was folded back into the parent. V Live went to Hybe's Weverse. Junior Naver, launched in 1999 alongside the search engine as Korea's first children's portal and for years the country's last, with a panel of educators screening out harmful content, went dark on May 27, 2025, two months after Lee came back to the board, a year after he stood up at a summit to worry about what single-answer machines would do to children's sense of history. The pieces of the company that taught a generation of Korean kids to search are gone. The worry stayed.
And some things have quietly survived. In 2009 Naver became the first company in Korea to digitize the country's newspapers from 1920 to 1999, eight decades running from Japanese colonial rule through war, dictatorship and democracy, keyword-searchable on a portal built by a man who quotes Orwell on the control of the past. The Green Factory in Seongnam, 27 floors above ground and seven below, is wrapped in vertical louvers that open at sunrise and close at sunset without anyone touching them.
Naver Dictionary launched in 1999 too, on the same day as the search engine. It supported two languages, Korean and English. It has added languages every few years since, Uzbek, Swahili, Khmer, Tamil, Georgian, Albanian, Nepali, and in 2016 Naver committed 10 billion won over five years to building and revising them. By 2024 it supported sixty-seven. In December 2018, among a batch of five minority-language dictionaries opened that month, it added ancient Hebrew, a language with no country to protect it and no living speakers to ask.
Part IIThe Playbook
What follows are twelve principles drawn from Lee Hae-jin's twenty-seven years building and defending Naver. Few of them are comfortable. Several of his strengths turned into liabilities at scale, so the playbook includes the costs alongside the moves. Read it as a manual for the underdog who has to stay an underdog, which means anyone competing against a rival with forty times the revenue, an absent home market, or both.
Twelve Principles
- 1.When algorithms converge, compete on proprietary data.
- 2.Keep fear as a daily discipline, not a mood.
- 3.Steal two hours a day from the employer.
- 4.Hand off the parts of the job you're bad at, including the title.
- 5.Turn a small home market into urgency.
- 6.Win abroad where the giant hasn't localized.
- 7.Build the infrastructure before you need it.
- 8.Make the company stand for something larger than its market share.
- 9.Separate founder from operator, and expect the law to disagree.
- 10.Under scrutiny, concede the principle and defend the position.
- 11.Come back only when the wave is as big as the last one.
- 12.Ration your public voice.
Principle 1
When algorithms converge, compete on proprietary data
Lee's central strategic insight is that the core technology of a category commoditizes, and what doesn't commoditize is the material it works on. Early search engines competed on ranking; within a few years they had converged on similar methods. Naver could not beat Google's algorithm in 2002 and did not try. It built Knowledge iN, Blog and Cafe, services that turned Korean users into producers of Korean-language content stored on Naver's servers, out of any outside crawler's reach. The 200 million answers that resulted were never going to be the best information on earth. They were information no one else had.
Lee spelled out the same thesis for AI in 2025: today's model race is about "performance and algorithms," but "there will be a time when all LLMs reach a similar level — and when that time comes, data will once again become a critical element for differentiation." Whether he is right about the models is an open question. What is certain is that he has been making the same bet for twenty-three years and was right the first time.
The cost is quality control. Naver's tolerance for loosely sourced content got Knowledge iN unfavorable comparisons with Wikipedia. Proprietary data is a moat. It is not automatically a good one.
Tactic: Identify the layer of your category that is converging, then put your investment into a data asset that only your users can create and only you can hold.
Principle 2
Keep fear as a daily discipline, not a mood
"For the past 17 years, I've been scared every morning." This is the most repeated sentence in Lee's public record, and it is easy to dismiss as false modesty from a billionaire. That reading misses its function. Lee's fear did not track Naver's actual position, which by most measures was dominant. It was a stance kept up against the evidence, a refusal to let a market share of 59 or 70 percent become a reason to relax.
The practical effect shows in the timing of his moves. He went after Japan in 2011, while Naver still ruled Korean search, because he feared a desktop company would not survive mobile. He talked about fear on the day of Line's IPO rather than celebrating, because the capital was "leeway," not victory. He returned in 2025 because AI was "a huge wave at the Internet and mobile level."
The 2012 "neighborhood recreational soccer club" lecture shows the limit. Fear is a founder's private fuel. Pushed onto an organization as a culture, it can curdle, and the 2021 crisis, whatever its specific causes, is a reminder that urgency at the top and cruelty in the middle can live in the same building.
Tactic: Write down, every quarter, the specific competitor or technology most likely to make your company irrelevant within five years. Act on that list, but don't hand the anxiety itself down the org chart.
Principle 3
Steal two hours a day from the employer
The 25 percent rule, two of every eight working hours spent on new technology outside his assigned work at Samsung SDS, sustained for about three years, is how Naver came to exist. It turned a salaried engineer into a founder without any dramatic resignation. By the time he left, he had a team (assembled in 1994), an internal competition win (1997), a 500-million-won venture, and a product. The risk had been taken in installments.
This principle is about the shape of a career more than about moonlighting. Lee de-risked founding by building the company inside a host organization that supplied salary, colleagues and eventually seed capital. Samsung got an in-house venture. Lee got a company.
Tactic: Set aside a fixed, non-negotiable block of every workday for the project you actually believe in, and let it accumulate into something your employer would rather back than block.
Principle 4
Hand off the parts of the job you're bad at, including the title
After NHN's 2002 listing, Lee reportedly admitted he was not good at socializing and played golf badly, and gave the chief executive job to Kim Beom-soo. By 2004 he had left any CEO title at NHN. He kept the board chairmanship and strategy, the parts of the job that suited a meticulous, introverted engineer, and gave away the parts that ran on relationships.
That pattern continued for two decades. Kim Sang-hun ran Naver for eight years. Han Seong-sook succeeded him. Choi Soo-yeon, a lawyer, succeeded her. Lee took the title GIO and spent his time on Line, Europe and deals. He has stayed, in effect, a founder-strategist who employs operators.
The cost is that capable partners can leave. Kim Beom-soo walked out in 2007 and built KakaoTalk, the product that beat Lee's in Korea. Handing off the CEO role means handing someone a CEO's education.
Tactic: List the parts of your role that depend on traits you don't have, give those parts away along with real authority, and plan for the possibility that the person you trained becomes your strongest competitor.
L
The Founder at a Distance
Lee's formal roles moved steadily away from operations before swinging back.
1999
Founds Naver Comm on June 2; becomes chairman of the board.
2000–02
Merges with Kim Beom-soo's Hangame; NHN lists on KOSDAQ; Lee cedes the CEO role.
2004
Steps down as NHN chief executive.
2011
Resigns as chief strategy officer in January to focus on Japan; Line launches in June.
2017
Leaves board chairmanship in March; FTC names him Naver's "same person" in September.
2018
Sells 195,000 shares; leaves board entirely and serves as Global Investment Officer.
2021
Emails all staff after employee's death; signals management overhaul and his own further withdrawal.
2025
Returns as board chairman on March 26 to lead Naver's AI push.
Principle 5
Turn a small home market into urgency
"Urgency was the secret of Line's success," Lee said in 2016, and he was specific about where it came from: "the South Korean market is too small to survive in." Many companies from small countries treat their size as a ceiling. Lee treated it as a deadline. When Kim Beom-soo's KakaoTalk took Korea, Lee did not spend years fighting over the remainder. He put Line where Kakao wasn't, and by 2012 more than 90 percent of Line's users were outside Korea.
The same instinct shaped Line Works. When Works Mobile drew up a domestic plan, Lee told its leaders to "win in the global market if you want to." It launched in Japan, a market more than three times Korea's in cloud software, and went from 200,000 won in first-month revenue to 16 billion yen in ARR.
The risk is that urgency can outrun judgment about where to spend it. Shareholders in 2025 were openly frustrated that the overseas acquisitions of Lee's GIO years, Poshmark above all, had not yet produced clear results.
Tactic: If your home market can't sustain a category leader, set an explicit date by which you'll compete abroad, and choose that market by where your strongest domestic rival is weakest.
Principle 6
Win abroad where the giant hasn't localized
Lee's foreign wins share a feature: each found the gap between a global product and a local habit. Knowledge iN filled a gap that Google's English-trained crawler couldn't, the thinness of the Korean web. Line launched in Japan in the weeks after the Tōhoku earthquake, when people urgently needed to know whether others were alive. Line Works was built around a quirk of Japanese work culture: reluctance to share personal phone numbers with colleagues and a field workforce that needed mobile chat more than office suites.
N
Naver Against the Giants
Where Naver chose to fight, and how.
| Arena | Incumbent advantage | Naver's wedge | Outcome |
|---|---|---|---|
| Korean search | Google's superior algorithm | User-generated Korean content (Knowledge iN, Blog, Cafe) | Won 2× Google's searches in 2025 |
| Korean messaging | KakaoTalk's first-mover lead | Aggressive investment and promotion | Lost 14M Korean Line users vs. Kakao's dominance |
| Asian messaging | WhatsApp's global scale | Japan-first launch; stickers, local partnerships | Won 470M users by June 2014 |
| Japanese work chat | Global collaboration suites | Mobile-first chat for field workers | Won No. 1 paid business messenger seven years running |
| Korean e-commerce | Coupang's logistics | Zero-fee Smart Store, Naver Pay, AI recommendations | Contested 20.7% vs. 22.7% (2024) |
| Foundation models | U.S. and Chinese capital and compute | "Sovereign AI" for non-English markets | Open Nvidia, Brookfield AI factory partnerships |
The table shows the limits as well. In Korean messaging Naver had no localization edge, since Kakao was just as Korean, and it lost. Localization is a wedge only against a foreign incumbent.
Tactic: Before entering a market dominated by a global player, name the specific local behavior the incumbent's product ignores, and build your first version only for that behavior.
Principle 7
Build the infrastructure before you need it
Lee spent the day of Line's IPO at a data center. The choice made a point. GAK Chuncheon, finished in 2013, made Naver the first Korean portal to own its data facility, years before "data sovereignty" was a policy phrase. A second GAK opened in Sejong in November 2023. Naver spent close to 1 trillion won on the two. Naver Labs (2017), the Xerox research center in Grenoble (2017), the 1784 headquarters (2022), and the 2026 gigawatt-scale AI factory with Nvidia and Brookfield all come from the same impulse.
His David-and-Goliath line from Palo Alto says it outright: before the stone, "there should be a giant language model, there should be a cloud, and all basic technologies should be prepared." The underdog can't rent the giant's infrastructure and expect to beat the giant.
This is the costliest principle in the playbook. In 2018 analysts warned that Naver's hiring of more than 2,000 people the previous year would squeeze profitability. Infrastructure built early looks like waste until the moment it becomes leverage.
Tactic: Identify the one piece of infrastructure your largest competitor could someday use to cut you off, whether compute, data storage, payments or distribution, and own it before you're forced to.
Lee did not describe Naver as a search company. He described it as a guardian of diversity, a company that would "resist the imperialism of global giants to the end, and survive to the last." He tied it to Hunminjeongeum and metal type, to Orwell, to the right of every country to have its own tea alongside Starbucks. This framing did real strategic work. It made Naver's survival a matter of national interest. It recruited allies in Europe, where U.S. tech firms had irritated regulators. And it handed Nvidia a reason to partner: "sovereign AI" means every nation needs its own models and its own chips.
The framing also turned a caretaker CEO into an evangelist. Choi Soo-yeon's line in 2024 about "a company that constantly fights against such a world and strives to preserve diversity" is Lee's argument in her voice.
The cost is exposure. A company that stands for diversity can be held to it at home, and Korean lawmakers did exactly that, likening Naver's news dominance to a single state newspaper. The mission becomes the measure of your hypocrisy.
Tactic: State the larger cause your company serves in terms that make allies of governments and partners, and before you say it, audit whether your own market behavior can survive the same standard.
Principle 9
Separate founder from operator, and expect the law to disagree
Lee's governance experiment was to build a Korean company of national scale with no chongsu: a founder holding 3 to 4 percent, no family involvement, no circular shareholding, professional CEOs, and the founder himself formally off the board from 2018. He argued for it in person at the FTC and in public at the 2019 symposium: "I own only 3 percent of the company's shares, so the system does not allow me to make a decision by myself."
The FTC rejected the argument, and its reasoning holds lessons. Formal distance did not cancel functional influence: Lee was the only major shareholder involved in top-level decisions, and he regularly received reports on companies tied to him and his relatives. The 2020 referral to prosecutors over 20 unreported affiliates, including his wholly owned Jieum, showed that the ownerless model needed tidiness that its founder had not entirely kept up.
The lesson isn't that the model fails. Lee's 2025 return was a two-top system with Choi Soo-yeon running operations and Lee setting AI direction, which is the same arrangement with the pretense taken out.
Tactic: If you build a founder-light governance structure, make your actual influence visible and formalize it rather than denying it, and keep your personal holdings as clean as the structure claims to be.
Principle 10
Under scrutiny, concede the principle and defend the position
At his first parliamentary audit in October 2017, Lee did three things in sequence. He apologized cleanly for the K League news manipulation. He made two concessions that shocked the industry, agreeing to pay into the broadcasting fund "if imposed fairly" and calling algorithm disclosure "the right direction" provided there was no risk of abuse or external attack. Then he defended the business itself: search ads are universal, Naver's 70 percent of Korea is small next to Google's 90 percent of the world, and shutting down news would hand the revenue to foreigners, "when Cyworld failed, Facebook took all of its revenue."
The concessions were built with conditions that could never quite be met, so they bought goodwill at almost no cost. Lee deferred operational questions to the CEO, which kept him from having to commit.
This principle has a shelf life. Five years later, an editorial titled "Naver head Lee Hae-jin's inconsistency" pointed out that the algorithm remained closed. A concession in principle works once. Repeated, it turns into evidence against you.
Tactic: When called to account, apologize specifically, concede the general principle your critics care about, defend your core economics with comparative data, and either follow through on one concession within a year or don't make it.
Principle 11
Come back only when the wave is as big as the last one
Lee stayed off the board for seven years through Naver's crises, including the 2020 prosecutor referral, the 2021 bullying scandal and repeated parliamentary summonses. What brought him back in 2025 was a judgment about scale: "AI is a huge wave at the Internet and mobile level, and the idea that it should fundamentally change was the trigger for a mode change." He had led Naver through the internet era and set its mobile strategy through Line. He came back for the third wave and no smaller one.
The return was preceded by a deliberate run-up: the AI Seoul Summit speech (May 2024), the first Nvidia meeting (June 2024), the Webtoon Nasdaq bell (June 2024). When he took the chair he already had the partnerships and the public framing in place. Within eighteen months there was a Dunamu deal, a gigawatt AI factory, and Jensen Huang touring 1784.
Tactic: Reserve your personal reentry for platform shifts of the same size as the one your company was born in, and spend the year before your return lining up the partners you'll need on day one.
Principle 12
Ration your public voice
By the JoongAng Daily's count in 2014, Lee had spoken publicly only twice in fifteen years: once in 2001 to announce Hangame becoming a paid service, and once in 2013 to celebrate Line's 300 million users. Then came a full lecture to 500 small-business CEOs in Jeju. His 2025 Palo Alto speech was his first overseas media appearance since 2016. The Korean press called him the hermit, the reclusive manager, the secret manager.
Scarcity made each appearance an event. When Lee spoke, about Hunminjeongeum or Orwell or Starbucks, his words were reported verbatim and read closely as signals. A founder who speaks weekly is commentary. A founder who speaks every few years is policy.
There's a real cost, which Korea Times columnists and lawmakers were glad to name: a silent founder looks evasive, even thin-skinned, and when he finally shows up it is often under subpoena, on someone else's terms. Silence carries power only if you choose when to break it.
Tactic: Appear in public only when you have a specific, durable argument to make, and break your silence before a crisis forces you to rather than after.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Lee Hae-jin, Naver Ventures launch, Palo Alto, June 5, 2025I was a new hire when Google first unveiled its search engine, and I thought back then that the information gap was going to lead to a gap in competitiveness — that is why I decided to develop our own Korean-language search engine.
— Lee Hae-jin, Korean Sociological Association symposium, June 2019A search engine service pursues diversity, and I want Naver to be a company that has resisted the imperialism of global giants to the end, and survived to the last.
— Lee Hae-jin, Naver Ventures launch, Palo Alto, June 2025Starbucks is known for delivering consistent quality worldwide. But it would be unfortunate if Starbucks were the only place to get coffee everywhere.
— Lee Hae-jin, to a Korean correspondent in Silicon Valley, 2025Naver's AI is insufficient. We have always been short. But we are used to the fight because everything, including searches, has been built up in short supply.
— Lee Hae-jin, AI Seoul Summit, May 2024If the very few AI dominate the present, the perception of past history and culture will be made only by the answer of the AI, and eventually the AI will have a great influence on the future.
Maxims
- Algorithms converge; data doesn't. Compete on the material only your users can produce, not on the method everyone will eventually share.
- Stay afraid on purpose. Treat fear as a daily practice that holds up against good news, not a reaction to bad news.
- Found inside the host. Take two hours of every eight and let the company you're building grow in borrowed time until it can stand on its own.
- Give away the golf. If the job runs on traits you lack, hand off the job and keep the strategy.
- A small market is a deadline. When home can't sustain you, the question is when you leave, not whether.
- Localize to the habit, not the language. Translation alone is easy to copy. Building for one specific local behavior is the real edge.
- Own the vault before the bell rings. Build data centers, clouds and models before you need them, because the underdog can't rent from Goliath.
- A mission is also a standard. The cause that recruits your allies will be used to measure your conduct, so live up to it at home first.
- Distance isn't the same as absence. Founder-light governance works only if the founder's real influence is visible and his own holdings are clean.
- Make silence count. Speak rarely, speak with a single argument, and break your silence before a subpoena breaks it for you.
