검색결과 총 6건
-
Lessons in Leadership from the Classics | Chapter 4: Nvidia
The Patience and the Ascent of Jensen Huang How the Alleyways of Korea and the Floors of Semiconductor Factories Forged the Aesthetics of 古枯孤高 [Economy Daily] At the beating heart of the civilizational upheaval we call artificial intelligence stands one company and one man: Nvidia and Jensen Huang. The world measures them in market capitalization and market share. But the deeper truth of great leadership outlasts any number. It is the power of time, long and unhurried. It is the discipline of subtraction. It is the courage of solitude. And it is, finally, the dignity that comes only from having endured. In the vocabulary of East Asian philosophy, these four qualities compress into a single phrase: 古枯孤高 — ancient (古), austere (枯), solitary (孤), elevated (高). Nvidia's rise is not the story of a stock that spiked overnight. It is the story of these four characters slowly calcifying into the bones of one man and the culture of one company, across thirty years of painstaking accumulation. 古 — The Ancient: Time as the First Discipline Every great enterprise, if it is truly great, eventually earns its face — but only through time. Jensen Huang had been walking this earth as a businessman long before the world knew his name. His relationship with Korea begins here, and it begins on foot. According to domestic industry accounts from that era, Huang made repeated visits to Yongsan Electronics Market in Seoul in the late 1990s and early 2000s — when Nvidia was still an obscure startup struggling to be taken seriously. He came not as a visiting dignitary but as a salesman: explaining graphics cards to shop owners, persuading assemblers, winning trust one transaction at a time. Huang himself has said his connection to Korea dates to 1996. He has spoken of how South Korea's explosion of high-speed internet, its PC-bang culture, and the nationwide fever for StarCraft formed a critical foundation for Nvidia's early growth. Korea, in those years, was the world's most electrified laboratory for digital culture — and the heat of its gaming rooms, the sharpness of its consumers, the velocity with which it embraced new technology, all of it nourished a company that had not yet found its footing. This detail matters enormously. The histories of great corporations are often rewritten to begin in gleaming boardrooms or on famous stages. But Jensen Huang's formation happened in narrow storefronts, surrounded by towers of component boxes, in a market where customers were price-sensitive and performance-obsessed and utterly unimpressed by brand mythology. In Yongsan, he did not sell a brand. He sold credibility. He sold product knowledge. He sold the felt experience of superior performance. The I Ching offers an image for this season of a man's life: 潛龍勿用 — "the hidden dragon does not yet act." The dragon submerged beneath the water has not yet ascended to the sky, but it is already gathering strength, already orienting itself toward its direction. Korea was that submerged time for Jensen Huang. It was where the dragon went quiet and grew. 枯 — The Austere: The Discipline of Withholding Austerity is not poverty. It is restraint. And few companies in the history of Silicon Valley have practiced restraint as rigorously or as consequentially as Nvidia. While its competitors raced to win the surface war — chasing specification numbers, upgrading the cosmetics of their products, playing to the gallery of consumer benchmarks — Huang kept his organization's attention trained on something less visible and far more consequential: the underlying architecture of computation, the logic of parallel processing, the infrastructure that would eventually become the indispensable engine of artificial intelligence. This is the aesthetic the Chinese literati call 枯淡 — a beauty that comes not from ornament but from essence. The Diamond Sutra puts it this way: 凡所有相 皆是虛妄 — "all that has form is ultimately illusion." In business terms: what catches the eye rarely determines a company's fate. What determines fate is the capability that cannot be seen. Nvidia understood this early. That is why the Nvidia of today rests not on the appearance of its products but on the depth of its software ecosystem, its developer base, and the intellectual architecture that competitors cannot easily replicate. This philosophy of austerity extends to Huang's understanding of human character. Speaking at Stanford, he told students that the most important trait for success is not intelligence but resilience — and went further, saying, "I hope you will have the experience of suffering and hardship." It is a startling thing to say, and deliberately so. His point is unambiguous: greatness is not the product of cleverness alone. Character is forged not in comfort but in friction. Huang speaks from experience. He has publicly described being bullied in an American boarding school as a boy, washing dishes and cleaning bathrooms at minimum wage. His philosophy of hardship is not rhetoric. It is autobiography. Most organizations today speak to their people endlessly about well-being and are afraid to speak about tempering. But Jensen Huang did not flinch from the uncomfortable truth: growth always requires some degree of resistance and endurance. He knows this in his body. 孤 — The Solitary: The Courage of the Unfashionable Conviction Solitude, properly understood, is not the condition of being alone. It is the willingness to choose a road that others have not taken — and to walk it long enough to find out whether you were right. Nvidia was, for a very long time, a company that received no particular applause. It was known as a graphics chip company, and in that category, it was formidable. But inside that public identity, Huang carried a private and lonely conviction: that the dominant paradigm of computing would shift — that the age of the general-purpose CPU would eventually yield to an age of accelerated computing. Markets demand the present moment. Leaders sometimes have to absorb today's contempt in exchange for tomorrow's vindication. Only those who sustain that solitude earn the right to the rewards of early arrival. The Analects of Confucius puts it plainly: 德不孤 必有隣 — "virtue is never truly alone; it will always find its neighbors." What appears solitary and eccentric at the beginning eventually draws its community. And in the story of Nvidia and Korea, this movement from isolation to alliance is almost perfectly illustrated. The partnership between Huang and South Korea has long since outgrown its origins in retail sales. SK Hynix began collaborating with Nvidia on High Bandwidth Memory in the uncertain early days of that technology — a bet made before the outcome was clear. That relationship has since deepened into something that resembles co-development more than supply chain. Nvidia has been advancing large-scale AI chip supply and infrastructure cooperation with the Korean government, Samsung, the SK Group, Hyundai Motor Group, and Naver. The lonely salesman who once walked the aisles of Yongsan is now at the table with the leaders of Korean industry and government, shaping the architecture of the nation's AI future. The solitary vigil became a strategic alliance. What was once walked alone is now walked together. 高 — The Elevated: Altitude as Accountability Elevation is not merely position. It is character — the capacity to see farther and to hold responsibility longer than others can or will. The Doctrine of the Mean speaks of 至誠無息 — "true sincerity never rests." This is, unexpectedly, one of the most precise descriptions of how Jensen Huang has run his company. He did not build Nvidia on a passing fashion. He crossed product failures, market cynicism, supply chain crises, and geopolitical headwinds, and climbed — slowly, deliberately, one foothold at a time — to the position the company occupies today. This is not a mountain ascended in a season. This is a summit reached in decades. Here, again, Korea re-enters the story. However regal the title "emperor of the AI era" may sound, the circuitry running through that crown is substantially Korean. Korea began as the consumption frontier — the PC-bang, the gaming market, the early adopter culture that gave Nvidia its first mass foothold. It has since become the strategic frontier: the partner in HBM and advanced memory, the co-architect of AI factories and digital transformation. Between the image of Jensen Huang persuading shop owners in Yongsan and the image of Jensen Huang discussing AI infrastructure with the heads of Korea's largest conglomerates, there runs a very long river. But the river is unbroken. What he first saw in Korea was not merely a sales opportunity. He saw a society with an extraordinary capacity for fast technical comprehension, for organizing technology into industry, for connecting the work of the mind to the work of the factory floor. That insight lives inside every partnership he has built here since. A Reckoning for Korean Business What, then, should Korean business leaders take from this? The lesson is not complicated, though it is demanding. Innovation does not arise from eloquent mission statements. It arises from time endured, from the discipline to discard the inessential, from the independence to pursue an unpopular answer, and from the accountability that eventually transforms all of it into something worthy of the word dignity. Jensen Huang's career is not a story of a man who happened to catch the AI wave at the right moment. It is a story of sediment — of years and decades of experience, discipline, and conviction accumulating until they were precisely aligned with the door that history opened. Which asks certain questions of Korean business. Do we still carry the original instinct of those years when we wrestled with the market on the ground floor — when we had no reputation to trade on, only our knowledge and our reliability? Do we have the austere courage to strip away what is not essential? Do we have the nerve to choose the lonely right answer over the popular wrong one? Management, at its best, is completed in the love of people, in the respect for the work done in the field, and in the refusal to defy the logic of time and nature. The tree that grows too fast is hollow at its core. The success that comes too easily has shallow roots. Nvidia — Jensen Huang's Nvidia — took the opposite path. It stood like an ancient tree, silent and unhurried, enduring the winds and the droughts, growing upward alone toward the high place it had decided, long ago, to reach. His success, for that reason, is not a flash of light. It is light that stays. That is the lesson of 古枯孤高. Only those who have endured long enough ascend high enough. Only those who have passed through austerity reach genuine depth. Only those who have borne solitude long enough find themselves, one day, at the center of their age. Jensen Huang's Korean story is one essential thread in that larger narrative. Today's glory is conceived in yesterday's alleyways. Even the history of the world's most powerful technology company is completed, in the end, only on the accumulated sweat and trust of human beings. He is demonstrating that, quietly, every day. The author is a contributing columnist covering business philosophy, technology, and economic history.
2026-04-22 11:57:23
-
구글, '젬마4' 출시로 오픈소스 AI 경쟁 참전…'제미나이' 투트랙 전략
[경제일보] 구글이 오픈형 인공지능(AI) 모델 '젬마 4'를 공개하며 개방형 AI 시장 공략에 속도를 내고 있다. 고급 추론 능력과 온디바이스 실행을 강화한 모델을 앞세워 개발자 생태계를 확대하고 폐쇄형 모델인 제미나이와의 투트랙 전략으로 AI 경쟁력 강화에 나서는 것으로 분석된다. 2일(현지시간) 구글은 고급 추론과 에이전트 기반 워크플로우를 지원하는 오픈형 모델 제품군 '젬마 4'를 출시했다고 밝혔다. 젬마 4는 아파치 2.0 라이선스로 제공되며 상업적 활용이 가능하다. 구글은 오픈형 모델을 통해 개발자와 기업이 다양한 환경에서 AI를 활용할 수 있도록 접근성을 확대한다는 전략이다. 젬마 4는 'Effective 2B(E2B)', 'Effective 4B(E4B)', '26B Mixture of Experts(MoE)', '31B Dense' 등 총 4가지 모델로 구성됐다. 해당 모델들은 단순한 대화 기능을 넘어 복잡한 논리 처리와 에이전트 기반 자동화 작업을 지원하도록 설계됐다. 특히 고급 추론 기능과 코드 생성 능력, 이미지·오디오 처리 기능 등 멀티모달 지원을 강화했으며 최대 256K 컨텍스트를 지원해 긴 문서 처리도 가능하다. 또한 140개 이상의 언어를 지원해 글로벌 애플리케이션 개발 환경을 제공한다. 구글은 이번 젬마 4를 통해 온디바이스 AI 전략도 강화했다. E2B와 E4B 모델은 스마트폰과 노트북, IoT 기기 등 엣지 환경에서 실행할 수 있도록 설계됐다. 이를 통해 클라우드 연결 없이도 AI 기능을 활용할 수 있으며 낮은 레이턴시와 개인정보 보호 측면에서도 장점을 제공한다. 특히 구글 픽셀을 비롯해 퀄컴, 미디어텍 등 주요 반도체 기업과 협력을 통해 스마트폰과 라즈베리 파이, 엔비디아 젯슨 나노 등 다양한 기기에서 오프라인 실행을 지원한다. 온디바이스 AI 확산이 가속화되는 가운데 구글이 모바일 중심 AI 경쟁력 확보에 나선 것으로 풀이된다. 오픈형 AI 모델 경쟁도 한층 치열해지고 있다. 메타의 'Llama' 시리즈를 비롯해 '미스트랄', 알리바바 'Qwen', 마이크로소프트 'Phi' 등 주요 기업들이 오픈형 모델을 잇따라 공개하며 개발자 생태계 확보 경쟁에 나서고 있다. 이에 구글 역시 젬마 4를 통해 오픈형 AI 시장에서 영향력을 확대하려는 전략으로 분석된다. 구글은 폐쇄형 모델인 제미나이를 중심으로 고성능 AI 서비스를 제공하는 동시에 젬마 시리즈를 통해 구글의 AI 전략 다변화를 진행하고 있다. 제미나이가 클라우드 기반 서비스와 기업 고객 중심이라면 젬마는 개발자와 스타트업을 대상으로 한 오픈형 생태계 구축에 초점이 맞춰져 있다. 두 모델을 병행 운영하며 AI 플랫폼 경쟁력을 강화하려는 전략으로 풀이된다. 또한 구글은 허깅페이스, vLLM, Ollama, NVIDIA NIM 등 주요 개발 도구에서 젬마 4를 출시 당일부터 지원해 개발자 생태계 확대에도 집중한다. 또한 구글 AI 스튜디오와 구글 콜랩, 버텍스 AI 등을 통해 모델 학습과 배포 환경을 확인할 수 있어 개발자는 로컬 환경뿐 아니라 구글 클라우드를 통해 확장 가능한 AI 서비스를 구축할 수 있을 것으로 전망된다. 하드웨어 지원 범위도 확대됐다. 젬마 4는 소비자용 GPU부터 엔비디아 H100 GPU, 구글 TPU까지 다양한 환경에서 실행할 수 있도록 최적화됐다. AMD GPU와의 연동도 지원하며 엣지 기기부터 클라우드까지 폭넓은 환경에서 활용 가능하도록 설계됐다. 이번 젬마 4 출시 이후 온디바이스 AI와 개방형 생태계를 중심으로 AI 시장 주도권 경쟁이 확대될 것으로 분석된다. 구글이 젬마 4를 통해 개발자 생태계 확보에 나서면서 AI 플랫폼 경쟁 구도도 한층 치열해질 전망이다.
2026-04-03 11:32:17
-
-
-
-