自從美國總統川普政府開始對中國實施嚴格的半導體與高科技出口限制以來，美中之間的科技競爭不僅愈演愈烈，也讓曾在中國市場幾乎一家獨大的美國晶片巨頭英偉達（NVIDIA）陷入前所未有的困境。這場由政治驅動的科技封鎖，讓英偉達在短短幾年內，從中國先進AI晶片市場的「霸主地位」急轉直下，如今市佔率幾乎歸零。

英偉達創辦人暨執行長黃仁勳（Jensen Huang）近日罕見地公開表達對美國政府政策的不滿。他在美國城堡證券（Citadel Securities）於紐約舉辦的一場論壇上直言，美國對中國的技術封鎖政策，最終不僅會傷害中國，更會對美國本身造成深遠且負面的影響。黃仁勳警告說：「美國決策者希望自己的國家能在AI競賽中取勝，但重要的是要記得，傷害中國的事情，往往也可能傷害美國，甚至更嚴重。」

這段談話於10月15日由城堡證券官方釋出影片後引發熱烈討論，標題為《英偉達的黃仁勳談人工智慧與下一個增長前沿》。影片中，紅杉資本（Sequoia Capital）合夥人康斯坦丁·布勒（Konstantine Buhler）向黃仁勳提問，美國應如何在AI運算力出口政策上找到正確平衡。黃仁勳回應時指出，美國在全球AI生態系中若採取過度排他性的策略，不僅會抑制中國的創新，同樣會削弱美國科技產業的長期競爭力。

他進一步強調，軟體與人工智慧產業的核心並非硬體設備本身，而是「開發者生態」。他直言，要掌握未來科技發展的主導權，必須贏得全球開發者的心與信任。然而，美國的政策卻正在阻止來自中國的大量優秀AI研究者與工程師使用美國的技術與平台，這無異於「自斷手腳」。黃仁勳指出：「中國擁有全球約50%的AI研究人員，擁有優秀的學術機構，對人工智慧懷抱極高熱情。如果不讓他們在美國技術架構上進行開發，那將是一個重大的錯誤。」

黃仁勳的這番言論，不僅反映出英偉達在中美科技戰中的尷尬處境，也揭示了企業界對華府政策的不安。過去數年，美國政府不斷升級對中國的晶片出口限制，要求企業不得向中國出口高階GPU晶片與AI運算設備，使得英偉達不得不重新設計性能較低、符合出口規定的特製版本。然而，這些產品在中國市場的競爭力明顯不足，導致英偉達的市佔率從巔峰時期的95%急跌至接近零。

在中國市場被迫退場後，英偉達的策略重心逐漸轉向其他地區與雲端運算企業，但中國這個全球第二大AI市場的流失，仍是公司發展的一大隱憂。黃仁勳的發言不僅是對現行政策的警告，也是一種呼籲——希望美國政府在科技競爭中能「理性思考」，不要讓短視近利的政治手段摧毀了整個產業的全球合作基礎。

整體而言，這起事件再次凸顯出美中科技戰的深層矛盾：美國試圖維持技術優勢，卻可能因封鎖與孤立而削弱自身的創新活力；而中國則在加速自主研發的同時，也在尋求新的供應鏈與技術突破。黃仁勳的話語，或許正是一個警鐘——在人工智慧這場全球性競賽中，真正的勝者，將是懂得合作與開放的那一方。

Since the beginning of this year, U.S. chip giant NVIDIA has found itself in a volatile and challenging situation in China, largely due to the Trump administration’s unpredictable and hardline export restrictions on advanced chips to the country. What was once a dominant position—controlling about 95% of China’s high-end AI chip market—has now collapsed to virtually zero.

Recently, NVIDIA founder and CEO Jensen Huang publicly voiced his frustration with Washington’s policies, warning that U.S. technological sanctions not only “hurt China” but “often hurt America, sometimes even more.” His remarks came during a fireside chat at an event hosted by Citadel Securities in New York earlier this month. On October 15, Citadel released a video of his conversation with Sequoia Capital partner Konstantine Buhler, titled “NVIDIA’s Jensen Huang on AI and the Next Frontier of Growth.”

During the discussion, Buhler asked Huang what the right approach should be for the U.S. when it comes to exporting AI computing power to China. Huang replied that while policymakers want America to win the AI race, they must remember that “hurting China often hurts America, sometimes even worse.” He added that before rushing to enact policies that harm others, leaders should pause and “reflect on what policies actually benefit America.”

Huang emphasized that the heart of the software and AI industry lies not in hardware but in developers—the innovators who build future platforms. To dominate the next technological era, the U.S. must win over the global developer community. However, he noted, current U.S. restrictions are preventing large numbers of talented Chinese AI researchers and engineers from using American technologies, which he called a strategic mistake. “China has about 50% of the world’s AI researchers, excellent universities, and enormous enthusiasm for AI,” Huang said. “Not allowing them to build on American technology is a major error.”

His comments reflect NVIDIA’s increasingly precarious position in the U.S.–China tech war and the growing unease within the American business community. Over the past several years, Washington has repeatedly tightened export controls on advanced GPUs and AI computing chips, forcing NVIDIA to redesign downgraded versions of its products to meet regulatory limits. These weakened versions, however, failed to compete effectively in China’s rapidly advancing market, causing NVIDIA’s dominance to vanish almost overnight.

With its presence in China now drastically reduced, NVIDIA has shifted focus toward other regions and global cloud service providers. Still, losing access to the world’s second-largest AI market remains a significant blow. Huang’s remarks serve as both a warning and an appeal—urging U.S. policymakers to think strategically rather than politically, and to avoid undermining global technological collaboration in pursuit of short-term national advantage.

Ultimately, this incident underscores the deeper tension at the core of the U.S.–China tech rivalry: America seeks to preserve its technological supremacy, but excessive restrictions risk eroding its own innovation ecosystem; China, meanwhile, is accelerating its self-sufficiency drive and investing heavily in domestic R&D. As Huang suggested, in the global AI race, the real winners will be those who choose cooperation and openness over isolation and control.