NVIDIA and SK Hynix Join Forces to Secure Next-Generation AI Memory

Source: Zhitong Finance
On June 7, NVIDIA and SK Hynix, the world's second-largest memory chip manufacturer, announced a multi-year technology partnership. The two companies will jointly develop next-generation memory products required for building global AI factories and deeply integrate AI technologies into semiconductor chip design and manufacturing. This collaboration directly addresses ongoing market concerns regarding storage supply bottlenecks amid the expansion of AI infrastructure.
Jensen Huang Urgently Clarifies HBM Supply; US Stock "Black Friday" Sends Global Tech Stocks Tumbling
The announcement came at a highly sensitive time for the market. On June 4, overseas research firm SemiAnalysis published a report alleging that NVIDIA’s new Vera Rubin AI server platform would feature downgraded memory. Combined with persistent market concerns over tight HBM supplies, this triggered a sharp pullback in the memory chip sector, sending Micron Technology plunging by over 10% on the same day.
In response to market anxieties, NVIDIA CEO Jensen Huang, who was visiting South Korea, quickly addressed the issue on June 5. He explicitly debunked the rumors during an interview, emphasizing that NVIDIA "will utilize massive amounts of HBM memory." He also confirmed that three major suppliers—Samsung Electronics, SK Hynix, and Micron Technology—have all passed qualification and entered mass production to ensure supply for the Vera Rubin platform. Huang further revealed that Vera Rubin chips are already in full production and are expected to begin deliveries in the third quarter of this year.
However, that same evening, US stocks experienced a "Black Friday." A combination of rising expectations for Federal Reserve interest rate hikes, the lingering fallout from the SemiAnalysis report, and Broadcom's disappointing earnings caused the Philadelphia Semiconductor Index to plunge by more than 10% in a single day, marking its largest daily drop since March 2020. NVIDIA plummeted 6.2%, wiping out approximately $328.3 billion (about RMB 2.2 trillion) in market value overnight, leading the tech stock selloff. Meanwhile, SK Hynix had already seen its shares tumble significantly earlier in the South Korean trading session.
NVIDIA and SK Hynix Join Forces to Secure Next-Generation AI Memory
Against this backdrop, the partnership announcement sent a much clearer strategic signal across the industrial chain. According to NVIDIA’s official statement, the core of this multi-year agreement lies in deeply aligning SK Hynix’s memory R&D roadmap with NVIDIA’s AI infrastructure roadmap. They will co-develop next-generation memory products tailored for NVIDIA’s Vera Rubin AI supercomputer, Vera CPU, RTX Spark personal computers, and Jetson Thor robot computing platforms, covering three major market directions: AI infrastructure, personal AI, and physical AI.
Beyond joint memory development, the collaboration extends into semiconductor design and manufacturing. SK Hynix will leverage NVIDIA’s CUDA-X libraries and PhysicsNeMo framework to accelerate semiconductor simulation, TCAD (Technology Computer-Aided Design) workflows, and internal engineering code efficiency. They will also explore tripartite collaboration models with EDA (Electronic Design Automation) software vendors.
In the realm of smart manufacturing, SK Hynix will integrate NVIDIA’s Omniverse platform, OpenUSD workflows, and cuOpt decision optimization engines to build digital twin systems for its wafer fabs, laying the foundation for fully autonomous fab operations. Both parties will also explore connecting these digital twins with legacy software and agentic AI workflows to drive automation and intelligence in manufacturing decisions.
NVIDIA emphasized in its statement that the agreement explicitly supports the secure supply of advanced memory. It aims to tackle structural challenges such as long development cycles, complex manufacturing processes, and capital-intensive investments, ensuring that memory supply keeps pace with the continuous global expansion of AI factories.
Jensen Huang stated that AI factories are the core engine of the new industrial revolution, and high-end memory is critical to safeguarding their performance. He noted that SK Hynix has always been an exceptional partner for NVIDIA, and together they will develop next-generation memory for AI factories. Chey Tae-won, Chairman of SK Group, remarked that SK Hynix and NVIDIA have been accumulating efforts toward this goal for years, and this partnership reflects the profound depth of their collaboration.
Jensen Huang Announces Entry into the PC Chip Market
Previously, on June 1 at the GTC Taipei 2026 conference, Jensen Huang announced NVIDIA’s official entry into the personal computer chip market with the launch of a brand-new processor. This move aims to break Intel's monopoly in the sector and adapt PC devices to the demands of the AI era.
Huang stated that starting this autumn, mainstream PC brands like Dell and Lenovo will sequentially launch laptops and desktops equipped with the RTX Spark super chip. Co-developed by NVIDIA and MediaTek, this product integrates a CPU and GPU, capable of running Microsoft’s Arm-based Windows system. The new product will transform how personal computers operate; users will simply state their needs, and the computer will handle the rest. Commenting on the current global AI boom, Huang asserted that the AI era has truly arrived.

英伟达(NVDA.US)携手SK海力士 锁定下一代AI存储
来源：智通财经网

6月7日，英伟达(NVDA.US)与全球第二大存储芯片制造商SK海力士宣布达成多年期技术合作伙伴关系，双方将围绕全球AI工厂建设所需的下一代存储产品展开联合研发，同时将AI技术深度应用于半导体芯片的设计与制造环节。这一合作落地，直接回应了市场对 AI 基础设施扩张过程中存储供应瓶颈的持续担忧。

黄仁勋紧急辟谣HBM供应 美股“黑五”引发全球科技股巨震

此次合作公告发布恰逢市场高度敏感时刻。6月4日，海外调研机构SemiAnalysis发布一则关于英伟达新一代Vera Rubin AI服务器平台存在内存降配的研报，该报告叠加市场对HBM供应紧张的持续担忧，导致存储芯片板块率先回调，美光科技当日应声暴跌超10%。

面对市场疑虑，正在韩国访问的英伟达CEO黄仁勋于6月5日迅速出面回应。他在接受采访时正面辟谣，强调英伟达“将使用大量的HBM内存”，并确认三星电子、SK海力士和美光科技三家主要供应商均已通过资格认证并投入量产，竞相支持Vera Rubin平台的供货保障。黄仁勋还透露，Vera Rubin芯片已全面投产，预计今年第三季度开始交付。

然而，同日晚间，美国股市遭遇“黑色星期五”——美联储加息预期升温、SemiAnalysis报告持续发酵以及博通财报不及预期等多重利空叠加共振，费城半导体指数单日重挫逾10%，创2020年3月以来最大单日跌幅。英伟达单日暴跌6.2%，市值一夜蒸发约3283亿美元（约合人民币2.2万亿元），成为领跌科技股的龙头之一。SK海力士股价则在韩国市场当日已率先大幅下挫。

英伟达与SK海力士携手 锁定下一代AI存储

在此背景下，这份合作公告释放了更为明确的产业链战略信号。根据英伟达官方公告，此次多年期协议的核心在于将 SK 海力士的存储研发路线图与英伟达的 AI 基础设施路线图深度绑定。双方将联合开发面向英伟达 Vera Rubin AI 超级计算机、Vera CPU、RTX Spark 个人电脑以及 Jetson Thor 机器人计算平台的下一代存储产品，覆盖 AI 基础设施、个人 AI 及物理 AI 三大市场方向。

除存储产品联合研发外，双方合作还延伸至半导体设计与制造环节。SK 海力士将采用英伟达 CUDA-X 库及 PhysicsNeMo 框架，加速半导体仿真、技术计算机辅助设计（TCAD）工作流及内部工程代码的运算效率，并探索与电子设计自动化软件供应商的三方协作模式。

在智能制造领域，SK 海力士将结合英伟达 Omniverse 平台、OpenUSD 流程及 cuOpt 决策优化引擎，构建晶圆厂数字孪生体系，为实现全自主晶圆厂运营奠定基础。双方还将探索将数字孪生与现有遗留软件及智能体 AI 工作流相连接，推动制造决策的自动化与智能化。

英伟达方面在声明中强调，协议明确支持先进存储的供应保障，旨在应对先进存储产品开发周期长、制造工艺复杂、资本投入密集等结构性挑战，确保存储供应能够跟上全球 AI 工厂持续扩张的节奏。

黄仁勋表示，AI工厂是新一轮工业革命的核心动力，高端存储是保障其性能的关键。SK海力士一直是英伟达极为出色的合作伙伴，双方将携手为AI工厂研发下一代存储产品。SK集团董事长Chey Tae-won则表示，SK 海力士与英伟达多年来一直在为此积累，这一合作伙伴关系体现了双方合作的深度。

黄仁勋宣布进军PC芯片市场

此前6月1日，在GTC Taipei2026大会上，黄仁勋宣布英伟达正式进军个人电脑芯片市场，推出全新处理器，意在打破英特尔在该领域的垄断地位，并推动PC设备适配人工智能时代的发展需求。

黄仁勋表示，今年秋季起，戴尔(DELL.US)、联想(00992)等主流PC品牌将陆续推出搭载RTX Spark超级芯片的笔记本及台式机。这款由英伟达联合联发科共同研发的产品，集成了处理器与显卡，可运行微软Arm架构版Windows系统。新产品将改变以往个人电脑的工作方式，用户只需提出需求，剩下的都将由电脑来完成。针对眼下全球AI产业的狂欢，黄仁勋认为，AI时代已经真正到来。