Nvidia CEO Huang confirms Vera Rubin in full production with Korean HBM chips
Seoul, June 1
Nvidia Chief Executive Jensen Huang said Monday that the company's next artificial intelligence accelerator, Vera Rubin, was in full production, confirming at his GTC Taipei keynote that high-bandwidth memory from Samsung Electronics, SK hynix, and Micron will go into the platform.
According to a news report by The Korea Herald, "The three suppliers were already expected. What the keynote, held alongside the Computex trade show in Taipei, settled is that the chip has moved from road map to volume production, which fixes the order each maker has actually won."
As per the news report, the order leaves SK hynix in front, with industry estimates putting it at roughly 60 to 70 per cent of Nvidia's HBM4 volume for the Rubin launch, the position it built supplying the current generation. Samsung's share is estimated at 25 to 30 per cent, with Micron taking a smaller, supplementary share.
For Samsung, that is still ground regained. After falling behind SK hynix on HBM, it cleared Nvidia's qualification tests at the 10 and 11 gigabit-per-second data rates the Rubin design demands and began shipping its 12-layer HBM4 early this year. Securing a quarter or more of the order on the most important AI platform marks a recovery from the rounds it lost.
High-bandwidth memory, the stacked memory that feeds data to AI processors, is the part Korea's two chipmakers compete hardest to sell, and the money at stake keeps climbing. Citing Morgan Stanley, the report estimated that memory now accounts for a share of the parts cost of a Rubin rack that is 435 per cent higher than in the previous generation.
The contest is already moving on. Samsung has begun shipping early samples of HBM4E, the seventh-generation memory aimed at the Vera Rubin Ultra due in late 2027.
Huang travels to Seoul this week to meet the heads of Korea's largest conglomerates, with SK Group Chairman Chey Tae-won expected to discuss memory among other topics.
— ANI
Reader Comments
The scale of investment here is mind-boggling. A 435% increase in memory cost share per rack? These semiconductor companies are minting money. And Samsung getting back into the game after being left behind earlier is a great comeback story. But I wonder about the geopolitical angle... what happens if there's a Taiwan contingency? India should position itself as an alternative hub. Our engineers are world-class, but the infrastructure needs a massive upgrade. 🤔
While everyone is excited about AI accelerators, I'm a bit concerned about the cost implications. These ultra-expensive Rubin racks will be used by big tech companies for training massive models. But what about inference at the edge? What about making AI accessible to smaller Indian startups? We're in danger of creating another digital divide. Nvidia is making billions, but the average user may never benefit from this hardware directly.
As someone working in the chip industry, this is huge news. The move from roadmap to volume production is the critical inflection point. SK Hynix maintaining 60-70% share is impressive, but Samsung clawing back to 25-30% after their HBM3 stumbles is a testament to Korean engineering. Jensen will have a busy week in Seoul. India needs to watch this closely and learn from how Korea built its memory ecosystem over decades.
It's incredible how quickly the technology is moving. HBM4E samples are already shipping while HBM4 production just started! But I do wonder about the environmental impact. These memory chips require massive amounts of energy to produce and operate. As AI adoption grows in India, we need to think about sustainable computing. Green data centers, water recycling in fabs, and renewable energy for AI workloads should be part of the conversation.
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.