"This is very difficult, but I believe in you!" Last week, Jensen Huang hosted a banquet for SK Hynix engineers, personally proposing a toast and urging for "on-time delivery of HBM4 without delay."

"This is very difficult, but I believe in you!" Last week, Jensen Huang hosted a banquet for SK Hynix engineers, personally proposing a toast and urging for "on-time delivery of HBM4 without delay."

Nvidia CEO Jensen Huang personally hosted a dinner for the engineering teams of its partners last week, a rare move that highlights the strategic importance of the next-generation high-bandwidth memory HBM4 to the AI chip giant. As Samsung becomes the first to ship in the HBM4 race, whether SK Hynix can deliver high-performance products on time will directly affect the market performance of Nvidia’s new-generation AI accelerator Vera Rubin, which is set to launch in the second half of this year.

On the evening of February 14, Jensen Huang appeared at a Korean fried chicken restaurant near Nvidia’s headquarters, spending about two hours mixing soju-beer and toasting 30-plus engineers from SK Hynix and Nvidia at each table. He repeatedly emphasized "We are one team" and "I am proud of you," urging the engineers to "deliver extraordinary results through relentless challenges and effort," with special mention of the sixth-generation HBM4 products promised by SK Hynix.

 

This impromptu gathering sent a clear signal: HBM4 is seen by Nvidia as a key differentiating component of the Vera Rubin accelerator. Nvidia has required HBM4 suppliers to meet specs of "over 11 Gbps operating speed" and "over 3.0 TB/s bandwidth," more than 30% higher than competitor AMD’s requirements for similar products. Industry analysts believe Jensen Huang's personal attendance increases SK Hynix’s likelihood of maintaining its leading supplier status for HBM4.

According to industry sources, Nvidia allocated its HBM supply share for this year in early December: SK Hynix received over 55%, Samsung Electronics took between just over 20% to nearly 30%, and Micron Technology about 20%. Although Samsung’s technical progress could shift the shares, it is widely believed that SK Hynix will secure the largest allocation after completing quality optimizations in the first quarter.

A Rare "Engineer Diplomacy"

The semiconductor industry regards Jensen Huang’s personal hosting of a dinner for partner engineers as highly unusual. The event was quickly arranged after Huang instructed Nvidia staff last week to "organize a dinner to encourage SK Hynix HBM engineers," underscoring the critical importance of SK Hynix’s HBM4 to Nvidia’s future business.

Huang arrived at the 99 Chicken restaurant in Santa Clara around 5:20 pm. During the final toast at the end of the dinner, he said: "AI accelerators and HBM4 represent extraordinary, world-leading challenges in technology. I am proud of all of you working day and night, and I believe you will deliver outstanding results." He added:

"I know the schedule for HBM4 and Vera Rubin development is tight, but I believe in you. Now is the great moment for SK Hynix and Nvidia to show the world together."

SK Hynix officially entered Nvidia’s supply chain in July 2020 with HBM2E (third-generation) products, later becoming the de facto exclusive supplier for HBM3 (fourth-generation) and HBM3E (fifth-generation), forming a close "AI Semiconductor Tri-Alliance" with TSMC.

Technical Barriers and Time Pressure

HBM4 is regarded as a key component that will determine the performance of Nvidia’s next-generation AI accelerator Vera Rubin. Planned for launch in the second half of this year, HBM4 is a high-performance memory module made from stacking 12 advanced DRAM chips, responsible for delivering massive amounts of data to the processing GPU in real-time.

Nvidia’s requirements for HBM4 far exceed those of its competitors. Its specs for operating speed and bandwidth are over 30% higher than AMD’s requirements for HBM4, effectively positioning HBM4 as the critical differentiator for Vera Rubin.

Unlike the HBM3E market, which is mainly monopolized by SK Hynix, the upcoming HBM4 market set to open in the second half of the year will feature a new competitive landscape. On February 12, Samsung shipped its first official HBM4 products to Nvidia, becoming the industry's first—with speeds at 11.7 Gbps (up to 13 Gbps) and a bandwidth of 3.3 TB/s.

SK Hynix has already ensured HBM4 performance reaches 11.7 Gbps or above, and is providing Nvidia with paid bulk samples while carrying out performance optimizations. Industry experts expect SK Hynix will soon receive formal "mass supply" approval from Nvidia.

At the dinner, Jensen Huang urged SK Hynix engineers to "deliver top-performing HBM4 without delay." This was interpreted as a clear requirement for the supply timeline.

HBM Share Battle Heats Up

As Samsung advances in the HBM4 technology race, this year’s supply share allocation may see changes. Still, the industry widely believes SK Hynix remains likely to secure the largest allocation after optimizing quality in the first quarter.

A semiconductor industry source explained to the media:

"Final HBM4 optimization work by all three memory makers won’t be done until around March. There’s recently been a trend to prioritize increasing general DRAM production for profitability, rather than competing for HBM market share."

According to a previous Wallstreetcn article, Evercore’s research pointed out differences in generational product evolution, with some industry insiders distinguishing between "minor" (small upgrades) and "major" (significant upgrades) transitions. The shift from H100 to B100/B200 is considered minor, while moving to GB-Series rack-level systems is major; going from GB-Series to VR-Series is minor, but the transition to Rubin Ultra, which expands cluster scale from 144 to 576, is a major upgrade.

This dinner hosted personally by Jensen Huang was both a recognition of a longtime partner and a push to secure a vital supply chain link. For SK Hynix, whether it can maintain its lead in fierce competition depends on execution in the coming months.

Risk Disclosure and DisclaimerThe market carries risks, and investments require caution. This article does not constitute individual investment advice, nor does it consider specific users’ unique investment goals, financial circumstances, or needs. Users should evaluate whether any opinions, viewpoints, or conclusions in this article suit their particular situation. Investing accordingly is at your own risk.