Silicon Valley "Fried Chicken Talks": SK Hynix Chairman Holds Secret Meeting with Jensen Huang, Secures 55% HBM4 Market Share and Begins AI Infrastructure Collaboration

In a fried chicken shop in Silicon Valley, there may not be exquisite tableware, but there are hundreds of billions of dollars' worth of business being served.

According to Korean media reports, on February 5 local time, SK Group Chairman Chey Tae-won, who is visiting the U.S., and Nvidia CEO Jensen Huang held an informal "chimaek" (chicken and beer) meeting at a Korean-style fried chicken restaurant called "99 Chicken" in Silicon Valley.

Observers believe that the content of their discussions likely covered not only negotiations over the supply volume of sixth-generation high-bandwidth memory (HBM4), but also strategic cooperation in building next-generation AI data centers.

Interestingly, reports said the level of privacy and intimacy at this meeting far exceeded standard business negotiations. Attendees included key executives from SK Hynix, Chey Tae-won's second daughter and IntegralHealth CEO Chey Min-jung, as well as Jensen Huang's daughter, Madison Huang, Senior Director of Nvidia's Robotics Division.

Analysts believe this “family-style” social scene sends a strong signal: the relationship between SK and Nvidia is upgrading from business partners to closer strategic allies.

Looking back through history, whenever Chey Tae-won and Jensen Huang meet, the semiconductor market is shaken. Their meeting in May 2021 directly led to the "AI Triangle Alliance" between SK, Nvidia, and TSMC.

This time, industry insiders believe Chey Tae-won's trip is not only about pledging on-time delivery of HBM4, but is also aimed at SK Group's strategic transition towards becoming a “comprehensive AI solutions provider.” An industry insider commented:

This signals SK Group's formal entry into the next-generation AI infrastructure market, using HBM as a lever.

HBM4 Supply Structure: SK Hynix Pledges Unimpeded Delivery

The focus of this meeting was assurance on the supply of HBM4. Nvidia’s next-generation AI accelerator "Vera Rubin" scheduled for release later this year will use HBM4 modules with a capacity of 288GB each.

As HBM production takes about four months, plus two to three months for packaging at TSMC, the entire cycle lasts six to seven months—making Nvidia more dependent on SK Hynix, which commands the largest capacity in the industry.

At the end of last year, SK Hynix reached an agreement with Nvidia to supply more than 55% of HBM4 demand and is currently optimizing performance. During the meeting, Chey Tae-won pledged "unimpeded supply" to Jensen Huang.

Reports indicate that while SK Hynix's HBM4 uses 12nm foundry and 1b DRAM (a relatively older technology), its performance matches that of Samsung Electronics' products, which use a 4nm foundry and 10nm 6th-gen DRAM.

The market landscape has shifted this year. Samsung Electronics passed Nvidia’s quality test for its 12-layer HBM3E products last September and became the first this month to begin mass producing and shipping HBM4.

Samsung's HBM4 transmission speed is 11.7Gbps, which exceeds Nvidia's standard requirement of 10-11Gbps, breaking SK Hynix's near-monopoly in the HBM3E 12-layer product segment last year.

Industry expectations suggest the two sides also discussed cooperation on seventh-generation HBM (HBM4E) to be officially launched in 2027, as well as custom HBM (cHBM) solutions.

From “Selling Memory” to “Selling AI Infrastructure”

If supplying HBM4 without barriers is about “defending the kingdom,” then Chey Tae-won's real ambition on this trip is to “expand the kingdom” into the AI data center infrastructure market.

Chey Tae-won is driving SK Group’s transformation into a "comprehensive AI solutions provider." His talks with Jensen Huang are believed to involve cooperation in AI semiconductors, servers, and data centers across multiple dimensions.

Further collaboration between SK Hynix and Nvidia in the enterprise solid-state drive (eSSD) space is getting attention.

In January, Nvidia announced its "Vera Rubin" would include a new memory solution called "ICMS," with each system equipped with 9,600TB of eSSD—16 times more demand than existing products. This represents a major opportunity for SK Hynix’s expansion in the storage sector.

Since February 3, Chairman Chey has been in the U.S., holding a series of meetings with major tech companies including Nvidia and Meta.

Reports also suggest the parties may have discussed next-generation server memory modules—SoCamm (low-power DRAM modules for servers)—and flash storage after HBM. Semiconductor industry insiders point out:

SoCamm is the next battlefield for changing AI server power architecture. SK Group is formally entering the next-generation AI infrastructure market, using HBM as a powerful lever.

SK Hynix has renamed its U.S. flash memory subsidiary Solidigm to "AI Company," making it the platform responsible for SK Group’s AI investment and solutions business.

The business scope will not only cover SK Hynix’s AI semiconductor operations, but also integrate SK Telecom’s AI technology and solution capabilities, establishing a full chain of services from AI data center design to semiconductor and server delivery. There are even plans to launch demonstration projects for data centers in North America this year.

This meeting exhibited signs of Chey Tai-won pitching his strategic vision to Nvidia. Industry observers believe the two sides may have explored specific plans for SK Group to provide comprehensive AI solutions to Nvidia.

This informal meeting at 99 Chicken could become the starting point for another major restructuring of the global semiconductor supply chain, following the 2021 “AI Triangle Alliance.”

Risk Warning and DisclaimerThe market entails risks and investments should be made with caution. This article does not constitute personal investment advice and does not take into account individual users’ particular investment goals, financial status, or needs. Users should consider whether any views, opinions, or conclusions expressed in this article fit their specific circumstances. Investment is at your own risk.