SK Group's Chey meets TSMC chief for first time in 2 years, deepening AI chip alliance [Computex 2026]

SK Group Chairman Chey Tae-won met with TSMC Chairman C.C. Wei in Taiwan on Wednesday, their first official encounter in two years. The two leaders shared views on next-generation AI technology trends and held in-depth discussions on how to lead the future AI ecosystem, drawing on the strong partnership between their companies.

The meeting was the first since June 2024 and reaffirmed the deep trust the two companies have built over time. They agreed to further strengthen broad-based cooperation spanning next-generation HBM development and advanced packaging, positioning themselves to respond swiftly to shifts in the global AI market.

With resolving supply bottlenecks in the global AI value chain emerging as a critical challenge, SK Hynix's industry-leading AI memory chip technology combined with TSMC's foundry capabilities is expected to offer a practical solution.

SK Hynix's HBM4 — its sixth-generation HBM — destined for Nvidia's next-generation AI accelerator Vera Rubin, uses TSMC's 12-nanometer base die and fifth-generation 10-nanometer-class DRAM (1b) process.

Industry observers expect the SK Hynix-TSMC collaboration to help stabilize the AI semiconductor supply chain, as easing those bottlenecks remains a top priority across the global AI value chain.

Going forward, the two companies plan to accelerate their push to capture the custom AI memory chip market tailored to the diverse demands of global big-tech firms. Through its partnership with TSMC, SK Hynix aims to cement its market leadership by delivering top-performing products on time for the AI era.

Making his first appearance at Computex 2026 — Asia's largest IT trade show, held in Taipei — Chey said Tuesday that "the memory bottleneck is expected to continue until 2030" and pledged to "double total wafer production capacity within the next five years."

Chey said the memory bottleneck would persist for at least another five years, adding that the company is "pushing capacity expansion at full speed."

"Building a new memory fab requires not only enormous investment but also a minimum of three years," he said. "Despite all these challenges, we plan to double our wafer production capacity over the next five years."

SK Hynix is channeling large-scale investment into its M15X and P&T7 facilities in Cheongju, its semiconductor cluster in Yongin, and an advanced packaging factory in Arizona to bolster its production capabilities.

SK Hynix unveiled samples of its HBM4E — seventh-generation HBM — for the first time at its Computex exhibition booth Wednesday.

Nvidia CEO Jensen Huang visited the SK Hynix booth and signed both an HBM4E wafer and a 192-gigabyte SOCAAM2. On the HBM4E wafer, he also left the message "Please Make More."

On the development timeline for HBM4E, Chey said: "Currently there is only one customer for HBM4E, so it entirely depends on the customer's schedule. We need to be ready whenever the customer is ready — and we will be."

Chey also outlined plans to build AI factories in collaboration with affiliates including SK Telecom. "We are producing memory chips for AI, but looking ahead, we are taking on the challenge of how we can build 'AI factories,'" he said. "In the future, many AI factories capable of producing more artificial intelligence will be needed, and because I believe this will benefit all of humanity, I want to build more AI factories."

On the three-way alliance with Nvidia and TSMC, Chey expressed confidence, saying it is "better than ever." He added: "We are collaborating with TSMC on the HBM4 base die. We rely on TSMC and have the best partnership we have ever had."

젠슨 황, 이르면 오늘 입국…피지컬AI 협력 위한 나흘간의 폭풍행보 예고

https://biz.heraldcorp.com/article/10763314?sec=027