Nvidia announced a multi-year agreement on Friday with South Korean firm SK Hynix to secure a supply of high-bandwidth memory, a component for its artificial intelligence processors. The deal includes plans for SK Hynix affiliate SK Telecom to build a cloud computing business using Nvidia’s technology.
The agreement, valued at a potential $500 billion over its duration, includes the construction of large-scale data centers scheduled to come online in 2027, according to a company statement. The planned infrastructure is targeted to require 2 gigawatts of power.
Under the terms of the collaboration, SK Telecom will utilize Nvidia’s Vera Rubin systems to build out its cloud business. SK Hynix will provide the high-bandwidth memory (HBM) necessary for the Nvidia graphics processing units (GPUs) that will power the new systems. The demand for HBM has grown alongside the expansion of AI infrastructure. According to industry analysts, SK Hynix is a primary producer of HBM.
“The expansion will include a co-develop opportunity for us on the next-generation SK Hynix AI memory, and this will help us secure a stable supply of HBM memory,” Raj Mirpuri, Nvidia’s enterprise vice president, said in a call with reporters.
The agreement was announced at an AI summit in San Francisco, which was attended by South Korean officials.
In a separate announcement the same day, Nvidia said it would invest $1 billion into Naver, another South Korean cloud company that is building data centers using Nvidia GPUs. According to Nvidia, this project will allow potential customers to secure AI computing capacity before the planned data centers are completed.
SK Hynix, South Korea’s second most valuable company, recently listed on the Nasdaq stock exchange to raise capital for infrastructure development.
