TMTPOST --OpenAI CEO Sam Altman recently held low-profile meetings in two major manufacturers based in Taiwan, according to local news media outlet Digital Times.
Credit：IBM
Altman reportedly took a low-profile trip to Taiwan on September 30 to separately meet representatives of Taiwan Semiconductor Manufacturing Company Ltd., (TSMC) and Foxconn, mainly discussing collaboration in AI chip-making and AI server infrastructure development.
The report suggested part of Altman's discussion aims to keep up with OpenAI’s increasing hardware demand for Stargate, an artificial intelligence (AI) infrastructure joint venture founded together with Oracle and SoftBank Group Corp. The venture plans five major datacenters across the United States, and is seeking to expand into the Middle East and South Korea. SoftBank’s CEO Masayoshi Son was also reported to visit Foxconn, underscoring the company’s critical role since it is Oracle’s largest supplier of AI servers.
Altman’s meeting with TSMC, the world’s largest and most advanced chip design and manufacturing company, was deemed as part of preparation for OpenAI’s in-house chip development. OpenAI reportedly formed an AI ASIC design team last year and is working with Broadcom to develop chips using TSMC’s advanced 3nm manufacturing process.These chips, with chip-on-wafer-on-substrate (CoWoS) packaging technology combined with high-bandwidth memory (HBM), are expected to start mass production in 2026.
Altman’s reported meetings reflected OpenAI’s latest effort to intensify its AI infrastructure, which relies heavily on Taiwan’s semiconductor supply chain.
OpenAI last week inked new strategic partnerships with two South Korean giants Samsung and SK s part of OpenAI’s Stargate initiative. OpenAI said these partnerships will focus on increasing the supply of advanced memory chips essential for next-generation AI and expanding data center capacity in Korea, positioning Samsung and SK as key contributors to global AI infrastructure and supporting Korea’s ambition to become a top-three global AI nation.
Through these partnerships, Samsung Electronics and SK Hynix, the world’s top two memory chip makers, plan to scale up production of advanced memory chips, targeting 900,000 Dynamic Random-Access Memory (DRAM) wafer starts per month at an accelerated capacity rollout, critical for powering OpenAI’s advanced AI models, according to OpenAI.
Samsung and SK separately announced their Letter of Intents (IOI) signed with OpenAI in Seoul. Starting with the landmark partnership with OpenAI, Samsung said it plans to fully support Korea’s goals to become one of the world’s top three nations in AI and create new opportunities in the field.
Besides the supply of DRAM waters, another Samsung’s unit Samsung SDS has entered into a potential partnership with OpenAI to jointly develop AI data centers and provide enterprise AI services. Samsung C&T and Samsung Heavy Industries will collaborate with OpenAI to advance global AI data centers, with a particular focus on the joint development of floating data centers. These two companies will also explore opportunities to pursue projects in floating power plants and control centers, in addition to floating data center infrastructure.
SK said it will supply High Bandwidth Memory (HBM) for the construction of global AI data center under the Stargate project under the IOL. Noting that OpenAI’s monthly HBM demand for up to 900,000 DARM wafers is more than double the current industry capacity, SK hynix will establish a production system capable of promptly meeting OpenAI’s HBM demand for up to 900,000 DRAM wafers per month.
SK Telecom signed an Memorandum of Understanding (MOU) with OpenAI to collaborate on building a AI Data Center in the southwest region of Korea as part of Stargate, also known as Stargate Korea. Leveraging its solid experience in data center construction and operation, SK Telecom and OpenAI will jointly develop next-generation computing and data center solutions. They will also explore various AI service use cases for consumers and enterprises.
