Altman secretly visited Hon Hai Precision Industry Co., Ltd. and Taiwan Semiconductor: Discussing details of the Stargate cooperation and self-developed ASIC chip project

Recently, OpenAI CEO Sam Altman was reported to have made a low-profile visit to Hon Hai and Taiwan Semiconductor.

According to a report by Taiwan's Economic Daily citing industry sources, Altman arrived in Taiwan on September 30 and held talks with Hon Hai and Taiwan Semiconductor respectively. The core topics of discussion focused on OpenAI's two strategic pillars: the "Star Gate" project and finalizing the details of its self-developed AI chip ASIC's foundry production.

Neither Taiwan Semiconductor nor Hon Hai has responded to this meeting.

Hon Hai Becomes a Key Partner for "Star Gate"

One of Altman's primary goals during this visit was to ensure that the ambitious "Star Gate" project has a stable hardware foundation. Reports indicate that the project plans to build five new data centers in the United States, making it one of the largest investments in computing power in the country.

Sources revealed that Hon Hai is the largest AI server supplier in the "Star Gate" plan. The project's partners include Oracle and SoftBank, and Hon Hai happens to be an important partner for both companies' AI servers and data centers.

Therefore, Altman's discussions with Hon Hai mainly revolved around ensuring the production and delivery capabilities of AI servers to support the massive infrastructure needs of "Star Gate."

Accelerating Self-Developed Chips, Targeting Taiwan Semiconductor

In addition to servers, another major focus of Altman's visit was OpenAI's self-developed AI chip plan. According to supply chain news, OpenAI is actively developing its own customized AI chips (ASIC) to reduce its high dependence on external suppliers like NVIDIA.

Reports indicate that OpenAI has recruited a team that was previously responsible for AI chip design at Google and plans to have the chip design results produced by Taiwan Semiconductor. The focus of the discussions between the two parties is believed to be the specific details of chip production at Taiwan Semiconductor, with the goal of achieving mass production by 2026.

Supply chain analysis suggests that the initial version of OpenAI's self-developed chips may primarily be used for inference rather than training, but their successful mass production will be an important validation of whether tech giants can replace existing market solutions with self-developed chips