Recent reports suggesting that TSMC has invested in OpenAI are inaccurate. The companies maintain a manufacturing partnership rather than a financial one, with TSMC serving as a foundry for OpenAI's custom AI chips.
OpenAI is currently developing its first-generation inference chip, codenamed Jalapeño. The processor is scheduled for initial production using TSMC’s 3 nm manufacturing process, with mass production expected to begin in 2026.
The company is also planning a second-generation chip design based on TSMC’s more advanced A16 technology. To secure necessary manufacturing capacity, OpenAI has partnered with Broadcom.
OpenAI CEO Sam Altman held meetings with TSMC Chairman C.C. Wei and other suppliers, including Foxconn, in Taiwan during late October and early November 2026 to discuss infrastructure scaling.
While OpenAI is developing custom silicon to reduce reliance on off-the-shelf accelerators, it continues to collaborate with Nvidia and AMD on multi-gigawatt data center projects.
