OpenAI to launch first AI chip with Broadcom in 2026

OpenAI is partnering with Broadcom to develop its first AI chip, which is expected to be ready in 2026.

The chip will reportedly be used for OpenAI’s internal operations and not sold to external customers.

OpenAI needs massive computing power for training and running its AI models, including ChatGPT.

The company has also worked with Broadcom and TSMC on chip development while still relying on AMD and Nvidia.

OpenAI aims to diversify its chip supply and cut costs, following the path of other tech giants like Google, Amazon, and Meta.

🔗 Source: Reuters

🧠 Food for thought

Implications, context, and why it matters.

Broadcom’s AI revenue surged 63% year-over-year in Q3 2025, largely driven by partnerships with companies like OpenAI seeking custom chip solutions2.
The company secured over $10 billion in AI infrastructure orders from new customers, with CEO Hock Tan expecting AI revenue growth to “improve significantly” for fiscal 20261.
Broadcom’s revenue forecasts highlight the opportunity, with expectations of capturing between $60 billion and $90 billion by fiscal 2027 as more AI companies move away from standard solutions2.
This positions Broadcom as a critical intermediary in the semiconductor ecosystem, enabling AI companies to develop custom hardware without building their own fabrication capabilities.

OpenAI’s 2026 timeline for its first custom chip using TSMC’s 3nm process demonstrates how AI companies are prioritizing hardware control alongside software development2.
Google has already leveraged machine learning to design its next-generation Tensor Processing Units in less than six hours—faster than traditional human-led design processes3.
This follows a broader pattern where tech giants like Apple, Tesla, and Amazon have developed custom chips to optimize performance for their specific workloads rather than relying on general-purpose solutions4.
The trend reflects companies’ desire to reduce dependence on external suppliers like Nvidia while gaining more control over their hardware roadmaps and cost structures.