- OpenAI partners tech giants for 2026 custom chip launch
- Company adds AMD to reduce Nvidia dependence
- Hardware strategy targets cost and supply chain challenges
Silicon Dreams Take Shape
OpenAI partners Broadcom and TSMC to design its first custom AI chip, targeting a 2026 launch.
The company shifted away from earlier plans to build manufacturing facilities, choosing instead to focus on chip design expertise.
Breaking the Nvidia Chain
In a strategic pivot, OpenAI expands beyond Nvidia GPUs by embracing AMD chips through Microsoft Azure for AI training.
The move responds to ongoing chip shortages and mounting training costs that challenge the industry.
Power Play in Progress
The company’s dual strategy – developing custom chips for deployment and diversifying training hardware – signals OpenAI’s push for greater independence in the AI infrastructure space.