OpenAI’s Head of Compute Infrastructure, Sachin Katti, says artificial intelligence will soon be designing the very chips and systems it needs to run.
The semiconductor supply chain that powers OpenAI’s ambitions is the same one that crypto miners, blockchain validators, and GPU-hungry DeFi protocols compete for. When the largest AI company on the planet starts building its own chips, the ripple effects hit every industry that depends on advanced silicon.
From Nvidia dependency to custom silicon
Katti, a former Intel executive, joined OpenAI in November 2025 to lead the company’s compute infrastructure efforts. His mandate is clear: reduce OpenAI’s reliance on off-the-shelf Nvidia GPUs while scaling compute capacity to meet what the company describes as demand in the tens of gigawatts.
The custom chip journey started in October 2025, when OpenAI partnered with semiconductor giant Broadcom to co-design specialized AI accelerators. That collaboration bore fruit in June 2026 with the unveiling of Jalapeño, OpenAI’s first custom chip built specifically for optimizing large language model inference workloads.
