Recognized as a leader in AI, OpenAI is making serious efforts to reduce its dependence on Nvidia AI accelerators. The company plans to finalize the design of its own chip in the coming months and begin production at TSMC factories using the latest technological processes.
According to analysts, “OpenAI is on track to achieve its ambitious goal of mass production at TSMC facilities in 2026.” A key step in this process is Tape-out—the transfer of a chip’s digital design to a photomask for production. This stage typically costs tens of millions of dollars and takes up to six months before the first chip is released. If issues arise, diagnosing the problem and repeating the process can be costly and time-consuming.
Strategic Goals and Future Developments
OpenAI sees its AI accelerator as a strategic tool to strengthen negotiations with chip suppliers. If the initial rollout is successful, the company could provide an alternative to Nvidia’s chips, which currently hold over 80% of the AI accelerator market.
Following the first chip’s success, OpenAI engineers plan to develop more advanced processors with each new iteration. The company has also joined the $500 billion Stargate infrastructure program announced by former U.S. President Donald Trump last month.
The chip is being developed by an internal OpenAI team led by Richard Ho, in collaboration with Broadcom. Ho, who previously worked at Google leading specialized AI chip projects, has expanded his team to 40 employees. However, this is still significantly smaller than the large-scale teams at companies like Google or Amazon.
Challenges and Industry Comparison
Analysts believe OpenAI’s new AI accelerator will initially have a limited role in its infrastructure, adds NIXSOLUTIONS. To build a program comparable to Google or Amazon, OpenAI would need to hire hundreds of engineers.
Industry sources estimate the cost of designing a high-performance AI chip at $500 million, potentially doubling when including software and necessary peripherals. In comparison, Meta plans to invest $60 billion in AI infrastructure in 2025, while Microsoft is allocating $80 billion annually.
OpenAI’s progress in AI chip development could reshape the competitive landscape. We’ll keep you updated as more details emerge.