Microsoft has the rights to the IP for OpenAI's custom AI semiconductors that the startup is developing with Broadcom.
In a recent appearance on author Dwarkesh Patel's podcast, and reported by Bloomberg, CEO Satya Nadella said regarding OpenAI's chip development: "As they innovate even at the system level, we get access to all of it. We first want to instantiate what they build for them, but then we’ll extend it.”
Whether that means Microsoft itself will—or can—use the chips, or offer them to others, is unclear. DCD has contacted Microsoft for clarification.
Nadella added that Microsoft will work with designs from OpenAI “knowing we have the IP rights,” though the cloud giant does not have property rights to OpenAI's consumer hardware.
This is part of the recently made agreement between OpenAI and Microsoft that enabled the AI startup to restructure into a for-profit. Microsoft retains a 27 percent stake worth $135bn in OpenAI, and will continue to have exclusive IP rights and Azure API exclusivity until Artificial General Intelligence (AGI) is achieved.
In addition, it saw Microsoft’s IP rights for both models and products extended through 2032 and now include models post-AGI, including any IP related to data center hardware and software, but exclude OpenAI’s consumer hardware.
Reports that OpenAI was exploring developing a custom AI chip with Broadcom first emerged in July 2024. In October 2025, it was announced that the two are aiming to deploy some 10GW of chips by 2029, with the first to roll-out in the second half of 2026. In addition to chips, Broadcom and OpenAI are developing Ethernet solutions.
What IP Broadcom owns and is not covered by Microsoft's agreement is unclear.
OpenAI is currently hiring for a senior legal counsel with experience in circuits, processors, sensors, mechanical systems, and AI hardware.
Microsoft also has its own custom chips, with CTO Kevin Scott saying last month that the company hopes to mainly use its own chips one day.
Microsoft announced its Azure Cobalt 100 and Azure Maia 100 chips in November 2023 - the former being a CPU, and the latter an "AI accelerator." The Maia accelerator is built on TSMC’s 5nm node and has 105 billion transistors on a monolithic die. The company claims a performance of 1,600 teraflops of MXInt8 and 3,200 teraflops of MXFP4.
Microsoft also announced plans for the Maia 200—but in June 2025, reports emerged that it was pushing back mass production to 2026.
Comments