OpenAI’s move to develop proprietary artificial intelligence chips is set to deliver significant advantages to SoftBank, a major shareholder and financier of the ChatGPT maker’s ambitious data center expansion.
The initiative, which places OpenAI at the center of an AI hardware race, involves deep collaboration with SoftBank’s Arm subsidiary and semiconductor giant Broadcom, industry sources close to Arm said.
According to a report by The Information, citing individuals familiar with the matter, OpenAI is partnering with Arm to produce a central processing unit (CPU) chip engineered to operate alongside OpenAI’s in-development AI server chip. While OpenAI is co-designing its AI processor with Broadcom, Arm is hopeful that OpenAI will deploy the new CPU not only with its bespoke chips but also in conjunction with leading AI chip offerings from Nvidia and AMD, one source disclosed.
CPUs play a crucial support role to AI chips, handling general computing tasks while specialized AI accelerators focus on heavy machine learning workloads. Notably, Arm has recently shifted from exclusively selling chip designs to actively developing its own CPUs, signaling a strategic evolution for the UK-based subsidiary.
On Monday, OpenAI and Broadcom confirmed plans to jointly produce OpenAI’s first proprietary AI processors. Under the arrangement, OpenAI will design the chips, with Broadcom leading their development and mass deployment beginning in the second half of 2026.
